{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.999297541394882, "eval_steps": 400, "global_step": 5604, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.002676032781401572, "grad_norm": 11.496960384029425, "learning_rate": 8.9126559714795e-09, "logits/chosen": -0.07004348933696747, "logits/rejected": 0.13604964315891266, "logps/chosen": -1.716159462928772, "logps/rejected": -1.8895246982574463, "loss": 1.9598, "rewards/accuracies": 0.5625, "rewards/chosen": -1.716159462928772, "rewards/margins": 0.17336519062519073, "rewards/rejected": -1.8895246982574463, "step": 5 }, { "epoch": 0.005352065562803144, "grad_norm": 24.62799600093432, "learning_rate": 1.7825311942959e-08, "logits/chosen": 0.011140784248709679, "logits/rejected": 0.13161656260490417, "logps/chosen": -1.8020946979522705, "logps/rejected": -1.8448978662490845, "loss": 2.0522, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.8020946979522705, "rewards/margins": 0.042803239077329636, "rewards/rejected": -1.8448978662490845, "step": 10 }, { "epoch": 0.008028098344204716, "grad_norm": 22.60780341155583, "learning_rate": 2.67379679144385e-08, "logits/chosen": -0.025929506868124008, "logits/rejected": 0.07177695631980896, "logps/chosen": -1.6351503133773804, "logps/rejected": -1.7645927667617798, "loss": 1.9222, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.6351503133773804, "rewards/margins": 0.1294424831867218, "rewards/rejected": -1.7645927667617798, "step": 15 }, { "epoch": 0.010704131125606288, "grad_norm": 12.75852372874483, "learning_rate": 3.5650623885918e-08, "logits/chosen": -0.031114911660552025, "logits/rejected": 0.05350674316287041, "logps/chosen": -1.7255808115005493, "logps/rejected": -1.8063666820526123, "loss": 2.0038, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.7255808115005493, "rewards/margins": 0.08078588545322418, "rewards/rejected": -1.8063666820526123, "step": 20 }, { "epoch": 0.013380163907007862, "grad_norm": 30.50287118337272, "learning_rate": 4.45632798573975e-08, "logits/chosen": -0.0557979941368103, "logits/rejected": 0.02790733054280281, "logps/chosen": -1.8691844940185547, "logps/rejected": -1.7790381908416748, "loss": 2.176, "rewards/accuracies": 0.38749998807907104, "rewards/chosen": -1.8691844940185547, "rewards/margins": -0.09014609456062317, "rewards/rejected": -1.7790381908416748, "step": 25 }, { "epoch": 0.016056196688409432, "grad_norm": 27.251095547414117, "learning_rate": 5.3475935828877e-08, "logits/chosen": -0.09042690694332123, "logits/rejected": 0.001267892075702548, "logps/chosen": -1.9062639474868774, "logps/rejected": -1.830116868019104, "loss": 2.1541, "rewards/accuracies": 0.4437499940395355, "rewards/chosen": -1.9062639474868774, "rewards/margins": -0.07614694535732269, "rewards/rejected": -1.830116868019104, "step": 30 }, { "epoch": 0.018732229469811006, "grad_norm": 19.649059504433637, "learning_rate": 6.23885918003565e-08, "logits/chosen": -0.038590312004089355, "logits/rejected": 0.12247200310230255, "logps/chosen": -1.8448536396026611, "logps/rejected": -1.9959800243377686, "loss": 2.1077, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.8448536396026611, "rewards/margins": 0.15112656354904175, "rewards/rejected": -1.9959800243377686, "step": 35 }, { "epoch": 0.021408262251212576, "grad_norm": 25.97747145346806, "learning_rate": 7.1301247771836e-08, "logits/chosen": 0.046131886541843414, "logits/rejected": 0.22134876251220703, "logps/chosen": -1.8790369033813477, "logps/rejected": -1.7409080266952515, "loss": 2.1579, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -1.8790369033813477, "rewards/margins": -0.13812878727912903, "rewards/rejected": -1.7409080266952515, "step": 40 }, { "epoch": 0.02408429503261415, "grad_norm": 23.426944055276945, "learning_rate": 8.021390374331551e-08, "logits/chosen": 0.03349178284406662, "logits/rejected": 0.23275916278362274, "logps/chosen": -1.8335708379745483, "logps/rejected": -1.8679958581924438, "loss": 2.1043, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.8335708379745483, "rewards/margins": 0.034425050020217896, "rewards/rejected": -1.8679958581924438, "step": 45 }, { "epoch": 0.026760327814015723, "grad_norm": 28.041207238046148, "learning_rate": 8.9126559714795e-08, "logits/chosen": -0.04500294104218483, "logits/rejected": 0.10422907024621964, "logps/chosen": -1.8920742273330688, "logps/rejected": -1.7743374109268188, "loss": 2.1584, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.8920742273330688, "rewards/margins": -0.1177368313074112, "rewards/rejected": -1.7743374109268188, "step": 50 }, { "epoch": 0.029436360595417294, "grad_norm": 22.230131366220668, "learning_rate": 9.80392156862745e-08, "logits/chosen": -0.11886776983737946, "logits/rejected": 0.09651211649179459, "logps/chosen": -1.8243519067764282, "logps/rejected": -1.8590061664581299, "loss": 2.0787, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.8243519067764282, "rewards/margins": 0.034654535353183746, "rewards/rejected": -1.8590061664581299, "step": 55 }, { "epoch": 0.032112393376818864, "grad_norm": 25.41905736053761, "learning_rate": 1.06951871657754e-07, "logits/chosen": -0.06625660508871078, "logits/rejected": 0.1267094761133194, "logps/chosen": -1.7776578664779663, "logps/rejected": -1.8829160928726196, "loss": 2.0253, "rewards/accuracies": 0.53125, "rewards/chosen": -1.7776578664779663, "rewards/margins": 0.10525822639465332, "rewards/rejected": -1.8829160928726196, "step": 60 }, { "epoch": 0.03478842615822044, "grad_norm": 24.019383898594057, "learning_rate": 1.158645276292335e-07, "logits/chosen": -0.026983071118593216, "logits/rejected": 0.11884398758411407, "logps/chosen": -1.628180742263794, "logps/rejected": -1.756899118423462, "loss": 1.8999, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.628180742263794, "rewards/margins": 0.12871839106082916, "rewards/rejected": -1.756899118423462, "step": 65 }, { "epoch": 0.03746445893962201, "grad_norm": 24.8180689671467, "learning_rate": 1.24777183600713e-07, "logits/chosen": -0.06905418634414673, "logits/rejected": 0.08335807174444199, "logps/chosen": -1.7545757293701172, "logps/rejected": -1.7981780767440796, "loss": 2.0317, "rewards/accuracies": 0.42500001192092896, "rewards/chosen": -1.7545757293701172, "rewards/margins": 0.043602414429187775, "rewards/rejected": -1.7981780767440796, "step": 70 }, { "epoch": 0.04014049172102358, "grad_norm": 20.509875985552213, "learning_rate": 1.3368983957219251e-07, "logits/chosen": -0.040019623935222626, "logits/rejected": 0.1444050371646881, "logps/chosen": -1.7451117038726807, "logps/rejected": -2.000904083251953, "loss": 1.9932, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.7451117038726807, "rewards/margins": 0.25579261779785156, "rewards/rejected": -2.000904083251953, "step": 75 }, { "epoch": 0.04281652450242515, "grad_norm": 19.0701022504621, "learning_rate": 1.42602495543672e-07, "logits/chosen": -0.011737307533621788, "logits/rejected": 0.09179870784282684, "logps/chosen": -1.6793285608291626, "logps/rejected": -1.7125403881072998, "loss": 1.9591, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.6793285608291626, "rewards/margins": 0.03321211040019989, "rewards/rejected": -1.7125403881072998, "step": 80 }, { "epoch": 0.04549255728382673, "grad_norm": 14.351006445287252, "learning_rate": 1.5151515151515152e-07, "logits/chosen": -0.16043472290039062, "logits/rejected": 0.08176468312740326, "logps/chosen": -1.74630868434906, "logps/rejected": -1.9133342504501343, "loss": 2.0241, "rewards/accuracies": 0.5, "rewards/chosen": -1.74630868434906, "rewards/margins": 0.16702556610107422, "rewards/rejected": -1.9133342504501343, "step": 85 }, { "epoch": 0.0481685900652283, "grad_norm": 25.696281391928444, "learning_rate": 1.6042780748663102e-07, "logits/chosen": 0.053802113980054855, "logits/rejected": 0.014244931749999523, "logps/chosen": -1.692808747291565, "logps/rejected": -1.732717752456665, "loss": 1.9809, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -1.692808747291565, "rewards/margins": 0.03990894928574562, "rewards/rejected": -1.732717752456665, "step": 90 }, { "epoch": 0.05084462284662987, "grad_norm": 29.848111715590246, "learning_rate": 1.693404634581105e-07, "logits/chosen": -0.10460035502910614, "logits/rejected": 0.04046844691038132, "logps/chosen": -1.7314746379852295, "logps/rejected": -1.857269287109375, "loss": 1.9942, "rewards/accuracies": 0.59375, "rewards/chosen": -1.7314746379852295, "rewards/margins": 0.12579476833343506, "rewards/rejected": -1.857269287109375, "step": 95 }, { "epoch": 0.05352065562803145, "grad_norm": 14.152630130218263, "learning_rate": 1.7825311942959e-07, "logits/chosen": -0.03871261700987816, "logits/rejected": 0.02377936616539955, "logps/chosen": -1.6145925521850586, "logps/rejected": -1.7207527160644531, "loss": 1.8863, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.6145925521850586, "rewards/margins": 0.10616017878055573, "rewards/rejected": -1.7207527160644531, "step": 100 }, { "epoch": 0.05619668840943302, "grad_norm": 20.414035787261877, "learning_rate": 1.8716577540106952e-07, "logits/chosen": 0.03783528506755829, "logits/rejected": 0.06228012964129448, "logps/chosen": -1.533184289932251, "logps/rejected": -1.7013075351715088, "loss": 1.813, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.533184289932251, "rewards/margins": 0.16812297701835632, "rewards/rejected": -1.7013075351715088, "step": 105 }, { "epoch": 0.05887272119083459, "grad_norm": 18.838681459332435, "learning_rate": 1.96078431372549e-07, "logits/chosen": -0.01743622124195099, "logits/rejected": 0.07553930580615997, "logps/chosen": -1.5218614339828491, "logps/rejected": -1.581130027770996, "loss": 1.8293, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.5218614339828491, "rewards/margins": 0.05926854535937309, "rewards/rejected": -1.581130027770996, "step": 110 }, { "epoch": 0.06154875397223616, "grad_norm": 20.681911095989836, "learning_rate": 2.049910873440285e-07, "logits/chosen": 0.02094849944114685, "logits/rejected": 0.22674188017845154, "logps/chosen": -1.5216385126113892, "logps/rejected": -1.7595542669296265, "loss": 1.7828, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.5216385126113892, "rewards/margins": 0.23791582882404327, "rewards/rejected": -1.7595542669296265, "step": 115 }, { "epoch": 0.06422478675363773, "grad_norm": 22.590931606448105, "learning_rate": 2.13903743315508e-07, "logits/chosen": -0.08596738427877426, "logits/rejected": 0.08397196233272552, "logps/chosen": -1.5576399564743042, "logps/rejected": -1.660080909729004, "loss": 1.8551, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.5576399564743042, "rewards/margins": 0.10244102776050568, "rewards/rejected": -1.660080909729004, "step": 120 }, { "epoch": 0.0669008195350393, "grad_norm": 7.514452992243219, "learning_rate": 2.2281639928698751e-07, "logits/chosen": -0.10349993407726288, "logits/rejected": 0.025051873177289963, "logps/chosen": -1.509615182876587, "logps/rejected": -1.4850200414657593, "loss": 1.8367, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.509615182876587, "rewards/margins": -0.02459501102566719, "rewards/rejected": -1.4850200414657593, "step": 125 }, { "epoch": 0.06957685231644088, "grad_norm": 26.418915644433355, "learning_rate": 2.31729055258467e-07, "logits/chosen": 0.009734541177749634, "logits/rejected": 0.1401519626379013, "logps/chosen": -1.5340511798858643, "logps/rejected": -1.6470483541488647, "loss": 1.8246, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.5340511798858643, "rewards/margins": 0.11299731582403183, "rewards/rejected": -1.6470483541488647, "step": 130 }, { "epoch": 0.07225288509784245, "grad_norm": 22.937112750822322, "learning_rate": 2.406417112299465e-07, "logits/chosen": -0.0796973779797554, "logits/rejected": 0.03641734644770622, "logps/chosen": -1.5494601726531982, "logps/rejected": -1.596419095993042, "loss": 1.8564, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.5494601726531982, "rewards/margins": 0.04695894196629524, "rewards/rejected": -1.596419095993042, "step": 135 }, { "epoch": 0.07492891787924402, "grad_norm": 14.481886404114174, "learning_rate": 2.49554367201426e-07, "logits/chosen": -0.08387432992458344, "logits/rejected": 0.07397343963384628, "logps/chosen": -1.4759531021118164, "logps/rejected": -1.5443933010101318, "loss": 1.8147, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.4759531021118164, "rewards/margins": 0.06844006478786469, "rewards/rejected": -1.5443933010101318, "step": 140 }, { "epoch": 0.0776049506606456, "grad_norm": 13.900630224506688, "learning_rate": 2.5846702317290554e-07, "logits/chosen": -0.06520480662584305, "logits/rejected": 0.08101221174001694, "logps/chosen": -1.3628003597259521, "logps/rejected": -1.4583301544189453, "loss": 1.7122, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3628003597259521, "rewards/margins": 0.09552977979183197, "rewards/rejected": -1.4583301544189453, "step": 145 }, { "epoch": 0.08028098344204716, "grad_norm": 14.456992031762525, "learning_rate": 2.6737967914438503e-07, "logits/chosen": -0.12199876457452774, "logits/rejected": 0.02296736277639866, "logps/chosen": -1.3004138469696045, "logps/rejected": -1.3106777667999268, "loss": 1.6961, "rewards/accuracies": 0.5, "rewards/chosen": -1.3004138469696045, "rewards/margins": 0.010263770818710327, "rewards/rejected": -1.3106777667999268, "step": 150 }, { "epoch": 0.08295701622344874, "grad_norm": 10.003665109023784, "learning_rate": 2.762923351158645e-07, "logits/chosen": -0.15272387862205505, "logits/rejected": -0.10898196697235107, "logps/chosen": -1.3114030361175537, "logps/rejected": -1.4270459413528442, "loss": 1.6758, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3114030361175537, "rewards/margins": 0.1156429648399353, "rewards/rejected": -1.4270459413528442, "step": 155 }, { "epoch": 0.0856330490048503, "grad_norm": 10.058938611024177, "learning_rate": 2.85204991087344e-07, "logits/chosen": -0.20483560860157013, "logits/rejected": -0.07443522661924362, "logps/chosen": -1.4005239009857178, "logps/rejected": -1.3797595500946045, "loss": 1.7876, "rewards/accuracies": 0.46875, "rewards/chosen": -1.4005239009857178, "rewards/margins": -0.020764362066984177, "rewards/rejected": -1.3797595500946045, "step": 160 }, { "epoch": 0.08830908178625188, "grad_norm": 12.766881003620881, "learning_rate": 2.941176470588235e-07, "logits/chosen": -0.115422323346138, "logits/rejected": 0.043187130242586136, "logps/chosen": -1.3161473274230957, "logps/rejected": -1.3928664922714233, "loss": 1.7146, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3161473274230957, "rewards/margins": 0.07671914994716644, "rewards/rejected": -1.3928664922714233, "step": 165 }, { "epoch": 0.09098511456765346, "grad_norm": 12.889165264352723, "learning_rate": 3.0303030303030305e-07, "logits/chosen": -0.1455298364162445, "logits/rejected": -0.09741321206092834, "logps/chosen": -1.4335427284240723, "logps/rejected": -1.5023107528686523, "loss": 1.7765, "rewards/accuracies": 0.5, "rewards/chosen": -1.4335427284240723, "rewards/margins": 0.0687679797410965, "rewards/rejected": -1.5023107528686523, "step": 170 }, { "epoch": 0.09366114734905502, "grad_norm": 10.673475631603585, "learning_rate": 3.1194295900178254e-07, "logits/chosen": -0.017373189330101013, "logits/rejected": -0.023483578115701675, "logps/chosen": -1.32071053981781, "logps/rejected": -1.407671570777893, "loss": 1.7025, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.32071053981781, "rewards/margins": 0.08696107566356659, "rewards/rejected": -1.407671570777893, "step": 175 }, { "epoch": 0.0963371801304566, "grad_norm": 8.027679837174334, "learning_rate": 3.2085561497326203e-07, "logits/chosen": -0.05327599123120308, "logits/rejected": -0.05693614482879639, "logps/chosen": -1.3389848470687866, "logps/rejected": -1.5398029088974, "loss": 1.6771, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3389848470687866, "rewards/margins": 0.20081813633441925, "rewards/rejected": -1.5398029088974, "step": 180 }, { "epoch": 0.09901321291185818, "grad_norm": 8.508747965236294, "learning_rate": 3.297682709447415e-07, "logits/chosen": -0.18951360881328583, "logits/rejected": -0.11156556755304337, "logps/chosen": -1.331371545791626, "logps/rejected": -1.378575325012207, "loss": 1.7309, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.331371545791626, "rewards/margins": 0.047203779220581055, "rewards/rejected": -1.378575325012207, "step": 185 }, { "epoch": 0.10168924569325974, "grad_norm": 10.290163511178912, "learning_rate": 3.38680926916221e-07, "logits/chosen": -0.1001301184296608, "logits/rejected": 0.008160894736647606, "logps/chosen": -1.2569787502288818, "logps/rejected": -1.3878567218780518, "loss": 1.6555, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2569787502288818, "rewards/margins": 0.13087785243988037, "rewards/rejected": -1.3878567218780518, "step": 190 }, { "epoch": 0.10436527847466132, "grad_norm": 8.298194138873017, "learning_rate": 3.475935828877005e-07, "logits/chosen": -0.016883565112948418, "logits/rejected": 0.1294752061367035, "logps/chosen": -1.2395333051681519, "logps/rejected": -1.4013464450836182, "loss": 1.6175, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2395333051681519, "rewards/margins": 0.16181306540966034, "rewards/rejected": -1.4013464450836182, "step": 195 }, { "epoch": 0.1070413112560629, "grad_norm": 21.57478799062816, "learning_rate": 3.5650623885918e-07, "logits/chosen": -0.10271809995174408, "logits/rejected": 0.027957985177636147, "logps/chosen": -1.3616082668304443, "logps/rejected": -1.4011781215667725, "loss": 1.7377, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3616082668304443, "rewards/margins": 0.03956984728574753, "rewards/rejected": -1.4011781215667725, "step": 200 }, { "epoch": 0.10971734403746446, "grad_norm": 15.11761693118118, "learning_rate": 3.654188948306595e-07, "logits/chosen": -0.10172367095947266, "logits/rejected": 0.028435533866286278, "logps/chosen": -1.272033452987671, "logps/rejected": -1.3431034088134766, "loss": 1.6765, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.272033452987671, "rewards/margins": 0.07106985151767731, "rewards/rejected": -1.3431034088134766, "step": 205 }, { "epoch": 0.11239337681886603, "grad_norm": 12.162560812099604, "learning_rate": 3.7433155080213904e-07, "logits/chosen": -0.17484010756015778, "logits/rejected": 0.00014435648336075246, "logps/chosen": -1.3493441343307495, "logps/rejected": -1.457792043685913, "loss": 1.7159, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3493441343307495, "rewards/margins": 0.10844788700342178, "rewards/rejected": -1.457792043685913, "step": 210 }, { "epoch": 0.1150694096002676, "grad_norm": 10.505960016810631, "learning_rate": 3.8324420677361853e-07, "logits/chosen": -0.21476414799690247, "logits/rejected": 0.015725497156381607, "logps/chosen": -1.3726469278335571, "logps/rejected": -1.4256699085235596, "loss": 1.7294, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3726469278335571, "rewards/margins": 0.053022850304841995, "rewards/rejected": -1.4256699085235596, "step": 215 }, { "epoch": 0.11774544238166917, "grad_norm": 16.8831046999415, "learning_rate": 3.92156862745098e-07, "logits/chosen": 0.012892027385532856, "logits/rejected": 0.10256718099117279, "logps/chosen": -1.294647455215454, "logps/rejected": -1.4366071224212646, "loss": 1.657, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.294647455215454, "rewards/margins": 0.14195974171161652, "rewards/rejected": -1.4366071224212646, "step": 220 }, { "epoch": 0.12042147516307075, "grad_norm": 7.742873673583737, "learning_rate": 4.010695187165775e-07, "logits/chosen": -0.11977468430995941, "logits/rejected": 0.03806857019662857, "logps/chosen": -1.3015865087509155, "logps/rejected": -1.4278619289398193, "loss": 1.6556, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3015865087509155, "rewards/margins": 0.12627527117729187, "rewards/rejected": -1.4278619289398193, "step": 225 }, { "epoch": 0.12309750794447231, "grad_norm": 8.81005179915772, "learning_rate": 4.09982174688057e-07, "logits/chosen": -0.0395035445690155, "logits/rejected": 0.0292628463357687, "logps/chosen": -1.3038604259490967, "logps/rejected": -1.4596668481826782, "loss": 1.6834, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3038604259490967, "rewards/margins": 0.1558062732219696, "rewards/rejected": -1.4596668481826782, "step": 230 }, { "epoch": 0.1257735407258739, "grad_norm": 12.371393950608745, "learning_rate": 4.188948306595365e-07, "logits/chosen": -0.0051063718274235725, "logits/rejected": 0.1200011596083641, "logps/chosen": -1.2733403444290161, "logps/rejected": -1.43751859664917, "loss": 1.6422, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2733403444290161, "rewards/margins": 0.16417834162712097, "rewards/rejected": -1.43751859664917, "step": 235 }, { "epoch": 0.12844957350727546, "grad_norm": 6.457958618778695, "learning_rate": 4.27807486631016e-07, "logits/chosen": -0.053750477731227875, "logits/rejected": 0.06239823251962662, "logps/chosen": -1.289251446723938, "logps/rejected": -1.4710654020309448, "loss": 1.6871, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.289251446723938, "rewards/margins": 0.1818140149116516, "rewards/rejected": -1.4710654020309448, "step": 240 }, { "epoch": 0.13112560628867703, "grad_norm": 8.860937119366293, "learning_rate": 4.3672014260249554e-07, "logits/chosen": 4.385709689813666e-05, "logits/rejected": 0.10791563987731934, "logps/chosen": -1.4089033603668213, "logps/rejected": -1.4336974620819092, "loss": 1.775, "rewards/accuracies": 0.5625, "rewards/chosen": -1.4089033603668213, "rewards/margins": 0.024794071912765503, "rewards/rejected": -1.4336974620819092, "step": 245 }, { "epoch": 0.1338016390700786, "grad_norm": 10.341642763213565, "learning_rate": 4.4563279857397503e-07, "logits/chosen": -0.06541736423969269, "logits/rejected": 0.09154494106769562, "logps/chosen": -1.2923893928527832, "logps/rejected": -1.3493139743804932, "loss": 1.701, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.2923893928527832, "rewards/margins": 0.056924544274806976, "rewards/rejected": -1.3493139743804932, "step": 250 }, { "epoch": 0.1364776718514802, "grad_norm": 8.995445203244538, "learning_rate": 4.545454545454545e-07, "logits/chosen": -0.03822184354066849, "logits/rejected": 0.09474680572748184, "logps/chosen": -1.2578426599502563, "logps/rejected": -1.3610799312591553, "loss": 1.6498, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2578426599502563, "rewards/margins": 0.10323736816644669, "rewards/rejected": -1.3610799312591553, "step": 255 }, { "epoch": 0.13915370463288176, "grad_norm": 7.808941045370267, "learning_rate": 4.63458110516934e-07, "logits/chosen": -0.2500733733177185, "logits/rejected": -0.1506270170211792, "logps/chosen": -1.3432053327560425, "logps/rejected": -1.4983842372894287, "loss": 1.6675, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3432053327560425, "rewards/margins": 0.15517887473106384, "rewards/rejected": -1.4983842372894287, "step": 260 }, { "epoch": 0.1418297374142833, "grad_norm": 11.263217166611032, "learning_rate": 4.723707664884135e-07, "logits/chosen": -0.11429083347320557, "logits/rejected": -0.03443082422018051, "logps/chosen": -1.333311676979065, "logps/rejected": -1.495581030845642, "loss": 1.6814, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.333311676979065, "rewards/margins": 0.16226927936077118, "rewards/rejected": -1.495581030845642, "step": 265 }, { "epoch": 0.1445057701956849, "grad_norm": 8.04536803191158, "learning_rate": 4.81283422459893e-07, "logits/chosen": -0.09445016086101532, "logits/rejected": 0.0262304600328207, "logps/chosen": -1.3086737394332886, "logps/rejected": -1.4043707847595215, "loss": 1.6979, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.3086737394332886, "rewards/margins": 0.09569709002971649, "rewards/rejected": -1.4043707847595215, "step": 270 }, { "epoch": 0.14718180297708647, "grad_norm": 9.923872351321576, "learning_rate": 4.901960784313725e-07, "logits/chosen": -0.028187870979309082, "logits/rejected": 0.06616386771202087, "logps/chosen": -1.2604142427444458, "logps/rejected": -1.4108034372329712, "loss": 1.6666, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2604142427444458, "rewards/margins": 0.15038911998271942, "rewards/rejected": -1.4108034372329712, "step": 275 }, { "epoch": 0.14985783575848804, "grad_norm": 10.338563914076289, "learning_rate": 4.99108734402852e-07, "logits/chosen": -0.10596181452274323, "logits/rejected": 0.043247222900390625, "logps/chosen": -1.3092851638793945, "logps/rejected": -1.4032866954803467, "loss": 1.6684, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3092851638793945, "rewards/margins": 0.09400156885385513, "rewards/rejected": -1.4032866954803467, "step": 280 }, { "epoch": 0.15253386853988962, "grad_norm": 9.301649752575466, "learning_rate": 5.080213903743315e-07, "logits/chosen": -0.07537764310836792, "logits/rejected": 0.05667581036686897, "logps/chosen": -1.339205026626587, "logps/rejected": -1.416961669921875, "loss": 1.7063, "rewards/accuracies": 0.5625, "rewards/chosen": -1.339205026626587, "rewards/margins": 0.07775678485631943, "rewards/rejected": -1.416961669921875, "step": 285 }, { "epoch": 0.1552099013212912, "grad_norm": 8.371480037795749, "learning_rate": 5.169340463458111e-07, "logits/chosen": -0.137905552983284, "logits/rejected": 0.1398901641368866, "logps/chosen": -1.3563731908798218, "logps/rejected": -1.47064208984375, "loss": 1.6818, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3563731908798218, "rewards/margins": 0.11426875740289688, "rewards/rejected": -1.47064208984375, "step": 290 }, { "epoch": 0.15788593410269275, "grad_norm": 12.564451448904155, "learning_rate": 5.258467023172905e-07, "logits/chosen": -0.08563229441642761, "logits/rejected": -0.029086634516716003, "logps/chosen": -1.2586605548858643, "logps/rejected": -1.3940225839614868, "loss": 1.6419, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2586605548858643, "rewards/margins": 0.13536228239536285, "rewards/rejected": -1.3940225839614868, "step": 295 }, { "epoch": 0.16056196688409433, "grad_norm": 8.503279331236856, "learning_rate": 5.347593582887701e-07, "logits/chosen": -0.08329950273036957, "logits/rejected": 0.07354854047298431, "logps/chosen": -1.2957587242126465, "logps/rejected": -1.3708839416503906, "loss": 1.6978, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2957587242126465, "rewards/margins": 0.07512523233890533, "rewards/rejected": -1.3708839416503906, "step": 300 }, { "epoch": 0.1632379996654959, "grad_norm": 7.082413964772183, "learning_rate": 5.436720142602496e-07, "logits/chosen": -0.031066909432411194, "logits/rejected": 0.0350339449942112, "logps/chosen": -1.3960840702056885, "logps/rejected": -1.4059756994247437, "loss": 1.7538, "rewards/accuracies": 0.5, "rewards/chosen": -1.3960840702056885, "rewards/margins": 0.009891606867313385, "rewards/rejected": -1.4059756994247437, "step": 305 }, { "epoch": 0.16591403244689748, "grad_norm": 8.740780332970166, "learning_rate": 5.52584670231729e-07, "logits/chosen": -0.2536133825778961, "logits/rejected": -0.17383472621440887, "logps/chosen": -1.362154483795166, "logps/rejected": -1.4443097114562988, "loss": 1.731, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.362154483795166, "rewards/margins": 0.08215537667274475, "rewards/rejected": -1.4443097114562988, "step": 310 }, { "epoch": 0.16859006522829906, "grad_norm": 10.359867234690551, "learning_rate": 5.614973262032086e-07, "logits/chosen": -0.03551404923200607, "logits/rejected": 0.10857124626636505, "logps/chosen": -1.3527511358261108, "logps/rejected": -1.4951927661895752, "loss": 1.6965, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3527511358261108, "rewards/margins": 0.14244171977043152, "rewards/rejected": -1.4951927661895752, "step": 315 }, { "epoch": 0.1712660980097006, "grad_norm": 6.856055712472328, "learning_rate": 5.70409982174688e-07, "logits/chosen": -0.09569795429706573, "logits/rejected": 0.026764288544654846, "logps/chosen": -1.3127437829971313, "logps/rejected": -1.3622338771820068, "loss": 1.7075, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3127437829971313, "rewards/margins": 0.04949017986655235, "rewards/rejected": -1.3622338771820068, "step": 320 }, { "epoch": 0.17394213079110218, "grad_norm": 9.565706670007815, "learning_rate": 5.793226381461676e-07, "logits/chosen": -0.15836867690086365, "logits/rejected": -0.05464999005198479, "logps/chosen": -1.3131455183029175, "logps/rejected": -1.5495985746383667, "loss": 1.6645, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3131455183029175, "rewards/margins": 0.23645305633544922, "rewards/rejected": -1.5495985746383667, "step": 325 }, { "epoch": 0.17661816357250376, "grad_norm": 9.880050272421682, "learning_rate": 5.88235294117647e-07, "logits/chosen": -0.062169916927814484, "logits/rejected": 0.06957022100687027, "logps/chosen": -1.3260176181793213, "logps/rejected": -1.4939022064208984, "loss": 1.6917, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3260176181793213, "rewards/margins": 0.16788452863693237, "rewards/rejected": -1.4939022064208984, "step": 330 }, { "epoch": 0.17929419635390534, "grad_norm": 9.66922885238171, "learning_rate": 5.971479500891266e-07, "logits/chosen": 0.006079360842704773, "logits/rejected": 0.09727548807859421, "logps/chosen": -1.3392069339752197, "logps/rejected": -1.3705189228057861, "loss": 1.7279, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3392069339752197, "rewards/margins": 0.031312037259340286, "rewards/rejected": -1.3705189228057861, "step": 335 }, { "epoch": 0.18197022913530692, "grad_norm": 11.033412727973458, "learning_rate": 6.060606060606061e-07, "logits/chosen": -0.03210308775305748, "logits/rejected": 0.10471953451633453, "logps/chosen": -1.3952913284301758, "logps/rejected": -1.4770443439483643, "loss": 1.759, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.3952913284301758, "rewards/margins": 0.08175303041934967, "rewards/rejected": -1.4770443439483643, "step": 340 }, { "epoch": 0.1846462619167085, "grad_norm": 10.896535078719086, "learning_rate": 6.149732620320855e-07, "logits/chosen": 0.03205911070108414, "logits/rejected": 0.057826824486255646, "logps/chosen": -1.3063201904296875, "logps/rejected": -1.4427855014801025, "loss": 1.6554, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3063201904296875, "rewards/margins": 0.13646534085273743, "rewards/rejected": -1.4427855014801025, "step": 345 }, { "epoch": 0.18732229469811004, "grad_norm": 10.28054144795725, "learning_rate": 6.238859180035651e-07, "logits/chosen": -0.010967472568154335, "logits/rejected": 0.07389771938323975, "logps/chosen": -1.2843072414398193, "logps/rejected": -1.399267554283142, "loss": 1.6701, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2843072414398193, "rewards/margins": 0.1149604544043541, "rewards/rejected": -1.399267554283142, "step": 350 }, { "epoch": 0.18999832747951162, "grad_norm": 11.039648213043902, "learning_rate": 6.327985739750445e-07, "logits/chosen": -0.10593543201684952, "logits/rejected": 0.10453915596008301, "logps/chosen": -1.3861037492752075, "logps/rejected": -1.4115387201309204, "loss": 1.7495, "rewards/accuracies": 0.46875, "rewards/chosen": -1.3861037492752075, "rewards/margins": 0.025434961542487144, "rewards/rejected": -1.4115387201309204, "step": 355 }, { "epoch": 0.1926743602609132, "grad_norm": 9.815585300420476, "learning_rate": 6.417112299465241e-07, "logits/chosen": -0.07406492531299591, "logits/rejected": 0.0021810054313391447, "logps/chosen": -1.315483808517456, "logps/rejected": -1.4323620796203613, "loss": 1.7339, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.315483808517456, "rewards/margins": 0.11687831580638885, "rewards/rejected": -1.4323620796203613, "step": 360 }, { "epoch": 0.19535039304231477, "grad_norm": 11.993286618551066, "learning_rate": 6.506238859180035e-07, "logits/chosen": 0.007736300118267536, "logits/rejected": 0.08413462340831757, "logps/chosen": -1.292368769645691, "logps/rejected": -1.3933672904968262, "loss": 1.6824, "rewards/accuracies": 0.59375, "rewards/chosen": -1.292368769645691, "rewards/margins": 0.10099846124649048, "rewards/rejected": -1.3933672904968262, "step": 365 }, { "epoch": 0.19802642582371635, "grad_norm": 6.775568693955582, "learning_rate": 6.59536541889483e-07, "logits/chosen": -0.03933990001678467, "logits/rejected": 0.043112464249134064, "logps/chosen": -1.288225531578064, "logps/rejected": -1.329222321510315, "loss": 1.6894, "rewards/accuracies": 0.5, "rewards/chosen": -1.288225531578064, "rewards/margins": 0.040996819734573364, "rewards/rejected": -1.329222321510315, "step": 370 }, { "epoch": 0.2007024586051179, "grad_norm": 15.797955341210931, "learning_rate": 6.684491978609626e-07, "logits/chosen": -0.0702676773071289, "logits/rejected": 0.08288853615522385, "logps/chosen": -1.2703843116760254, "logps/rejected": -1.4104673862457275, "loss": 1.6667, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2703843116760254, "rewards/margins": 0.1400829255580902, "rewards/rejected": -1.4104673862457275, "step": 375 }, { "epoch": 0.20337849138651948, "grad_norm": 9.3923973841865, "learning_rate": 6.77361853832442e-07, "logits/chosen": -0.05306627228856087, "logits/rejected": 0.02266976237297058, "logps/chosen": -1.2857197523117065, "logps/rejected": -1.4401063919067383, "loss": 1.6482, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2857197523117065, "rewards/margins": 0.1543864607810974, "rewards/rejected": -1.4401063919067383, "step": 380 }, { "epoch": 0.20605452416792105, "grad_norm": 5.749278727008655, "learning_rate": 6.862745098039216e-07, "logits/chosen": 0.024824131280183792, "logits/rejected": 0.10414397716522217, "logps/chosen": -1.383319616317749, "logps/rejected": -1.372910976409912, "loss": 1.7442, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.383319616317749, "rewards/margins": -0.010408599860966206, "rewards/rejected": -1.372910976409912, "step": 385 }, { "epoch": 0.20873055694932263, "grad_norm": 10.617790928113745, "learning_rate": 6.95187165775401e-07, "logits/chosen": 0.06647435575723648, "logits/rejected": 0.22994470596313477, "logps/chosen": -1.3744819164276123, "logps/rejected": -1.4345366954803467, "loss": 1.7584, "rewards/accuracies": 0.46875, "rewards/chosen": -1.3744819164276123, "rewards/margins": 0.06005479767918587, "rewards/rejected": -1.4345366954803467, "step": 390 }, { "epoch": 0.2114065897307242, "grad_norm": 7.745401478461128, "learning_rate": 7.040998217468806e-07, "logits/chosen": -0.0460563488304615, "logits/rejected": 0.10764249414205551, "logps/chosen": -1.3257935047149658, "logps/rejected": -1.3358434438705444, "loss": 1.7054, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3257935047149658, "rewards/margins": 0.010050171986222267, "rewards/rejected": -1.3358434438705444, "step": 395 }, { "epoch": 0.2140826225121258, "grad_norm": 7.754185347608746, "learning_rate": 7.1301247771836e-07, "logits/chosen": 0.05485847592353821, "logits/rejected": 0.14413060247898102, "logps/chosen": -1.3048384189605713, "logps/rejected": -1.3969600200653076, "loss": 1.6555, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3048384189605713, "rewards/margins": 0.09212164580821991, "rewards/rejected": -1.3969600200653076, "step": 400 }, { "epoch": 0.2140826225121258, "eval_logits/chosen": 0.3137153089046478, "eval_logits/rejected": 0.40302950143814087, "eval_logps/chosen": -1.3383382558822632, "eval_logps/rejected": -1.4639986753463745, "eval_loss": 1.6940995454788208, "eval_rewards/accuracies": 0.5556379556655884, "eval_rewards/chosen": -1.3383382558822632, "eval_rewards/margins": 0.12566043436527252, "eval_rewards/rejected": -1.4639986753463745, "eval_runtime": 41.5943, "eval_samples_per_second": 32.336, "eval_steps_per_second": 8.102, "step": 400 }, { "epoch": 0.21675865529352734, "grad_norm": 8.754336696951281, "learning_rate": 7.219251336898395e-07, "logits/chosen": -0.005072087049484253, "logits/rejected": 0.08546868711709976, "logps/chosen": -1.3079750537872314, "logps/rejected": -1.371268630027771, "loss": 1.7065, "rewards/accuracies": 0.5, "rewards/chosen": -1.3079750537872314, "rewards/margins": 0.0632937103509903, "rewards/rejected": -1.371268630027771, "step": 405 }, { "epoch": 0.2194346880749289, "grad_norm": 10.519751722000656, "learning_rate": 7.30837789661319e-07, "logits/chosen": 0.0018633157014846802, "logits/rejected": 0.12912698090076447, "logps/chosen": -1.2822954654693604, "logps/rejected": -1.3665344715118408, "loss": 1.6734, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2822954654693604, "rewards/margins": 0.08423889428377151, "rewards/rejected": -1.3665344715118408, "step": 410 }, { "epoch": 0.2221107208563305, "grad_norm": 6.074341910212351, "learning_rate": 7.397504456327985e-07, "logits/chosen": 0.004594428930431604, "logits/rejected": 0.03903172165155411, "logps/chosen": -1.2817447185516357, "logps/rejected": -1.4341824054718018, "loss": 1.6418, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2817447185516357, "rewards/margins": 0.15243759751319885, "rewards/rejected": -1.4341824054718018, "step": 415 }, { "epoch": 0.22478675363773207, "grad_norm": 7.604611473074622, "learning_rate": 7.486631016042781e-07, "logits/chosen": -0.015866711735725403, "logits/rejected": 0.16937878727912903, "logps/chosen": -1.2663681507110596, "logps/rejected": -1.3563402891159058, "loss": 1.6679, "rewards/accuracies": 0.5, "rewards/chosen": -1.2663681507110596, "rewards/margins": 0.08997206389904022, "rewards/rejected": -1.3563402891159058, "step": 420 }, { "epoch": 0.22746278641913364, "grad_norm": 6.717082718541665, "learning_rate": 7.575757575757575e-07, "logits/chosen": -0.0675196424126625, "logits/rejected": 0.12153605371713638, "logps/chosen": -1.306672215461731, "logps/rejected": -1.4570428133010864, "loss": 1.6619, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.306672215461731, "rewards/margins": 0.15037064254283905, "rewards/rejected": -1.4570428133010864, "step": 425 }, { "epoch": 0.2301388192005352, "grad_norm": 7.371899282836315, "learning_rate": 7.664884135472371e-07, "logits/chosen": -0.08296699821949005, "logits/rejected": 0.10469052940607071, "logps/chosen": -1.3389928340911865, "logps/rejected": -1.4607927799224854, "loss": 1.6907, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3389928340911865, "rewards/margins": 0.12179992347955704, "rewards/rejected": -1.4607927799224854, "step": 430 }, { "epoch": 0.23281485198193677, "grad_norm": 12.06561551172007, "learning_rate": 7.754010695187165e-07, "logits/chosen": -0.012138782069087029, "logits/rejected": 0.07224924117326736, "logps/chosen": -1.2074350118637085, "logps/rejected": -1.3375585079193115, "loss": 1.6144, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2074350118637085, "rewards/margins": 0.13012345135211945, "rewards/rejected": -1.3375585079193115, "step": 435 }, { "epoch": 0.23549088476333835, "grad_norm": 7.575184535702604, "learning_rate": 7.84313725490196e-07, "logits/chosen": -0.0038666813634335995, "logits/rejected": 0.08154430240392685, "logps/chosen": -1.2854907512664795, "logps/rejected": -1.368265986442566, "loss": 1.6762, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2854907512664795, "rewards/margins": 0.08277542889118195, "rewards/rejected": -1.368265986442566, "step": 440 }, { "epoch": 0.23816691754473993, "grad_norm": 8.138193319450046, "learning_rate": 7.932263814616755e-07, "logits/chosen": -0.061522066593170166, "logits/rejected": 0.04313060641288757, "logps/chosen": -1.3112833499908447, "logps/rejected": -1.433841347694397, "loss": 1.6972, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3112833499908447, "rewards/margins": 0.12255801260471344, "rewards/rejected": -1.433841347694397, "step": 445 }, { "epoch": 0.2408429503261415, "grad_norm": 8.640186349405695, "learning_rate": 8.02139037433155e-07, "logits/chosen": -0.020474979653954506, "logits/rejected": 0.09682993590831757, "logps/chosen": -1.2927807569503784, "logps/rejected": -1.4289346933364868, "loss": 1.6518, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2927807569503784, "rewards/margins": 0.13615401089191437, "rewards/rejected": -1.4289346933364868, "step": 450 }, { "epoch": 0.24351898310754308, "grad_norm": 9.540609815035786, "learning_rate": 8.110516934046346e-07, "logits/chosen": -0.01543851476162672, "logits/rejected": 0.07130376249551773, "logps/chosen": -1.2424728870391846, "logps/rejected": -1.41982102394104, "loss": 1.6114, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2424728870391846, "rewards/margins": 0.1773480623960495, "rewards/rejected": -1.41982102394104, "step": 455 }, { "epoch": 0.24619501588894463, "grad_norm": 8.326447292688538, "learning_rate": 8.19964349376114e-07, "logits/chosen": -0.16184619069099426, "logits/rejected": -0.051874689757823944, "logps/chosen": -1.3679503202438354, "logps/rejected": -1.4119255542755127, "loss": 1.7356, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.3679503202438354, "rewards/margins": 0.0439753457903862, "rewards/rejected": -1.4119255542755127, "step": 460 }, { "epoch": 0.2488710486703462, "grad_norm": 9.473883720691502, "learning_rate": 8.288770053475936e-07, "logits/chosen": 0.11290111392736435, "logits/rejected": 0.12598662078380585, "logps/chosen": -1.2719142436981201, "logps/rejected": -1.4320366382598877, "loss": 1.6722, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2719142436981201, "rewards/margins": 0.160122349858284, "rewards/rejected": -1.4320366382598877, "step": 465 }, { "epoch": 0.2515470814517478, "grad_norm": 9.555764432813708, "learning_rate": 8.37789661319073e-07, "logits/chosen": 0.1309710443019867, "logits/rejected": 0.07963626086711884, "logps/chosen": -1.234974980354309, "logps/rejected": -1.4127203226089478, "loss": 1.6056, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.234974980354309, "rewards/margins": 0.17774544656276703, "rewards/rejected": -1.4127203226089478, "step": 470 }, { "epoch": 0.25422311423314936, "grad_norm": 6.846149735650059, "learning_rate": 8.467023172905525e-07, "logits/chosen": -0.0759991854429245, "logits/rejected": 0.0541856586933136, "logps/chosen": -1.299948811531067, "logps/rejected": -1.4895131587982178, "loss": 1.6333, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.299948811531067, "rewards/margins": 0.18956440687179565, "rewards/rejected": -1.4895131587982178, "step": 475 }, { "epoch": 0.2568991470145509, "grad_norm": 12.67935847973589, "learning_rate": 8.55614973262032e-07, "logits/chosen": -0.06495614349842072, "logits/rejected": 0.12569716572761536, "logps/chosen": -1.2725584506988525, "logps/rejected": -1.3375532627105713, "loss": 1.7014, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2725584506988525, "rewards/margins": 0.06499501317739487, "rewards/rejected": -1.3375532627105713, "step": 480 }, { "epoch": 0.2595751797959525, "grad_norm": 11.367772552764144, "learning_rate": 8.645276292335115e-07, "logits/chosen": -0.0036151937674731016, "logits/rejected": 0.030773457139730453, "logps/chosen": -1.3656898736953735, "logps/rejected": -1.435654878616333, "loss": 1.7304, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3656898736953735, "rewards/margins": 0.06996499001979828, "rewards/rejected": -1.435654878616333, "step": 485 }, { "epoch": 0.26225121257735406, "grad_norm": 8.665607148566286, "learning_rate": 8.734402852049911e-07, "logits/chosen": -0.01699291542172432, "logits/rejected": 0.05099986121058464, "logps/chosen": -1.302870750427246, "logps/rejected": -1.3726005554199219, "loss": 1.7032, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.302870750427246, "rewards/margins": 0.06972992420196533, "rewards/rejected": -1.3726005554199219, "step": 490 }, { "epoch": 0.26492724535875567, "grad_norm": 9.49301727299634, "learning_rate": 8.823529411764705e-07, "logits/chosen": -0.05295737460255623, "logits/rejected": -0.03520811349153519, "logps/chosen": -1.3093057870864868, "logps/rejected": -1.4187067747116089, "loss": 1.6828, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3093057870864868, "rewards/margins": 0.10940104722976685, "rewards/rejected": -1.4187067747116089, "step": 495 }, { "epoch": 0.2676032781401572, "grad_norm": 8.428671515813905, "learning_rate": 8.912655971479501e-07, "logits/chosen": -0.04793981835246086, "logits/rejected": 0.04994889721274376, "logps/chosen": -1.2275326251983643, "logps/rejected": -1.379403829574585, "loss": 1.6185, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2275326251983643, "rewards/margins": 0.15187107026576996, "rewards/rejected": -1.379403829574585, "step": 500 }, { "epoch": 0.27027931092155877, "grad_norm": 7.9895540595139956, "learning_rate": 9.001782531194295e-07, "logits/chosen": -0.062299322336912155, "logits/rejected": 0.07209259271621704, "logps/chosen": -1.3436644077301025, "logps/rejected": -1.377514362335205, "loss": 1.6953, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3436644077301025, "rewards/margins": 0.03384983539581299, "rewards/rejected": -1.377514362335205, "step": 505 }, { "epoch": 0.2729553437029604, "grad_norm": 8.131935548042284, "learning_rate": 9.09090909090909e-07, "logits/chosen": 0.07910905033349991, "logits/rejected": 0.13863492012023926, "logps/chosen": -1.3027557134628296, "logps/rejected": -1.454535722732544, "loss": 1.6514, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3027557134628296, "rewards/margins": 0.15178008377552032, "rewards/rejected": -1.454535722732544, "step": 510 }, { "epoch": 0.2756313764843619, "grad_norm": 8.590703247363546, "learning_rate": 9.180035650623885e-07, "logits/chosen": 0.050558000802993774, "logits/rejected": 0.14124636352062225, "logps/chosen": -1.2460167407989502, "logps/rejected": -1.3840495347976685, "loss": 1.6088, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2460167407989502, "rewards/margins": 0.1380326896905899, "rewards/rejected": -1.3840495347976685, "step": 515 }, { "epoch": 0.27830740926576353, "grad_norm": 6.893933372683863, "learning_rate": 9.26916221033868e-07, "logits/chosen": -0.08430425822734833, "logits/rejected": 0.04368329793214798, "logps/chosen": -1.2867071628570557, "logps/rejected": -1.3750287294387817, "loss": 1.695, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2867071628570557, "rewards/margins": 0.08832161128520966, "rewards/rejected": -1.3750287294387817, "step": 520 }, { "epoch": 0.2809834420471651, "grad_norm": 19.67350727764414, "learning_rate": 9.358288770053476e-07, "logits/chosen": 0.09718038141727448, "logits/rejected": 0.16004881262779236, "logps/chosen": -1.264098882675171, "logps/rejected": -1.4222582578659058, "loss": 1.6412, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.264098882675171, "rewards/margins": 0.1581595242023468, "rewards/rejected": -1.4222582578659058, "step": 525 }, { "epoch": 0.2836594748285666, "grad_norm": 6.274997373019293, "learning_rate": 9.44741532976827e-07, "logits/chosen": 0.0687488541007042, "logits/rejected": 0.14788378775119781, "logps/chosen": -1.2366673946380615, "logps/rejected": -1.3275758028030396, "loss": 1.6578, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2366673946380615, "rewards/margins": 0.09090828895568848, "rewards/rejected": -1.3275758028030396, "step": 530 }, { "epoch": 0.28633550760996823, "grad_norm": 7.0945040486516175, "learning_rate": 9.536541889483066e-07, "logits/chosen": -0.10038735717535019, "logits/rejected": 0.14553984999656677, "logps/chosen": -1.2521215677261353, "logps/rejected": -1.3115968704223633, "loss": 1.6602, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2521215677261353, "rewards/margins": 0.059475433081388474, "rewards/rejected": -1.3115968704223633, "step": 535 }, { "epoch": 0.2890115403913698, "grad_norm": 7.789872109139343, "learning_rate": 9.62566844919786e-07, "logits/chosen": 0.016909483820199966, "logits/rejected": 0.08182911574840546, "logps/chosen": -1.378409743309021, "logps/rejected": -1.4266738891601562, "loss": 1.7337, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.378409743309021, "rewards/margins": 0.04826412349939346, "rewards/rejected": -1.4266738891601562, "step": 540 }, { "epoch": 0.2916875731727714, "grad_norm": 7.359174830858011, "learning_rate": 9.714795008912655e-07, "logits/chosen": -0.10508112609386444, "logits/rejected": 0.08393336832523346, "logps/chosen": -1.2827831506729126, "logps/rejected": -1.3865883350372314, "loss": 1.6549, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2827831506729126, "rewards/margins": 0.10380531847476959, "rewards/rejected": -1.3865883350372314, "step": 545 }, { "epoch": 0.29436360595417294, "grad_norm": 8.778259311834221, "learning_rate": 9.80392156862745e-07, "logits/chosen": 0.04168703407049179, "logits/rejected": 0.10389252007007599, "logps/chosen": -1.280343770980835, "logps/rejected": -1.402133822441101, "loss": 1.6367, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.280343770980835, "rewards/margins": 0.12179000675678253, "rewards/rejected": -1.402133822441101, "step": 550 }, { "epoch": 0.2970396387355745, "grad_norm": 11.267802618540536, "learning_rate": 9.893048128342244e-07, "logits/chosen": -0.07766524702310562, "logits/rejected": 0.04163938760757446, "logps/chosen": -1.356514573097229, "logps/rejected": -1.4130576848983765, "loss": 1.7217, "rewards/accuracies": 0.53125, "rewards/chosen": -1.356514573097229, "rewards/margins": 0.056543171405792236, "rewards/rejected": -1.4130576848983765, "step": 555 }, { "epoch": 0.2997156715169761, "grad_norm": 9.734534917386588, "learning_rate": 9.98217468805704e-07, "logits/chosen": 0.0625699833035469, "logits/rejected": 0.07177621126174927, "logps/chosen": -1.2119487524032593, "logps/rejected": -1.33185875415802, "loss": 1.6101, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2119487524032593, "rewards/margins": 0.11990992724895477, "rewards/rejected": -1.33185875415802, "step": 560 }, { "epoch": 0.30239170429837764, "grad_norm": 6.1694254592049536, "learning_rate": 9.999984476788462e-07, "logits/chosen": 0.030029237270355225, "logits/rejected": 0.07460276782512665, "logps/chosen": -1.3244996070861816, "logps/rejected": -1.4376353025436401, "loss": 1.6988, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3244996070861816, "rewards/margins": 0.11313574016094208, "rewards/rejected": -1.4376353025436401, "step": 565 }, { "epoch": 0.30506773707977924, "grad_norm": 8.854668170263972, "learning_rate": 9.999921413906797e-07, "logits/chosen": -0.04299437254667282, "logits/rejected": 0.1687401980161667, "logps/chosen": -1.3059120178222656, "logps/rejected": -1.37355637550354, "loss": 1.702, "rewards/accuracies": 0.5, "rewards/chosen": -1.3059120178222656, "rewards/margins": 0.0676443800330162, "rewards/rejected": -1.37355637550354, "step": 570 }, { "epoch": 0.3077437698611808, "grad_norm": 6.858926964971097, "learning_rate": 9.999809841765644e-07, "logits/chosen": -0.0256123635917902, "logits/rejected": 0.029277730733156204, "logps/chosen": -1.2328436374664307, "logps/rejected": -1.3388514518737793, "loss": 1.6605, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2328436374664307, "rewards/margins": 0.1060078889131546, "rewards/rejected": -1.3388514518737793, "step": 575 }, { "epoch": 0.3104198026425824, "grad_norm": 7.451636113265376, "learning_rate": 9.999649761447477e-07, "logits/chosen": -0.03498077020049095, "logits/rejected": 0.11152464151382446, "logps/chosen": -1.2289667129516602, "logps/rejected": -1.3867911100387573, "loss": 1.5999, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2289667129516602, "rewards/margins": 0.1578243523836136, "rewards/rejected": -1.3867911100387573, "step": 580 }, { "epoch": 0.31309583542398395, "grad_norm": 7.454532761341038, "learning_rate": 9.999441174505398e-07, "logits/chosen": -0.07113450020551682, "logits/rejected": 0.022057075053453445, "logps/chosen": -1.3707550764083862, "logps/rejected": -1.412688136100769, "loss": 1.7413, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3707550764083862, "rewards/margins": 0.04193300008773804, "rewards/rejected": -1.412688136100769, "step": 585 }, { "epoch": 0.3157718682053855, "grad_norm": 15.759533284704874, "learning_rate": 9.999184082963116e-07, "logits/chosen": -0.02356426976621151, "logits/rejected": 0.09675613045692444, "logps/chosen": -1.3437496423721313, "logps/rejected": -1.3699607849121094, "loss": 1.7185, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3437496423721313, "rewards/margins": 0.0262112058699131, "rewards/rejected": -1.3699607849121094, "step": 590 }, { "epoch": 0.3184479009867871, "grad_norm": 8.753436869586725, "learning_rate": 9.998878489314937e-07, "logits/chosen": 0.001271332846954465, "logits/rejected": 0.118097685277462, "logps/chosen": -1.275848150253296, "logps/rejected": -1.3325821161270142, "loss": 1.681, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -1.275848150253296, "rewards/margins": 0.05673399567604065, "rewards/rejected": -1.3325821161270142, "step": 595 }, { "epoch": 0.32112393376818865, "grad_norm": 8.206758711816612, "learning_rate": 9.99852439652573e-07, "logits/chosen": -0.05460699647665024, "logits/rejected": 0.08076707273721695, "logps/chosen": -1.2695443630218506, "logps/rejected": -1.2824230194091797, "loss": 1.6864, "rewards/accuracies": 0.46875, "rewards/chosen": -1.2695443630218506, "rewards/margins": 0.012878738343715668, "rewards/rejected": -1.2824230194091797, "step": 600 }, { "epoch": 0.32379996654959026, "grad_norm": 9.827295432839174, "learning_rate": 9.998121808030904e-07, "logits/chosen": -0.08948203921318054, "logits/rejected": -0.007546691689640284, "logps/chosen": -1.330176591873169, "logps/rejected": -1.4655649662017822, "loss": 1.6795, "rewards/accuracies": 0.59375, "rewards/chosen": -1.330176591873169, "rewards/margins": 0.1353883445262909, "rewards/rejected": -1.4655649662017822, "step": 605 }, { "epoch": 0.3264759993309918, "grad_norm": 12.364970388660888, "learning_rate": 9.997670727736379e-07, "logits/chosen": 0.0591292604804039, "logits/rejected": 0.2056918442249298, "logps/chosen": -1.308626651763916, "logps/rejected": -1.3645381927490234, "loss": 1.7032, "rewards/accuracies": 0.53125, "rewards/chosen": -1.308626651763916, "rewards/margins": 0.05591156333684921, "rewards/rejected": -1.3645381927490234, "step": 610 }, { "epoch": 0.32915203211239336, "grad_norm": 5.644018971156467, "learning_rate": 9.99717116001853e-07, "logits/chosen": -0.05800892040133476, "logits/rejected": 0.04104765132069588, "logps/chosen": -1.3062307834625244, "logps/rejected": -1.432071328163147, "loss": 1.6772, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3062307834625244, "rewards/margins": 0.1258404701948166, "rewards/rejected": -1.432071328163147, "step": 615 }, { "epoch": 0.33182806489379496, "grad_norm": 7.3898691520823725, "learning_rate": 9.996623109724173e-07, "logits/chosen": 0.05058063194155693, "logits/rejected": 0.11339068412780762, "logps/chosen": -1.3660705089569092, "logps/rejected": -1.4618982076644897, "loss": 1.7012, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.3660705089569092, "rewards/margins": 0.09582777321338654, "rewards/rejected": -1.4618982076644897, "step": 620 }, { "epoch": 0.3345040976751965, "grad_norm": 8.959693793832603, "learning_rate": 9.996026582170488e-07, "logits/chosen": 0.06929932534694672, "logits/rejected": 0.1785743236541748, "logps/chosen": -1.2847315073013306, "logps/rejected": -1.4012258052825928, "loss": 1.6582, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2847315073013306, "rewards/margins": 0.11649449169635773, "rewards/rejected": -1.4012258052825928, "step": 625 }, { "epoch": 0.3371801304565981, "grad_norm": 7.628006754940694, "learning_rate": 9.995381583144996e-07, "logits/chosen": -0.01316717267036438, "logits/rejected": 0.08945786952972412, "logps/chosen": -1.3198153972625732, "logps/rejected": -1.448467493057251, "loss": 1.6546, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3198153972625732, "rewards/margins": 0.12865200638771057, "rewards/rejected": -1.448467493057251, "step": 630 }, { "epoch": 0.33985616323799966, "grad_norm": 6.656545125521265, "learning_rate": 9.994688118905471e-07, "logits/chosen": -0.012319590896368027, "logits/rejected": 0.21458666026592255, "logps/chosen": -1.3832319974899292, "logps/rejected": -1.4100427627563477, "loss": 1.7693, "rewards/accuracies": 0.44999998807907104, "rewards/chosen": -1.3832319974899292, "rewards/margins": 0.0268106609582901, "rewards/rejected": -1.4100427627563477, "step": 635 }, { "epoch": 0.3425321960194012, "grad_norm": 8.970597384974177, "learning_rate": 9.993946196179912e-07, "logits/chosen": -0.08106163144111633, "logits/rejected": 0.10695508867502213, "logps/chosen": -1.3231474161148071, "logps/rejected": -1.4096291065216064, "loss": 1.7402, "rewards/accuracies": 0.45625001192092896, "rewards/chosen": -1.3231474161148071, "rewards/margins": 0.08648180961608887, "rewards/rejected": -1.4096291065216064, "step": 640 }, { "epoch": 0.3452082288008028, "grad_norm": 7.346230858532709, "learning_rate": 9.993155822166455e-07, "logits/chosen": -0.08081404119729996, "logits/rejected": 0.0013829886447638273, "logps/chosen": -1.2329657077789307, "logps/rejected": -1.3853070735931396, "loss": 1.6505, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2329657077789307, "rewards/margins": 0.1523413360118866, "rewards/rejected": -1.3853070735931396, "step": 645 }, { "epoch": 0.34788426158220437, "grad_norm": 10.907122686264634, "learning_rate": 9.992317004533313e-07, "logits/chosen": -0.04056433588266373, "logits/rejected": 0.09426826983690262, "logps/chosen": -1.3744598627090454, "logps/rejected": -1.517823338508606, "loss": 1.713, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3744598627090454, "rewards/margins": 0.14336350560188293, "rewards/rejected": -1.517823338508606, "step": 650 }, { "epoch": 0.350560294363606, "grad_norm": 8.246432749787882, "learning_rate": 9.991429751418696e-07, "logits/chosen": 0.024029621854424477, "logits/rejected": 0.028746861964464188, "logps/chosen": -1.3122899532318115, "logps/rejected": -1.4840376377105713, "loss": 1.6704, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.3122899532318115, "rewards/margins": 0.17174772918224335, "rewards/rejected": -1.4840376377105713, "step": 655 }, { "epoch": 0.3532363271450075, "grad_norm": 7.0970140924944225, "learning_rate": 9.99049407143074e-07, "logits/chosen": 0.014008665457367897, "logits/rejected": 0.13609251379966736, "logps/chosen": -1.270634412765503, "logps/rejected": -1.2979482412338257, "loss": 1.6808, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.270634412765503, "rewards/margins": 0.027314016595482826, "rewards/rejected": -1.2979482412338257, "step": 660 }, { "epoch": 0.35591235992640907, "grad_norm": 7.628095065228471, "learning_rate": 9.989509973647416e-07, "logits/chosen": -0.008538028225302696, "logits/rejected": 0.12382189929485321, "logps/chosen": -1.2361500263214111, "logps/rejected": -1.3574538230895996, "loss": 1.6301, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2361500263214111, "rewards/margins": 0.1213037520647049, "rewards/rejected": -1.3574538230895996, "step": 665 }, { "epoch": 0.3585883927078107, "grad_norm": 8.552725286884353, "learning_rate": 9.988477467616445e-07, "logits/chosen": -0.02346823923289776, "logits/rejected": 0.17321622371673584, "logps/chosen": -1.2735487222671509, "logps/rejected": -1.3107669353485107, "loss": 1.669, "rewards/accuracies": 0.46875, "rewards/chosen": -1.2735487222671509, "rewards/margins": 0.0372183695435524, "rewards/rejected": -1.3107669353485107, "step": 670 }, { "epoch": 0.3612644254892122, "grad_norm": 10.528437312230285, "learning_rate": 9.987396563355205e-07, "logits/chosen": -0.036085594445466995, "logits/rejected": 0.03549571335315704, "logps/chosen": -1.2650729417800903, "logps/rejected": -1.4736168384552002, "loss": 1.6303, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2650729417800903, "rewards/margins": 0.20854397118091583, "rewards/rejected": -1.4736168384552002, "step": 675 }, { "epoch": 0.36394045827061383, "grad_norm": 9.271902916867035, "learning_rate": 9.986267271350631e-07, "logits/chosen": 0.06981401890516281, "logits/rejected": 0.22414882481098175, "logps/chosen": -1.3095825910568237, "logps/rejected": -1.3565170764923096, "loss": 1.7206, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.3095825910568237, "rewards/margins": 0.04693450406193733, "rewards/rejected": -1.3565170764923096, "step": 680 }, { "epoch": 0.3666164910520154, "grad_norm": 9.197512258817639, "learning_rate": 9.985089602559123e-07, "logits/chosen": 0.020719774067401886, "logits/rejected": 0.17003081738948822, "logps/chosen": -1.2913005352020264, "logps/rejected": -1.3424879312515259, "loss": 1.6958, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2913005352020264, "rewards/margins": 0.051187384873628616, "rewards/rejected": -1.3424879312515259, "step": 685 }, { "epoch": 0.369292523833417, "grad_norm": 8.633540226735317, "learning_rate": 9.983863568406428e-07, "logits/chosen": 0.03548329323530197, "logits/rejected": 0.0648711770772934, "logps/chosen": -1.2927361726760864, "logps/rejected": -1.4053599834442139, "loss": 1.6573, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2927361726760864, "rewards/margins": 0.11262372881174088, "rewards/rejected": -1.4053599834442139, "step": 690 }, { "epoch": 0.37196855661481854, "grad_norm": 7.219278802274095, "learning_rate": 9.982589180787532e-07, "logits/chosen": -0.00833642203360796, "logits/rejected": 0.07940270006656647, "logps/chosen": -1.1803183555603027, "logps/rejected": -1.3500235080718994, "loss": 1.5644, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.1803183555603027, "rewards/margins": 0.16970540583133698, "rewards/rejected": -1.3500235080718994, "step": 695 }, { "epoch": 0.3746445893962201, "grad_norm": 6.240252885623886, "learning_rate": 9.981266452066553e-07, "logits/chosen": -0.10447759926319122, "logits/rejected": 0.024963444098830223, "logps/chosen": -1.3531849384307861, "logps/rejected": -1.4271669387817383, "loss": 1.7094, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3531849384307861, "rewards/margins": 0.07398195564746857, "rewards/rejected": -1.4271669387817383, "step": 700 }, { "epoch": 0.3773206221776217, "grad_norm": 6.504425773436158, "learning_rate": 9.979895395076608e-07, "logits/chosen": -0.09504284709692001, "logits/rejected": 0.06818296760320663, "logps/chosen": -1.3101884126663208, "logps/rejected": -1.4512856006622314, "loss": 1.6811, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3101884126663208, "rewards/margins": 0.1410975158214569, "rewards/rejected": -1.4512856006622314, "step": 705 }, { "epoch": 0.37999665495902324, "grad_norm": 7.997459613296311, "learning_rate": 9.9784760231197e-07, "logits/chosen": 0.028507554903626442, "logits/rejected": 0.1171996220946312, "logps/chosen": -1.2455408573150635, "logps/rejected": -1.3672510385513306, "loss": 1.6293, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2455408573150635, "rewards/margins": 0.12171021848917007, "rewards/rejected": -1.3672510385513306, "step": 710 }, { "epoch": 0.38267268774042484, "grad_norm": 8.537076673133495, "learning_rate": 9.97700834996658e-07, "logits/chosen": -0.02836451306939125, "logits/rejected": 0.126664400100708, "logps/chosen": -1.3226584196090698, "logps/rejected": -1.4415714740753174, "loss": 1.7028, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3226584196090698, "rewards/margins": 0.11891325563192368, "rewards/rejected": -1.4415714740753174, "step": 715 }, { "epoch": 0.3853487205218264, "grad_norm": 7.544234541581894, "learning_rate": 9.97549238985662e-07, "logits/chosen": 0.041222043335437775, "logits/rejected": 0.2183455228805542, "logps/chosen": -1.3687158823013306, "logps/rejected": -1.4198092222213745, "loss": 1.7279, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3687158823013306, "rewards/margins": 0.05109328031539917, "rewards/rejected": -1.4198092222213745, "step": 720 }, { "epoch": 0.38802475330322794, "grad_norm": 9.122685189648141, "learning_rate": 9.973928157497674e-07, "logits/chosen": -0.0802018791437149, "logits/rejected": 0.049556490033864975, "logps/chosen": -1.2097431421279907, "logps/rejected": -1.4569720029830933, "loss": 1.5884, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2097431421279907, "rewards/margins": 0.24722866714000702, "rewards/rejected": -1.4569720029830933, "step": 725 }, { "epoch": 0.39070078608462955, "grad_norm": 9.481217491829481, "learning_rate": 9.972315668065927e-07, "logits/chosen": -0.09887029230594635, "logits/rejected": 0.06117262318730354, "logps/chosen": -1.3061532974243164, "logps/rejected": -1.3930120468139648, "loss": 1.6769, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3061532974243164, "rewards/margins": 0.08685862272977829, "rewards/rejected": -1.3930120468139648, "step": 730 }, { "epoch": 0.3933768188660311, "grad_norm": 6.656490793599791, "learning_rate": 9.97065493720576e-07, "logits/chosen": -0.0811881572008133, "logits/rejected": 0.014002988114953041, "logps/chosen": -1.328331708908081, "logps/rejected": -1.384516716003418, "loss": 1.6862, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.328331708908081, "rewards/margins": 0.05618508532643318, "rewards/rejected": -1.384516716003418, "step": 735 }, { "epoch": 0.3960528516474327, "grad_norm": 10.559224036803714, "learning_rate": 9.968945981029594e-07, "logits/chosen": -0.04994923993945122, "logits/rejected": 0.1147971898317337, "logps/chosen": -1.3782548904418945, "logps/rejected": -1.4143259525299072, "loss": 1.748, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3782548904418945, "rewards/margins": 0.03607102110981941, "rewards/rejected": -1.4143259525299072, "step": 740 }, { "epoch": 0.39872888442883425, "grad_norm": 7.189987022376075, "learning_rate": 9.967188816117726e-07, "logits/chosen": 0.036188650876283646, "logits/rejected": 0.09928051382303238, "logps/chosen": -1.359431266784668, "logps/rejected": -1.5481358766555786, "loss": 1.7021, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.359431266784668, "rewards/margins": 0.18870458006858826, "rewards/rejected": -1.5481358766555786, "step": 745 }, { "epoch": 0.4014049172102358, "grad_norm": 6.989374751704205, "learning_rate": 9.965383459518179e-07, "logits/chosen": -0.022238023579120636, "logits/rejected": 0.13665586709976196, "logps/chosen": -1.2956759929656982, "logps/rejected": -1.4461919069290161, "loss": 1.6496, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2956759929656982, "rewards/margins": 0.15051576495170593, "rewards/rejected": -1.4461919069290161, "step": 750 }, { "epoch": 0.4040809499916374, "grad_norm": 5.70105952630306, "learning_rate": 9.963529928746533e-07, "logits/chosen": 0.004465815611183643, "logits/rejected": 0.12207271158695221, "logps/chosen": -1.327978491783142, "logps/rejected": -1.3998768329620361, "loss": 1.6933, "rewards/accuracies": 0.5, "rewards/chosen": -1.327978491783142, "rewards/margins": 0.07189829647541046, "rewards/rejected": -1.3998768329620361, "step": 755 }, { "epoch": 0.40675698277303896, "grad_norm": 5.94205764786591, "learning_rate": 9.961628241785746e-07, "logits/chosen": -0.06844659149646759, "logits/rejected": -0.002805382013320923, "logps/chosen": -1.3424692153930664, "logps/rejected": -1.4756808280944824, "loss": 1.6882, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3424692153930664, "rewards/margins": 0.13321153819561005, "rewards/rejected": -1.4756808280944824, "step": 760 }, { "epoch": 0.40943301555444056, "grad_norm": 14.34219789446751, "learning_rate": 9.959678417085998e-07, "logits/chosen": -0.04711277782917023, "logits/rejected": 0.03717537969350815, "logps/chosen": -1.2987074851989746, "logps/rejected": -1.3974730968475342, "loss": 1.659, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2987074851989746, "rewards/margins": 0.09876558184623718, "rewards/rejected": -1.3974730968475342, "step": 765 }, { "epoch": 0.4121090483358421, "grad_norm": 7.668216776232713, "learning_rate": 9.957680473564493e-07, "logits/chosen": 0.07918961346149445, "logits/rejected": 0.1951749622821808, "logps/chosen": -1.257340669631958, "logps/rejected": -1.4603685140609741, "loss": 1.6128, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.257340669631958, "rewards/margins": 0.2030281275510788, "rewards/rejected": -1.4603685140609741, "step": 770 }, { "epoch": 0.41478508111724366, "grad_norm": 9.20239361618736, "learning_rate": 9.95563443060529e-07, "logits/chosen": -0.07869169116020203, "logits/rejected": 0.08225543797016144, "logps/chosen": -1.3372876644134521, "logps/rejected": -1.5053128004074097, "loss": 1.6958, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3372876644134521, "rewards/margins": 0.16802528500556946, "rewards/rejected": -1.5053128004074097, "step": 775 }, { "epoch": 0.41746111389864526, "grad_norm": 7.2244031080730755, "learning_rate": 9.95354030805911e-07, "logits/chosen": -0.1303524672985077, "logits/rejected": 0.009751476347446442, "logps/chosen": -1.2609707117080688, "logps/rejected": -1.414001226425171, "loss": 1.6255, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2609707117080688, "rewards/margins": 0.1530303955078125, "rewards/rejected": -1.414001226425171, "step": 780 }, { "epoch": 0.4201371466800468, "grad_norm": 8.457832296987833, "learning_rate": 9.951398126243133e-07, "logits/chosen": 0.022420603781938553, "logits/rejected": 0.1414371132850647, "logps/chosen": -1.2476838827133179, "logps/rejected": -1.4334955215454102, "loss": 1.6246, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2476838827133179, "rewards/margins": 0.18581156432628632, "rewards/rejected": -1.4334955215454102, "step": 785 }, { "epoch": 0.4228131794614484, "grad_norm": 7.083834282044994, "learning_rate": 9.94920790594082e-07, "logits/chosen": -0.04662860929965973, "logits/rejected": 0.06980902701616287, "logps/chosen": -1.3007665872573853, "logps/rejected": -1.3650496006011963, "loss": 1.6828, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3007665872573853, "rewards/margins": 0.0642831102013588, "rewards/rejected": -1.3650496006011963, "step": 790 }, { "epoch": 0.42548921224284997, "grad_norm": 5.925629002132357, "learning_rate": 9.946969668401696e-07, "logits/chosen": -0.07388118654489517, "logits/rejected": 0.1055014580488205, "logps/chosen": -1.2804076671600342, "logps/rejected": -1.4107162952423096, "loss": 1.6566, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2804076671600342, "rewards/margins": 0.1303088366985321, "rewards/rejected": -1.4107162952423096, "step": 795 }, { "epoch": 0.4281652450242516, "grad_norm": 7.440409871322956, "learning_rate": 9.944683435341155e-07, "logits/chosen": -0.029027258977293968, "logits/rejected": 0.046039216220378876, "logps/chosen": -1.2842944860458374, "logps/rejected": -1.3327921628952026, "loss": 1.6693, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2842944860458374, "rewards/margins": 0.048497579991817474, "rewards/rejected": -1.3327921628952026, "step": 800 }, { "epoch": 0.4281652450242516, "eval_logits/chosen": 0.2642241418361664, "eval_logits/rejected": 0.3440663516521454, "eval_logps/chosen": -1.3149185180664062, "eval_logps/rejected": -1.4532232284545898, "eval_loss": 1.671885371208191, "eval_rewards/accuracies": 0.5578634738922119, "eval_rewards/chosen": -1.3149185180664062, "eval_rewards/margins": 0.13830474019050598, "eval_rewards/rejected": -1.4532232284545898, "eval_runtime": 40.5458, "eval_samples_per_second": 33.172, "eval_steps_per_second": 8.312, "step": 800 }, { "epoch": 0.4308412778056531, "grad_norm": 7.508015672817706, "learning_rate": 9.942349228940236e-07, "logits/chosen": -0.0667061060667038, "logits/rejected": 0.07563512772321701, "logps/chosen": -1.3263849020004272, "logps/rejected": -1.476535439491272, "loss": 1.68, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.3263849020004272, "rewards/margins": 0.1501503884792328, "rewards/rejected": -1.476535439491272, "step": 805 }, { "epoch": 0.43351731058705467, "grad_norm": 7.655604039956669, "learning_rate": 9.939967071845424e-07, "logits/chosen": 0.02369997464120388, "logits/rejected": 0.09098490327596664, "logps/chosen": -1.2372820377349854, "logps/rejected": -1.3665462732315063, "loss": 1.6153, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2372820377349854, "rewards/margins": 0.12926435470581055, "rewards/rejected": -1.3665462732315063, "step": 810 }, { "epoch": 0.4361933433684563, "grad_norm": 8.11934052284553, "learning_rate": 9.937536987168413e-07, "logits/chosen": 0.04012486711144447, "logits/rejected": 0.1624753475189209, "logps/chosen": -1.2311725616455078, "logps/rejected": -1.4302477836608887, "loss": 1.5796, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2311725616455078, "rewards/margins": 0.19907523691654205, "rewards/rejected": -1.4302477836608887, "step": 815 }, { "epoch": 0.4388693761498578, "grad_norm": 7.76861185601798, "learning_rate": 9.935058998485896e-07, "logits/chosen": 0.035312581807374954, "logits/rejected": 0.0773046463727951, "logps/chosen": -1.277048110961914, "logps/rejected": -1.4330211877822876, "loss": 1.6463, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.277048110961914, "rewards/margins": 0.15597307682037354, "rewards/rejected": -1.4330211877822876, "step": 820 }, { "epoch": 0.44154540893125943, "grad_norm": 7.860471019420282, "learning_rate": 9.932533129839333e-07, "logits/chosen": -0.014185379259288311, "logits/rejected": 0.10354729741811752, "logps/chosen": -1.2197961807250977, "logps/rejected": -1.306098222732544, "loss": 1.6297, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.2197961807250977, "rewards/margins": 0.08630184829235077, "rewards/rejected": -1.306098222732544, "step": 825 }, { "epoch": 0.444221441712661, "grad_norm": 6.980346415002472, "learning_rate": 9.929959405734711e-07, "logits/chosen": 0.044074997305870056, "logits/rejected": 0.19551844894886017, "logps/chosen": -1.3337340354919434, "logps/rejected": -1.3899133205413818, "loss": 1.7003, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3337340354919434, "rewards/margins": 0.0561792366206646, "rewards/rejected": -1.3899133205413818, "step": 830 }, { "epoch": 0.44689747449406253, "grad_norm": 10.643519397721393, "learning_rate": 9.927337851142314e-07, "logits/chosen": -0.013499056920409203, "logits/rejected": 0.10668818652629852, "logps/chosen": -1.2430568933486938, "logps/rejected": -1.3656256198883057, "loss": 1.6498, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2430568933486938, "rewards/margins": 0.12256884574890137, "rewards/rejected": -1.3656256198883057, "step": 835 }, { "epoch": 0.44957350727546413, "grad_norm": 8.177028406683556, "learning_rate": 9.924668491496474e-07, "logits/chosen": -0.004524660296738148, "logits/rejected": 0.1376916915178299, "logps/chosen": -1.284436583518982, "logps/rejected": -1.4487638473510742, "loss": 1.666, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.284436583518982, "rewards/margins": 0.1643272340297699, "rewards/rejected": -1.4487638473510742, "step": 840 }, { "epoch": 0.4522495400568657, "grad_norm": 5.537721258154138, "learning_rate": 9.92195135269533e-07, "logits/chosen": 0.06299177557229996, "logits/rejected": 0.12591898441314697, "logps/chosen": -1.288023829460144, "logps/rejected": -1.341961145401001, "loss": 1.6896, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.288023829460144, "rewards/margins": 0.05393745377659798, "rewards/rejected": -1.341961145401001, "step": 845 }, { "epoch": 0.4549255728382673, "grad_norm": 9.106385983775208, "learning_rate": 9.919186461100574e-07, "logits/chosen": 0.0094298105686903, "logits/rejected": 0.06498949229717255, "logps/chosen": -1.248669147491455, "logps/rejected": -1.3809727430343628, "loss": 1.6115, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.248669147491455, "rewards/margins": 0.13230359554290771, "rewards/rejected": -1.3809727430343628, "step": 850 }, { "epoch": 0.45760160561966884, "grad_norm": 6.43224040221898, "learning_rate": 9.9163738435372e-07, "logits/chosen": -0.04182653874158859, "logits/rejected": 0.084588423371315, "logps/chosen": -1.3051223754882812, "logps/rejected": -1.4761320352554321, "loss": 1.6954, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3051223754882812, "rewards/margins": 0.17100976407527924, "rewards/rejected": -1.4761320352554321, "step": 855 }, { "epoch": 0.4602776384010704, "grad_norm": 5.868384556976196, "learning_rate": 9.913513527293234e-07, "logits/chosen": -0.06548645347356796, "logits/rejected": 0.07703325897455215, "logps/chosen": -1.3435872793197632, "logps/rejected": -1.5250961780548096, "loss": 1.6786, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3435872793197632, "rewards/margins": 0.1815088838338852, "rewards/rejected": -1.5250961780548096, "step": 860 }, { "epoch": 0.462953671182472, "grad_norm": 10.824301867524007, "learning_rate": 9.910605540119474e-07, "logits/chosen": -0.002961373422294855, "logits/rejected": 0.08694703131914139, "logps/chosen": -1.2438809871673584, "logps/rejected": -1.4252790212631226, "loss": 1.6348, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2438809871673584, "rewards/margins": 0.18139800429344177, "rewards/rejected": -1.4252790212631226, "step": 865 }, { "epoch": 0.46562970396387354, "grad_norm": 7.012684753512225, "learning_rate": 9.907649910229227e-07, "logits/chosen": -0.08534906059503555, "logits/rejected": 0.15127788484096527, "logps/chosen": -1.3012317419052124, "logps/rejected": -1.3915889263153076, "loss": 1.676, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3012317419052124, "rewards/margins": 0.09035725891590118, "rewards/rejected": -1.3915889263153076, "step": 870 }, { "epoch": 0.46830573674527515, "grad_norm": 7.511187500576787, "learning_rate": 9.90464666629803e-07, "logits/chosen": 0.022366385906934738, "logits/rejected": 0.08713110536336899, "logps/chosen": -1.3248341083526611, "logps/rejected": -1.4425128698349, "loss": 1.7236, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3248341083526611, "rewards/margins": 0.11767883598804474, "rewards/rejected": -1.4425128698349, "step": 875 }, { "epoch": 0.4709817695266767, "grad_norm": 6.033858000811316, "learning_rate": 9.901595837463363e-07, "logits/chosen": 0.012679760344326496, "logits/rejected": 0.15422797203063965, "logps/chosen": -1.3769097328186035, "logps/rejected": -1.4846652746200562, "loss": 1.7246, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3769097328186035, "rewards/margins": 0.10775542259216309, "rewards/rejected": -1.4846652746200562, "step": 880 }, { "epoch": 0.47365780230807825, "grad_norm": 8.760506894298192, "learning_rate": 9.898497453324384e-07, "logits/chosen": -0.054704517126083374, "logits/rejected": 0.016666647046804428, "logps/chosen": -1.2689878940582275, "logps/rejected": -1.4531781673431396, "loss": 1.6311, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2689878940582275, "rewards/margins": 0.18419024348258972, "rewards/rejected": -1.4531781673431396, "step": 885 }, { "epoch": 0.47633383508947985, "grad_norm": 7.6920383686095795, "learning_rate": 9.895351543941628e-07, "logits/chosen": -0.14391431212425232, "logits/rejected": -0.03333331272006035, "logps/chosen": -1.318817377090454, "logps/rejected": -1.4459335803985596, "loss": 1.6889, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.318817377090454, "rewards/margins": 0.1271161586046219, "rewards/rejected": -1.4459335803985596, "step": 890 }, { "epoch": 0.4790098678708814, "grad_norm": 7.794206463550445, "learning_rate": 9.892158139836724e-07, "logits/chosen": 0.05477297306060791, "logits/rejected": 0.16170786321163177, "logps/chosen": -1.2167054414749146, "logps/rejected": -1.3180320262908936, "loss": 1.6156, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2167054414749146, "rewards/margins": 0.10132656991481781, "rewards/rejected": -1.3180320262908936, "step": 895 }, { "epoch": 0.481685900652283, "grad_norm": 7.239663409256137, "learning_rate": 9.88891727199209e-07, "logits/chosen": -0.06828771531581879, "logits/rejected": -0.012270675972104073, "logps/chosen": -1.2079464197158813, "logps/rejected": -1.4431251287460327, "loss": 1.5744, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2079464197158813, "rewards/margins": 0.23517878353595734, "rewards/rejected": -1.4431251287460327, "step": 900 }, { "epoch": 0.48436193343368455, "grad_norm": 7.101761732783572, "learning_rate": 9.885628971850641e-07, "logits/chosen": 0.034347210079431534, "logits/rejected": 0.20339736342430115, "logps/chosen": -1.2839748859405518, "logps/rejected": -1.4656829833984375, "loss": 1.6507, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2839748859405518, "rewards/margins": 0.18170809745788574, "rewards/rejected": -1.4656829833984375, "step": 905 }, { "epoch": 0.48703796621508616, "grad_norm": 5.129875399961622, "learning_rate": 9.882293271315481e-07, "logits/chosen": -0.022476624697446823, "logits/rejected": 0.08038485795259476, "logps/chosen": -1.3369861841201782, "logps/rejected": -1.4090746641159058, "loss": 1.7334, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3369861841201782, "rewards/margins": 0.07208836078643799, "rewards/rejected": -1.4090746641159058, "step": 910 }, { "epoch": 0.4897139989964877, "grad_norm": 6.423648732496239, "learning_rate": 9.878910202749589e-07, "logits/chosen": 0.00816698931157589, "logits/rejected": 0.1703716367483139, "logps/chosen": -1.2553752660751343, "logps/rejected": -1.3933888673782349, "loss": 1.6275, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2553752660751343, "rewards/margins": 0.1380135416984558, "rewards/rejected": -1.3933888673782349, "step": 915 }, { "epoch": 0.49239003177788926, "grad_norm": 9.911753176821456, "learning_rate": 9.875479798975512e-07, "logits/chosen": 0.1201072558760643, "logits/rejected": 0.2435188591480255, "logps/chosen": -1.221426248550415, "logps/rejected": -1.3659164905548096, "loss": 1.6165, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.221426248550415, "rewards/margins": 0.1444903314113617, "rewards/rejected": -1.3659164905548096, "step": 920 }, { "epoch": 0.49506606455929086, "grad_norm": 6.945541841571725, "learning_rate": 9.87200209327504e-07, "logits/chosen": -0.014080618508160114, "logits/rejected": 0.1393960863351822, "logps/chosen": -1.305490255355835, "logps/rejected": -1.3544299602508545, "loss": 1.6717, "rewards/accuracies": 0.53125, "rewards/chosen": -1.305490255355835, "rewards/margins": 0.048939622938632965, "rewards/rejected": -1.3544299602508545, "step": 925 }, { "epoch": 0.4977420973406924, "grad_norm": 11.57724749033797, "learning_rate": 9.868477119388894e-07, "logits/chosen": -0.054570745676755905, "logits/rejected": 0.04997735470533371, "logps/chosen": -1.2871347665786743, "logps/rejected": -1.4756519794464111, "loss": 1.6784, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2871347665786743, "rewards/margins": 0.18851712346076965, "rewards/rejected": -1.4756519794464111, "step": 930 }, { "epoch": 0.500418130122094, "grad_norm": 8.932820080999921, "learning_rate": 9.864904911516383e-07, "logits/chosen": 0.04412764683365822, "logits/rejected": 0.07467509061098099, "logps/chosen": -1.2148029804229736, "logps/rejected": -1.4018632173538208, "loss": 1.6217, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2148029804229736, "rewards/margins": 0.18706026673316956, "rewards/rejected": -1.4018632173538208, "step": 935 }, { "epoch": 0.5030941629034956, "grad_norm": 8.357914495043543, "learning_rate": 9.861285504315084e-07, "logits/chosen": -0.028966009616851807, "logits/rejected": 0.07452841848134995, "logps/chosen": -1.2796359062194824, "logps/rejected": -1.3434219360351562, "loss": 1.6655, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2796359062194824, "rewards/margins": 0.06378618627786636, "rewards/rejected": -1.3434219360351562, "step": 940 }, { "epoch": 0.5057701956848971, "grad_norm": 7.554214115567587, "learning_rate": 9.857618932900502e-07, "logits/chosen": -0.0378776416182518, "logits/rejected": 0.07617872953414917, "logps/chosen": -1.2562286853790283, "logps/rejected": -1.412217617034912, "loss": 1.6175, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2562286853790283, "rewards/margins": 0.15598897635936737, "rewards/rejected": -1.412217617034912, "step": 945 }, { "epoch": 0.5084462284662987, "grad_norm": 6.11351422417312, "learning_rate": 9.853905232845727e-07, "logits/chosen": -0.016596803441643715, "logits/rejected": 0.13409195840358734, "logps/chosen": -1.3554424047470093, "logps/rejected": -1.384244680404663, "loss": 1.7487, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.3554424047470093, "rewards/margins": 0.028802240267395973, "rewards/rejected": -1.384244680404663, "step": 950 }, { "epoch": 0.5111222612477003, "grad_norm": 5.893037811337725, "learning_rate": 9.850144440181095e-07, "logits/chosen": -0.007894618436694145, "logits/rejected": 0.18766812980175018, "logps/chosen": -1.3416087627410889, "logps/rejected": -1.3942476511001587, "loss": 1.738, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.3416087627410889, "rewards/margins": 0.052638787776231766, "rewards/rejected": -1.3942476511001587, "step": 955 }, { "epoch": 0.5137982940291018, "grad_norm": 7.4002169151715815, "learning_rate": 9.846336591393832e-07, "logits/chosen": -0.03831852972507477, "logits/rejected": 0.09143869578838348, "logps/chosen": -1.2990386486053467, "logps/rejected": -1.3768006563186646, "loss": 1.6703, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2990386486053467, "rewards/margins": 0.07776188850402832, "rewards/rejected": -1.3768006563186646, "step": 960 }, { "epoch": 0.5164743268105034, "grad_norm": 7.213188120271872, "learning_rate": 9.842481723427704e-07, "logits/chosen": 0.026158872991800308, "logits/rejected": 0.0104296263307333, "logps/chosen": -1.3406366109848022, "logps/rejected": -1.4973324537277222, "loss": 1.6826, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3406366109848022, "rewards/margins": 0.1566958725452423, "rewards/rejected": -1.4973324537277222, "step": 965 }, { "epoch": 0.519150359591905, "grad_norm": 13.922579926966854, "learning_rate": 9.838579873682658e-07, "logits/chosen": 0.04408761113882065, "logits/rejected": 0.041850827634334564, "logps/chosen": -1.223388910293579, "logps/rejected": -1.309316873550415, "loss": 1.6618, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.223388910293579, "rewards/margins": 0.08592810481786728, "rewards/rejected": -1.309316873550415, "step": 970 }, { "epoch": 0.5218263923733065, "grad_norm": 7.122156841654528, "learning_rate": 9.834631080014457e-07, "logits/chosen": -0.09621085971593857, "logits/rejected": 0.05997641757130623, "logps/chosen": -1.2935442924499512, "logps/rejected": -1.385911226272583, "loss": 1.6683, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2935442924499512, "rewards/margins": 0.09236685931682587, "rewards/rejected": -1.385911226272583, "step": 975 }, { "epoch": 0.5245024251547081, "grad_norm": 12.369058104273913, "learning_rate": 9.830635380734312e-07, "logits/chosen": -0.09873241931200027, "logits/rejected": 0.07674108445644379, "logps/chosen": -1.3272755146026611, "logps/rejected": -1.425257921218872, "loss": 1.7112, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3272755146026611, "rewards/margins": 0.09798254072666168, "rewards/rejected": -1.425257921218872, "step": 980 }, { "epoch": 0.5271784579361097, "grad_norm": 9.248028376973298, "learning_rate": 9.826592814608517e-07, "logits/chosen": 0.0055808136239647865, "logits/rejected": 0.17035320401191711, "logps/chosen": -1.3002865314483643, "logps/rejected": -1.4181766510009766, "loss": 1.6661, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3002865314483643, "rewards/margins": 0.11789016425609589, "rewards/rejected": -1.4181766510009766, "step": 985 }, { "epoch": 0.5298544907175113, "grad_norm": 7.1786141635335765, "learning_rate": 9.822503420858067e-07, "logits/chosen": 0.04005901888012886, "logits/rejected": 0.07813599705696106, "logps/chosen": -1.1756422519683838, "logps/rejected": -1.385040283203125, "loss": 1.5564, "rewards/accuracies": 0.5625, "rewards/chosen": -1.1756422519683838, "rewards/margins": 0.2093980312347412, "rewards/rejected": -1.385040283203125, "step": 990 }, { "epoch": 0.5325305234989128, "grad_norm": 7.924804989519409, "learning_rate": 9.818367239158277e-07, "logits/chosen": 0.059029437601566315, "logits/rejected": 0.11863996833562851, "logps/chosen": -1.2930384874343872, "logps/rejected": -1.311851978302002, "loss": 1.7088, "rewards/accuracies": 0.45625001192092896, "rewards/chosen": -1.2930384874343872, "rewards/margins": 0.018813494592905045, "rewards/rejected": -1.311851978302002, "step": 995 }, { "epoch": 0.5352065562803144, "grad_norm": 7.403299342167094, "learning_rate": 9.8141843096384e-07, "logits/chosen": 0.05010901764035225, "logits/rejected": 0.1410784125328064, "logps/chosen": -1.3167353868484497, "logps/rejected": -1.435499668121338, "loss": 1.6718, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3167353868484497, "rewards/margins": 0.11876416206359863, "rewards/rejected": -1.435499668121338, "step": 1000 }, { "epoch": 0.537882589061716, "grad_norm": 9.467155946645837, "learning_rate": 9.809954672881237e-07, "logits/chosen": 0.011445349082350731, "logits/rejected": 0.1570497453212738, "logps/chosen": -1.3170721530914307, "logps/rejected": -1.3973153829574585, "loss": 1.7062, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3170721530914307, "rewards/margins": 0.08024324476718903, "rewards/rejected": -1.3973153829574585, "step": 1005 }, { "epoch": 0.5405586218431175, "grad_norm": 6.196994007674593, "learning_rate": 9.80567836992274e-07, "logits/chosen": -0.00795581191778183, "logits/rejected": 0.14944200217723846, "logps/chosen": -1.1793162822723389, "logps/rejected": -1.3786524534225464, "loss": 1.5712, "rewards/accuracies": 0.53125, "rewards/chosen": -1.1793162822723389, "rewards/margins": 0.19933614134788513, "rewards/rejected": -1.3786524534225464, "step": 1010 }, { "epoch": 0.5432346546245191, "grad_norm": 8.822340000877439, "learning_rate": 9.801355442251625e-07, "logits/chosen": -0.002402575220912695, "logits/rejected": 0.1537180095911026, "logps/chosen": -1.2487362623214722, "logps/rejected": -1.4060825109481812, "loss": 1.6326, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2487362623214722, "rewards/margins": 0.15734639763832092, "rewards/rejected": -1.4060825109481812, "step": 1015 }, { "epoch": 0.5459106874059207, "grad_norm": 8.590755779674463, "learning_rate": 9.796985931808949e-07, "logits/chosen": 0.014068256132304668, "logits/rejected": 0.14183005690574646, "logps/chosen": -1.3020049333572388, "logps/rejected": -1.4582527875900269, "loss": 1.6691, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3020049333572388, "rewards/margins": 0.15624786913394928, "rewards/rejected": -1.4582527875900269, "step": 1020 }, { "epoch": 0.5485867201873222, "grad_norm": 7.752168879995463, "learning_rate": 9.792569880987724e-07, "logits/chosen": -0.04073721542954445, "logits/rejected": 0.060070596635341644, "logps/chosen": -1.203337550163269, "logps/rejected": -1.4392364025115967, "loss": 1.5607, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.203337550163269, "rewards/margins": 0.23589889705181122, "rewards/rejected": -1.4392364025115967, "step": 1025 }, { "epoch": 0.5512627529687238, "grad_norm": 8.062073517399023, "learning_rate": 9.788107332632493e-07, "logits/chosen": -0.02512373961508274, "logits/rejected": 0.05384986475110054, "logps/chosen": -1.2794196605682373, "logps/rejected": -1.3432122468948364, "loss": 1.6633, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2794196605682373, "rewards/margins": 0.06379257887601852, "rewards/rejected": -1.3432122468948364, "step": 1030 }, { "epoch": 0.5539387857501255, "grad_norm": 7.008867071946321, "learning_rate": 9.783598330038924e-07, "logits/chosen": -0.035729292780160904, "logits/rejected": 0.06643488258123398, "logps/chosen": -1.3670885562896729, "logps/rejected": -1.400439977645874, "loss": 1.7529, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3670885562896729, "rewards/margins": 0.033351391553878784, "rewards/rejected": -1.400439977645874, "step": 1035 }, { "epoch": 0.5566148185315271, "grad_norm": 7.184521705306499, "learning_rate": 9.779042916953376e-07, "logits/chosen": -0.01730963960289955, "logits/rejected": 0.10935752093791962, "logps/chosen": -1.2887794971466064, "logps/rejected": -1.4097373485565186, "loss": 1.6628, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2887794971466064, "rewards/margins": 0.12095792591571808, "rewards/rejected": -1.4097373485565186, "step": 1040 }, { "epoch": 0.5592908513129285, "grad_norm": 6.354357386659293, "learning_rate": 9.774441137572487e-07, "logits/chosen": -0.07398457825183868, "logits/rejected": 0.04312765598297119, "logps/chosen": -1.2651703357696533, "logps/rejected": -1.4347505569458008, "loss": 1.6325, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2651703357696533, "rewards/margins": 0.16958020627498627, "rewards/rejected": -1.4347505569458008, "step": 1045 }, { "epoch": 0.5619668840943302, "grad_norm": 12.460384564896605, "learning_rate": 9.76979303654274e-07, "logits/chosen": -0.0898417979478836, "logits/rejected": -0.005552306771278381, "logps/chosen": -1.308148980140686, "logps/rejected": -1.4421528577804565, "loss": 1.6773, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.308148980140686, "rewards/margins": 0.1340039223432541, "rewards/rejected": -1.4421528577804565, "step": 1050 }, { "epoch": 0.5646429168757318, "grad_norm": 11.386979354888513, "learning_rate": 9.765098658960035e-07, "logits/chosen": -0.015278068371117115, "logits/rejected": 0.05204174667596817, "logps/chosen": -1.2988243103027344, "logps/rejected": -1.4415223598480225, "loss": 1.6623, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2988243103027344, "rewards/margins": 0.14269819855690002, "rewards/rejected": -1.4415223598480225, "step": 1055 }, { "epoch": 0.5673189496571333, "grad_norm": 8.994374756793784, "learning_rate": 9.76035805036924e-07, "logits/chosen": 0.011766267940402031, "logits/rejected": 0.16604655981063843, "logps/chosen": -1.367139458656311, "logps/rejected": -1.4718599319458008, "loss": 1.7199, "rewards/accuracies": 0.5625, "rewards/chosen": -1.367139458656311, "rewards/margins": 0.10472049564123154, "rewards/rejected": -1.4718599319458008, "step": 1060 }, { "epoch": 0.5699949824385349, "grad_norm": 5.90780125638111, "learning_rate": 9.755571256763764e-07, "logits/chosen": 0.04043951630592346, "logits/rejected": 0.15448689460754395, "logps/chosen": -1.2304548025131226, "logps/rejected": -1.4319727420806885, "loss": 1.5995, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2304548025131226, "rewards/margins": 0.20151777565479279, "rewards/rejected": -1.4319727420806885, "step": 1065 }, { "epoch": 0.5726710152199365, "grad_norm": 7.0689647235553394, "learning_rate": 9.750738324585097e-07, "logits/chosen": -0.09330417960882187, "logits/rejected": 0.12764659523963928, "logps/chosen": -1.286632776260376, "logps/rejected": -1.462348461151123, "loss": 1.6161, "rewards/accuracies": 0.5625, "rewards/chosen": -1.286632776260376, "rewards/margins": 0.1757156103849411, "rewards/rejected": -1.462348461151123, "step": 1070 }, { "epoch": 0.5753470480013381, "grad_norm": 6.9557637626714435, "learning_rate": 9.74585930072237e-07, "logits/chosen": -0.008208471350371838, "logits/rejected": 0.10426203161478043, "logps/chosen": -1.2359673976898193, "logps/rejected": -1.4388471841812134, "loss": 1.6272, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2359673976898193, "rewards/margins": 0.20287981629371643, "rewards/rejected": -1.4388471841812134, "step": 1075 }, { "epoch": 0.5780230807827396, "grad_norm": 7.520732608296653, "learning_rate": 9.740934232511892e-07, "logits/chosen": -0.08178670704364777, "logits/rejected": 0.015749135985970497, "logps/chosen": -1.3555552959442139, "logps/rejected": -1.3882399797439575, "loss": 1.7303, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3555552959442139, "rewards/margins": 0.03268469125032425, "rewards/rejected": -1.3882399797439575, "step": 1080 }, { "epoch": 0.5806991135641412, "grad_norm": 6.986884533733415, "learning_rate": 9.735963167736698e-07, "logits/chosen": -0.022600090131163597, "logits/rejected": 0.12094497680664062, "logps/chosen": -1.3131974935531616, "logps/rejected": -1.345354676246643, "loss": 1.7244, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.3131974935531616, "rewards/margins": 0.032157205045223236, "rewards/rejected": -1.345354676246643, "step": 1085 }, { "epoch": 0.5833751463455428, "grad_norm": 5.930443648116194, "learning_rate": 9.730946154626078e-07, "logits/chosen": -0.00549353426322341, "logits/rejected": 0.08679809421300888, "logps/chosen": -1.2895848751068115, "logps/rejected": -1.3172898292541504, "loss": 1.699, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.2895848751068115, "rewards/margins": 0.027705058455467224, "rewards/rejected": -1.3172898292541504, "step": 1090 }, { "epoch": 0.5860511791269443, "grad_norm": 9.926583525201936, "learning_rate": 9.725883241855117e-07, "logits/chosen": -0.14290958642959595, "logits/rejected": -0.02356121316552162, "logps/chosen": -1.2431046962738037, "logps/rejected": -1.3776899576187134, "loss": 1.6306, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2431046962738037, "rewards/margins": 0.1345853954553604, "rewards/rejected": -1.3776899576187134, "step": 1095 }, { "epoch": 0.5887272119083459, "grad_norm": 8.687027405990982, "learning_rate": 9.720774478544218e-07, "logits/chosen": -0.004207250662147999, "logits/rejected": 0.08541065454483032, "logps/chosen": -1.1548147201538086, "logps/rejected": -1.4498443603515625, "loss": 1.5286, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1548147201538086, "rewards/margins": 0.295029878616333, "rewards/rejected": -1.4498443603515625, "step": 1100 }, { "epoch": 0.5914032446897475, "grad_norm": 6.933596671987193, "learning_rate": 9.715619914258624e-07, "logits/chosen": -0.05270897224545479, "logits/rejected": 0.010835884138941765, "logps/chosen": -1.2801262140274048, "logps/rejected": -1.3806558847427368, "loss": 1.6644, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2801262140274048, "rewards/margins": 0.10052965581417084, "rewards/rejected": -1.3806558847427368, "step": 1105 }, { "epoch": 0.594079277471149, "grad_norm": 8.646665805004428, "learning_rate": 9.710419599007937e-07, "logits/chosen": -0.04489428550004959, "logits/rejected": 0.06177765130996704, "logps/chosen": -1.26051926612854, "logps/rejected": -1.2954927682876587, "loss": 1.6631, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.26051926612854, "rewards/margins": 0.03497341275215149, "rewards/rejected": -1.2954927682876587, "step": 1110 }, { "epoch": 0.5967553102525506, "grad_norm": 12.184135617650634, "learning_rate": 9.705173583245643e-07, "logits/chosen": 0.06277371942996979, "logits/rejected": 0.1672242432832718, "logps/chosen": -1.184098482131958, "logps/rejected": -1.371530532836914, "loss": 1.5668, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.184098482131958, "rewards/margins": 0.18743225932121277, "rewards/rejected": -1.371530532836914, "step": 1115 }, { "epoch": 0.5994313430339522, "grad_norm": 8.431296788619152, "learning_rate": 9.699881917868609e-07, "logits/chosen": -0.13257569074630737, "logits/rejected": -0.0393402986228466, "logps/chosen": -1.2535715103149414, "logps/rejected": -1.3950719833374023, "loss": 1.6096, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2535715103149414, "rewards/margins": 0.1415005922317505, "rewards/rejected": -1.3950719833374023, "step": 1120 }, { "epoch": 0.6021073758153538, "grad_norm": 9.393854795053596, "learning_rate": 9.694544654216594e-07, "logits/chosen": -0.13203957676887512, "logits/rejected": 0.030075013637542725, "logps/chosen": -1.2522079944610596, "logps/rejected": -1.3912581205368042, "loss": 1.63, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2522079944610596, "rewards/margins": 0.1390499621629715, "rewards/rejected": -1.3912581205368042, "step": 1125 }, { "epoch": 0.6047834085967553, "grad_norm": 7.378137478880466, "learning_rate": 9.689161844071755e-07, "logits/chosen": 0.03407396003603935, "logits/rejected": 0.08551694452762604, "logps/chosen": -1.293287992477417, "logps/rejected": -1.435396432876587, "loss": 1.6505, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.293287992477417, "rewards/margins": 0.14210841059684753, "rewards/rejected": -1.435396432876587, "step": 1130 }, { "epoch": 0.6074594413781569, "grad_norm": 10.528202132639327, "learning_rate": 9.683733539658138e-07, "logits/chosen": -0.03661159425973892, "logits/rejected": 0.10714194923639297, "logps/chosen": -1.3035916090011597, "logps/rejected": -1.490254282951355, "loss": 1.6403, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3035916090011597, "rewards/margins": 0.18666283786296844, "rewards/rejected": -1.490254282951355, "step": 1135 }, { "epoch": 0.6101354741595585, "grad_norm": 7.904780641376041, "learning_rate": 9.678259793641178e-07, "logits/chosen": -0.03270817548036575, "logits/rejected": -0.005903895013034344, "logps/chosen": -1.281888723373413, "logps/rejected": -1.306801199913025, "loss": 1.6899, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.281888723373413, "rewards/margins": 0.024912385269999504, "rewards/rejected": -1.306801199913025, "step": 1140 }, { "epoch": 0.61281150694096, "grad_norm": 7.430297222491973, "learning_rate": 9.672740659127183e-07, "logits/chosen": -0.15106946229934692, "logits/rejected": -0.05194350332021713, "logps/chosen": -1.3074098825454712, "logps/rejected": -1.4172364473342896, "loss": 1.6766, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3074098825454712, "rewards/margins": 0.10982663929462433, "rewards/rejected": -1.4172364473342896, "step": 1145 }, { "epoch": 0.6154875397223616, "grad_norm": 8.04457179946613, "learning_rate": 9.667176189662818e-07, "logits/chosen": -0.14218200743198395, "logits/rejected": -0.017552126199007034, "logps/chosen": -1.1962858438491821, "logps/rejected": -1.341317892074585, "loss": 1.5791, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.1962858438491821, "rewards/margins": 0.14503201842308044, "rewards/rejected": -1.341317892074585, "step": 1150 }, { "epoch": 0.6181635725037632, "grad_norm": 6.891796025143772, "learning_rate": 9.661566439234592e-07, "logits/chosen": -0.0063630566000938416, "logits/rejected": 0.06813793629407883, "logps/chosen": -1.2960070371627808, "logps/rejected": -1.3874499797821045, "loss": 1.6781, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2960070371627808, "rewards/margins": 0.09144283831119537, "rewards/rejected": -1.3874499797821045, "step": 1155 }, { "epoch": 0.6208396052851648, "grad_norm": 7.694183281262976, "learning_rate": 9.655911462268327e-07, "logits/chosen": 0.039759453386068344, "logits/rejected": 0.11061426252126694, "logps/chosen": -1.2482908964157104, "logps/rejected": -1.3631278276443481, "loss": 1.619, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2482908964157104, "rewards/margins": 0.11483697593212128, "rewards/rejected": -1.3631278276443481, "step": 1160 }, { "epoch": 0.6235156380665663, "grad_norm": 5.952230367438864, "learning_rate": 9.650211313628636e-07, "logits/chosen": -0.033538371324539185, "logits/rejected": 0.03427625447511673, "logps/chosen": -1.1877871751785278, "logps/rejected": -1.3554905652999878, "loss": 1.5718, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.1877871751785278, "rewards/margins": 0.16770341992378235, "rewards/rejected": -1.3554905652999878, "step": 1165 }, { "epoch": 0.6261916708479679, "grad_norm": 5.896613098675456, "learning_rate": 9.644466048618386e-07, "logits/chosen": -0.07759127020835876, "logits/rejected": 0.0643915981054306, "logps/chosen": -1.4012242555618286, "logps/rejected": -1.4206397533416748, "loss": 1.7616, "rewards/accuracies": 0.5, "rewards/chosen": -1.4012242555618286, "rewards/margins": 0.0194155965000391, "rewards/rejected": -1.4206397533416748, "step": 1170 }, { "epoch": 0.6288677036293695, "grad_norm": 6.190941374554223, "learning_rate": 9.63867572297816e-07, "logits/chosen": -0.03675876185297966, "logits/rejected": 0.1220620647072792, "logps/chosen": -1.2355022430419922, "logps/rejected": -1.3272840976715088, "loss": 1.6463, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2355022430419922, "rewards/margins": 0.09178170561790466, "rewards/rejected": -1.3272840976715088, "step": 1175 }, { "epoch": 0.631543736410771, "grad_norm": 5.84549471884131, "learning_rate": 9.632840392885727e-07, "logits/chosen": -0.04186251014471054, "logits/rejected": 0.071475088596344, "logps/chosen": -1.336317539215088, "logps/rejected": -1.5061604976654053, "loss": 1.6709, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.336317539215088, "rewards/margins": 0.16984297335147858, "rewards/rejected": -1.5061604976654053, "step": 1180 }, { "epoch": 0.6342197691921726, "grad_norm": 9.110628218510094, "learning_rate": 9.626960114955483e-07, "logits/chosen": 0.004259307868778706, "logits/rejected": 0.11975675821304321, "logps/chosen": -1.3721723556518555, "logps/rejected": -1.4441169500350952, "loss": 1.7389, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.3721723556518555, "rewards/margins": 0.07194463908672333, "rewards/rejected": -1.4441169500350952, "step": 1185 }, { "epoch": 0.6368958019735742, "grad_norm": 8.604588520350909, "learning_rate": 9.621034946237909e-07, "logits/chosen": -0.06812702864408493, "logits/rejected": 0.053222037851810455, "logps/chosen": -1.32889723777771, "logps/rejected": -1.4761818647384644, "loss": 1.6724, "rewards/accuracies": 0.53125, "rewards/chosen": -1.32889723777771, "rewards/margins": 0.147284597158432, "rewards/rejected": -1.4761818647384644, "step": 1190 }, { "epoch": 0.6395718347549757, "grad_norm": 9.709864140064102, "learning_rate": 9.615064944219021e-07, "logits/chosen": -0.0022974193561822176, "logits/rejected": 0.10469532012939453, "logps/chosen": -1.2060132026672363, "logps/rejected": -1.4014453887939453, "loss": 1.5608, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2060132026672363, "rewards/margins": 0.19543221592903137, "rewards/rejected": -1.4014453887939453, "step": 1195 }, { "epoch": 0.6422478675363773, "grad_norm": 7.482539490205642, "learning_rate": 9.609050166819803e-07, "logits/chosen": -0.06645162403583527, "logits/rejected": -0.013482004404067993, "logps/chosen": -1.2427408695220947, "logps/rejected": -1.3890235424041748, "loss": 1.6204, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2427408695220947, "rewards/margins": 0.1462826430797577, "rewards/rejected": -1.3890235424041748, "step": 1200 }, { "epoch": 0.6422478675363773, "eval_logits/chosen": 0.27460145950317383, "eval_logits/rejected": 0.3559272885322571, "eval_logps/chosen": -1.308514952659607, "eval_logps/rejected": -1.4525140523910522, "eval_loss": 1.6640385389328003, "eval_rewards/accuracies": 0.5556379556655884, "eval_rewards/chosen": -1.308514952659607, "eval_rewards/margins": 0.14399926364421844, "eval_rewards/rejected": -1.4525140523910522, "eval_runtime": 40.6531, "eval_samples_per_second": 33.085, "eval_steps_per_second": 8.29, "step": 1200 }, { "epoch": 0.6449239003177789, "grad_norm": 10.047042764127452, "learning_rate": 9.602990672395653e-07, "logits/chosen": -0.13160620629787445, "logits/rejected": 0.02522926963865757, "logps/chosen": -1.2614212036132812, "logps/rejected": -1.3818312883377075, "loss": 1.6262, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2614212036132812, "rewards/margins": 0.12041006237268448, "rewards/rejected": -1.3818312883377075, "step": 1205 }, { "epoch": 0.6475999330991805, "grad_norm": 9.75320853357286, "learning_rate": 9.59688651973581e-07, "logits/chosen": -0.06755046546459198, "logits/rejected": 0.10317499935626984, "logps/chosen": -1.298020362854004, "logps/rejected": -1.405084490776062, "loss": 1.6968, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.298020362854004, "rewards/margins": 0.1070641279220581, "rewards/rejected": -1.405084490776062, "step": 1210 }, { "epoch": 0.650275965880582, "grad_norm": 6.399735782427771, "learning_rate": 9.590737768062792e-07, "logits/chosen": -0.12315372377634048, "logits/rejected": -0.013821613974869251, "logps/chosen": -1.2987453937530518, "logps/rejected": -1.325515866279602, "loss": 1.6791, "rewards/accuracies": 0.5, "rewards/chosen": -1.2987453937530518, "rewards/margins": 0.026770705357193947, "rewards/rejected": -1.325515866279602, "step": 1215 }, { "epoch": 0.6529519986619836, "grad_norm": 8.322102796728917, "learning_rate": 9.584544477031816e-07, "logits/chosen": 0.055944450199604034, "logits/rejected": 0.15044358372688293, "logps/chosen": -1.2174725532531738, "logps/rejected": -1.3485443592071533, "loss": 1.6243, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2174725532531738, "rewards/margins": 0.13107183575630188, "rewards/rejected": -1.3485443592071533, "step": 1220 }, { "epoch": 0.6556280314433852, "grad_norm": 5.972083585488595, "learning_rate": 9.578306706730215e-07, "logits/chosen": -0.15188264846801758, "logits/rejected": 0.03867144137620926, "logps/chosen": -1.3097045421600342, "logps/rejected": -1.4055533409118652, "loss": 1.6674, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3097045421600342, "rewards/margins": 0.09584885090589523, "rewards/rejected": -1.4055533409118652, "step": 1225 }, { "epoch": 0.6583040642247867, "grad_norm": 8.093006246489878, "learning_rate": 9.572024517676865e-07, "logits/chosen": -0.0809139758348465, "logits/rejected": 0.012979112565517426, "logps/chosen": -1.2431614398956299, "logps/rejected": -1.4049303531646729, "loss": 1.6231, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2431614398956299, "rewards/margins": 0.1617688238620758, "rewards/rejected": -1.4049303531646729, "step": 1230 }, { "epoch": 0.6609800970061883, "grad_norm": 6.848159379031791, "learning_rate": 9.565697970821593e-07, "logits/chosen": -0.050653569400310516, "logits/rejected": 0.06407668441534042, "logps/chosen": -1.2996481657028198, "logps/rejected": -1.378269910812378, "loss": 1.6866, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2996481657028198, "rewards/margins": 0.07862164080142975, "rewards/rejected": -1.378269910812378, "step": 1235 }, { "epoch": 0.6636561297875899, "grad_norm": 8.510686496424805, "learning_rate": 9.559327127544585e-07, "logits/chosen": -0.16195550560951233, "logits/rejected": -0.04807741194963455, "logps/chosen": -1.269803762435913, "logps/rejected": -1.3921595811843872, "loss": 1.6484, "rewards/accuracies": 0.53125, "rewards/chosen": -1.269803762435913, "rewards/margins": 0.12235573679208755, "rewards/rejected": -1.3921595811843872, "step": 1240 }, { "epoch": 0.6663321625689914, "grad_norm": 7.838294023870456, "learning_rate": 9.552912049655789e-07, "logits/chosen": -0.07872875034809113, "logits/rejected": 0.07285197079181671, "logps/chosen": -1.3556663990020752, "logps/rejected": -1.3947060108184814, "loss": 1.713, "rewards/accuracies": 0.46875, "rewards/chosen": -1.3556663990020752, "rewards/margins": 0.039039671421051025, "rewards/rejected": -1.3947060108184814, "step": 1245 }, { "epoch": 0.669008195350393, "grad_norm": 11.117903128068692, "learning_rate": 9.546452799394315e-07, "logits/chosen": -0.06848958134651184, "logits/rejected": 0.09921123087406158, "logps/chosen": -1.3285969495773315, "logps/rejected": -1.3818318843841553, "loss": 1.7038, "rewards/accuracies": 0.45625001192092896, "rewards/chosen": -1.3285969495773315, "rewards/margins": 0.05323495343327522, "rewards/rejected": -1.3818318843841553, "step": 1250 }, { "epoch": 0.6716842281317946, "grad_norm": 9.426299726196365, "learning_rate": 9.539949439427846e-07, "logits/chosen": -0.10870027542114258, "logits/rejected": 0.004453069064766169, "logps/chosen": -1.2898695468902588, "logps/rejected": -1.387575387954712, "loss": 1.6565, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2898695468902588, "rewards/margins": 0.0977058857679367, "rewards/rejected": -1.387575387954712, "step": 1255 }, { "epoch": 0.6743602609131962, "grad_norm": 7.7439778835708255, "learning_rate": 9.533402032852002e-07, "logits/chosen": -0.12049994617700577, "logits/rejected": -0.005878242664039135, "logps/chosen": -1.2310149669647217, "logps/rejected": -1.411268711090088, "loss": 1.6042, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2310149669647217, "rewards/margins": 0.18025359511375427, "rewards/rejected": -1.411268711090088, "step": 1260 }, { "epoch": 0.6770362936945977, "grad_norm": 6.7439236704393535, "learning_rate": 9.526810643189754e-07, "logits/chosen": -0.02287997305393219, "logits/rejected": 0.09953973442316055, "logps/chosen": -1.2751572132110596, "logps/rejected": -1.3866431713104248, "loss": 1.6341, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2751572132110596, "rewards/margins": 0.11148588359355927, "rewards/rejected": -1.3866431713104248, "step": 1265 }, { "epoch": 0.6797123264759993, "grad_norm": 7.899839971390841, "learning_rate": 9.52017533439079e-07, "logits/chosen": -0.08607939630746841, "logits/rejected": 0.0121527761220932, "logps/chosen": -1.2371551990509033, "logps/rejected": -1.392277717590332, "loss": 1.6179, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2371551990509033, "rewards/margins": 0.15512245893478394, "rewards/rejected": -1.392277717590332, "step": 1270 }, { "epoch": 0.6823883592574009, "grad_norm": 5.628595303817242, "learning_rate": 9.513496170830909e-07, "logits/chosen": -0.0971599891781807, "logits/rejected": -0.005085567478090525, "logps/chosen": -1.2943452596664429, "logps/rejected": -1.3745836019515991, "loss": 1.6751, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.2943452596664429, "rewards/margins": 0.08023835718631744, "rewards/rejected": -1.3745836019515991, "step": 1275 }, { "epoch": 0.6850643920388024, "grad_norm": 8.352724391218983, "learning_rate": 9.506773217311382e-07, "logits/chosen": -0.0870608538389206, "logits/rejected": 0.03902773559093475, "logps/chosen": -1.3659374713897705, "logps/rejected": -1.4287694692611694, "loss": 1.7331, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3659374713897705, "rewards/margins": 0.06283222138881683, "rewards/rejected": -1.4287694692611694, "step": 1280 }, { "epoch": 0.687740424820204, "grad_norm": 7.466254419479338, "learning_rate": 9.500006539058334e-07, "logits/chosen": -0.028805231675505638, "logits/rejected": 0.08828462660312653, "logps/chosen": -1.241590976715088, "logps/rejected": -1.3411766290664673, "loss": 1.6123, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.241590976715088, "rewards/margins": 0.09958568960428238, "rewards/rejected": -1.3411766290664673, "step": 1285 }, { "epoch": 0.6904164576016056, "grad_norm": 7.996329692726276, "learning_rate": 9.493196201722109e-07, "logits/chosen": -0.179660826921463, "logits/rejected": -0.055936299264431, "logps/chosen": -1.310693621635437, "logps/rejected": -1.3532706499099731, "loss": 1.7178, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.310693621635437, "rewards/margins": 0.04257718473672867, "rewards/rejected": -1.3532706499099731, "step": 1290 }, { "epoch": 0.6930924903830072, "grad_norm": 6.1239123973147676, "learning_rate": 9.486342271376628e-07, "logits/chosen": -0.08025787770748138, "logits/rejected": -0.07135315984487534, "logps/chosen": -1.2761461734771729, "logps/rejected": -1.4287402629852295, "loss": 1.6428, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2761461734771729, "rewards/margins": 0.15259432792663574, "rewards/rejected": -1.4287402629852295, "step": 1295 }, { "epoch": 0.6957685231644087, "grad_norm": 7.432784798775175, "learning_rate": 9.479444814518755e-07, "logits/chosen": -0.0681220144033432, "logits/rejected": 0.14305290579795837, "logps/chosen": -1.2733452320098877, "logps/rejected": -1.4136388301849365, "loss": 1.6306, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2733452320098877, "rewards/margins": 0.14029362797737122, "rewards/rejected": -1.4136388301849365, "step": 1300 }, { "epoch": 0.6984445559458103, "grad_norm": 7.02831430011763, "learning_rate": 9.472503898067645e-07, "logits/chosen": 0.015904754400253296, "logits/rejected": 0.06446841359138489, "logps/chosen": -1.2825438976287842, "logps/rejected": -1.4184726476669312, "loss": 1.6303, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2825438976287842, "rewards/margins": 0.13592875003814697, "rewards/rejected": -1.4184726476669312, "step": 1305 }, { "epoch": 0.701120588727212, "grad_norm": 7.109526609365743, "learning_rate": 9.465519589364099e-07, "logits/chosen": 0.019358232617378235, "logits/rejected": 0.08662734180688858, "logps/chosen": -1.2953920364379883, "logps/rejected": -1.4193115234375, "loss": 1.6836, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2953920364379883, "rewards/margins": 0.12391936779022217, "rewards/rejected": -1.4193115234375, "step": 1310 }, { "epoch": 0.7037966215086134, "grad_norm": 6.365121027589216, "learning_rate": 9.458491956169914e-07, "logits/chosen": -0.07012242078781128, "logits/rejected": 0.09191621094942093, "logps/chosen": -1.2381283044815063, "logps/rejected": -1.4403287172317505, "loss": 1.6023, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2381283044815063, "rewards/margins": 0.2022004872560501, "rewards/rejected": -1.4403287172317505, "step": 1315 }, { "epoch": 0.706472654290015, "grad_norm": 5.815899579642899, "learning_rate": 9.451421066667215e-07, "logits/chosen": -0.16489481925964355, "logits/rejected": 0.016862286254763603, "logps/chosen": -1.2204926013946533, "logps/rejected": -1.3832234144210815, "loss": 1.5897, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2204926013946533, "rewards/margins": 0.16273091733455658, "rewards/rejected": -1.3832234144210815, "step": 1320 }, { "epoch": 0.7091486870714167, "grad_norm": 11.57452564485908, "learning_rate": 9.444306989457805e-07, "logits/chosen": -0.0183856338262558, "logits/rejected": 0.07245060801506042, "logps/chosen": -1.2934316396713257, "logps/rejected": -1.3811691999435425, "loss": 1.6827, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2934316396713257, "rewards/margins": 0.08773749321699142, "rewards/rejected": -1.3811691999435425, "step": 1325 }, { "epoch": 0.7118247198528181, "grad_norm": 5.568127237800386, "learning_rate": 9.437149793562489e-07, "logits/chosen": -0.05760542303323746, "logits/rejected": 0.04698445647954941, "logps/chosen": -1.2826694250106812, "logps/rejected": -1.3580313920974731, "loss": 1.6626, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2826694250106812, "rewards/margins": 0.07536178827285767, "rewards/rejected": -1.3580313920974731, "step": 1330 }, { "epoch": 0.7145007526342197, "grad_norm": 6.356676163069989, "learning_rate": 9.429949548420417e-07, "logits/chosen": -0.03230949491262436, "logits/rejected": 0.036061473190784454, "logps/chosen": -1.353149652481079, "logps/rejected": -1.481315016746521, "loss": 1.6798, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.353149652481079, "rewards/margins": 0.12816523015499115, "rewards/rejected": -1.481315016746521, "step": 1335 }, { "epoch": 0.7171767854156214, "grad_norm": 7.862975720515515, "learning_rate": 9.422706323888396e-07, "logits/chosen": -0.02949601784348488, "logits/rejected": -0.0037048873491585255, "logps/chosen": -1.3224132061004639, "logps/rejected": -1.4355237483978271, "loss": 1.6931, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3224132061004639, "rewards/margins": 0.11311064660549164, "rewards/rejected": -1.4355237483978271, "step": 1340 }, { "epoch": 0.719852818197023, "grad_norm": 5.176058235706188, "learning_rate": 9.415420190240225e-07, "logits/chosen": 0.0031900019384920597, "logits/rejected": 0.15508271753787994, "logps/chosen": -1.3075135946273804, "logps/rejected": -1.3928053379058838, "loss": 1.6681, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3075135946273804, "rewards/margins": 0.0852917954325676, "rewards/rejected": -1.3928053379058838, "step": 1345 }, { "epoch": 0.7225288509784245, "grad_norm": 7.611410454276417, "learning_rate": 9.408091218166002e-07, "logits/chosen": 0.005640983581542969, "logits/rejected": 0.05313152074813843, "logps/chosen": -1.2921403646469116, "logps/rejected": -1.3184449672698975, "loss": 1.7019, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2921403646469116, "rewards/margins": 0.02630445919930935, "rewards/rejected": -1.3184449672698975, "step": 1350 }, { "epoch": 0.7252048837598261, "grad_norm": 7.152231165146874, "learning_rate": 9.400719478771449e-07, "logits/chosen": -0.010153415612876415, "logits/rejected": 0.25364306569099426, "logps/chosen": -1.3463757038116455, "logps/rejected": -1.4059813022613525, "loss": 1.727, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3463757038116455, "rewards/margins": 0.05960559844970703, "rewards/rejected": -1.4059813022613525, "step": 1355 }, { "epoch": 0.7278809165412277, "grad_norm": 8.501974366271751, "learning_rate": 9.393305043577209e-07, "logits/chosen": -0.14782077074050903, "logits/rejected": -0.020702391862869263, "logps/chosen": -1.359283685684204, "logps/rejected": -1.4838364124298096, "loss": 1.7063, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.359283685684204, "rewards/margins": 0.1245526447892189, "rewards/rejected": -1.4838364124298096, "step": 1360 }, { "epoch": 0.7305569493226292, "grad_norm": 5.371286371507929, "learning_rate": 9.38584798451817e-07, "logits/chosen": -0.02379719726741314, "logits/rejected": 0.09597743302583694, "logps/chosen": -1.2888532876968384, "logps/rejected": -1.4119470119476318, "loss": 1.6511, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2888532876968384, "rewards/margins": 0.12309379875659943, "rewards/rejected": -1.4119470119476318, "step": 1365 }, { "epoch": 0.7332329821040308, "grad_norm": 8.362674846669057, "learning_rate": 9.37834837394275e-07, "logits/chosen": -0.04941604286432266, "logits/rejected": 0.049748942255973816, "logps/chosen": -1.3442224264144897, "logps/rejected": -1.5253331661224365, "loss": 1.6698, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3442224264144897, "rewards/margins": 0.1811104714870453, "rewards/rejected": -1.5253331661224365, "step": 1370 }, { "epoch": 0.7359090148854324, "grad_norm": 6.495126111078663, "learning_rate": 9.370806284612203e-07, "logits/chosen": -0.06998618692159653, "logits/rejected": 0.05238911509513855, "logps/chosen": -1.2650034427642822, "logps/rejected": -1.4815846681594849, "loss": 1.6215, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2650034427642822, "rewards/margins": 0.21658115088939667, "rewards/rejected": -1.4815846681594849, "step": 1375 }, { "epoch": 0.738585047666834, "grad_norm": 9.380202598349682, "learning_rate": 9.363221789699912e-07, "logits/chosen": -0.12380628287792206, "logits/rejected": -0.02891998365521431, "logps/chosen": -1.2920019626617432, "logps/rejected": -1.347071647644043, "loss": 1.695, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2920019626617432, "rewards/margins": 0.055069684982299805, "rewards/rejected": -1.347071647644043, "step": 1380 }, { "epoch": 0.7412610804482355, "grad_norm": 7.368251054782439, "learning_rate": 9.355594962790682e-07, "logits/chosen": -0.10633337497711182, "logits/rejected": -0.004128370434045792, "logps/chosen": -1.2442162036895752, "logps/rejected": -1.3430384397506714, "loss": 1.647, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2442162036895752, "rewards/margins": 0.09882227331399918, "rewards/rejected": -1.3430384397506714, "step": 1385 }, { "epoch": 0.7439371132296371, "grad_norm": 9.269784762432563, "learning_rate": 9.34792587788002e-07, "logits/chosen": 0.009227467700839043, "logits/rejected": 0.10784689337015152, "logps/chosen": -1.2933768033981323, "logps/rejected": -1.4177485704421997, "loss": 1.6632, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2933768033981323, "rewards/margins": 0.12437163293361664, "rewards/rejected": -1.4177485704421997, "step": 1390 }, { "epoch": 0.7466131460110387, "grad_norm": 6.200345019295354, "learning_rate": 9.34021460937342e-07, "logits/chosen": -0.014314393512904644, "logits/rejected": 0.07106490433216095, "logps/chosen": -1.2545225620269775, "logps/rejected": -1.330452561378479, "loss": 1.6528, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2545225620269775, "rewards/margins": 0.07592994719743729, "rewards/rejected": -1.330452561378479, "step": 1395 }, { "epoch": 0.7492891787924402, "grad_norm": 5.113076051211028, "learning_rate": 9.332461232085646e-07, "logits/chosen": -0.1846596896648407, "logits/rejected": -0.076918825507164, "logps/chosen": -1.3319851160049438, "logps/rejected": -1.3965907096862793, "loss": 1.6926, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3319851160049438, "rewards/margins": 0.06460557878017426, "rewards/rejected": -1.3965907096862793, "step": 1400 }, { "epoch": 0.7519652115738418, "grad_norm": 5.78917820670399, "learning_rate": 9.324665821239998e-07, "logits/chosen": -0.07831187546253204, "logits/rejected": 0.07580214738845825, "logps/chosen": -1.189420223236084, "logps/rejected": -1.4074944257736206, "loss": 1.5701, "rewards/accuracies": 0.5625, "rewards/chosen": -1.189420223236084, "rewards/margins": 0.21807405352592468, "rewards/rejected": -1.4074944257736206, "step": 1405 }, { "epoch": 0.7546412443552434, "grad_norm": 5.6554854963582235, "learning_rate": 9.316828452467583e-07, "logits/chosen": -0.13492147624492645, "logits/rejected": 0.018135586753487587, "logps/chosen": -1.2970507144927979, "logps/rejected": -1.44521164894104, "loss": 1.6474, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2970507144927979, "rewards/margins": 0.14816097915172577, "rewards/rejected": -1.44521164894104, "step": 1410 }, { "epoch": 0.7573172771366449, "grad_norm": 9.906896734902677, "learning_rate": 9.30894920180659e-07, "logits/chosen": -0.03486012667417526, "logits/rejected": 0.09014071524143219, "logps/chosen": -1.3414547443389893, "logps/rejected": -1.3082062005996704, "loss": 1.7412, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3414547443389893, "rewards/margins": -0.033248428255319595, "rewards/rejected": -1.3082062005996704, "step": 1415 }, { "epoch": 0.7599933099180465, "grad_norm": 5.945288403773963, "learning_rate": 9.301028145701543e-07, "logits/chosen": -0.030293172225356102, "logits/rejected": 0.07679029554128647, "logps/chosen": -1.2248847484588623, "logps/rejected": -1.4750934839248657, "loss": 1.5933, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2248847484588623, "rewards/margins": 0.25020867586135864, "rewards/rejected": -1.4750934839248657, "step": 1420 }, { "epoch": 0.7626693426994481, "grad_norm": 6.2924481165233015, "learning_rate": 9.293065361002563e-07, "logits/chosen": -0.0005339503404684365, "logits/rejected": 0.05334051325917244, "logps/chosen": -1.2537484169006348, "logps/rejected": -1.514533281326294, "loss": 1.6072, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2537484169006348, "rewards/margins": 0.26078492403030396, "rewards/rejected": -1.514533281326294, "step": 1425 }, { "epoch": 0.7653453754808497, "grad_norm": 6.747990687916865, "learning_rate": 9.285060924964622e-07, "logits/chosen": -0.1154429093003273, "logits/rejected": -0.004071956966072321, "logps/chosen": -1.3178616762161255, "logps/rejected": -1.3854830265045166, "loss": 1.6837, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.3178616762161255, "rewards/margins": 0.06762126833200455, "rewards/rejected": -1.3854830265045166, "step": 1430 }, { "epoch": 0.7680214082622512, "grad_norm": 7.379312272553948, "learning_rate": 9.277014915246792e-07, "logits/chosen": -0.001880859606899321, "logits/rejected": 0.03771301731467247, "logps/chosen": -1.2543996572494507, "logps/rejected": -1.4419598579406738, "loss": 1.6479, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2543996572494507, "rewards/margins": 0.1875600814819336, "rewards/rejected": -1.4419598579406738, "step": 1435 }, { "epoch": 0.7706974410436528, "grad_norm": 5.879818791750725, "learning_rate": 9.268927409911498e-07, "logits/chosen": -0.06734560430049896, "logits/rejected": 0.010824119672179222, "logps/chosen": -1.3159987926483154, "logps/rejected": -1.3603847026824951, "loss": 1.6746, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3159987926483154, "rewards/margins": 0.04438580945134163, "rewards/rejected": -1.3603847026824951, "step": 1440 }, { "epoch": 0.7733734738250544, "grad_norm": 8.30102687568028, "learning_rate": 9.260798487423749e-07, "logits/chosen": -0.11533383280038834, "logits/rejected": 0.06859040260314941, "logps/chosen": -1.3441131114959717, "logps/rejected": -1.4326902627944946, "loss": 1.7039, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3441131114959717, "rewards/margins": 0.08857724070549011, "rewards/rejected": -1.4326902627944946, "step": 1445 }, { "epoch": 0.7760495066064559, "grad_norm": 16.99885339856131, "learning_rate": 9.252628226650389e-07, "logits/chosen": -0.005267086438834667, "logits/rejected": 0.0751887634396553, "logps/chosen": -1.2348849773406982, "logps/rejected": -1.3485267162322998, "loss": 1.6351, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2348849773406982, "rewards/margins": 0.1136416420340538, "rewards/rejected": -1.3485267162322998, "step": 1450 }, { "epoch": 0.7787255393878575, "grad_norm": 9.381136737150669, "learning_rate": 9.244416706859321e-07, "logits/chosen": -0.05502817779779434, "logits/rejected": 0.07302253693342209, "logps/chosen": -1.2619531154632568, "logps/rejected": -1.4499928951263428, "loss": 1.6239, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2619531154632568, "rewards/margins": 0.1880396157503128, "rewards/rejected": -1.4499928951263428, "step": 1455 }, { "epoch": 0.7814015721692591, "grad_norm": 5.68119111046384, "learning_rate": 9.23616400771875e-07, "logits/chosen": -0.051288902759552, "logits/rejected": 0.06629034131765366, "logps/chosen": -1.2232128381729126, "logps/rejected": -1.3965691328048706, "loss": 1.5924, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2232128381729126, "rewards/margins": 0.17335614562034607, "rewards/rejected": -1.3965691328048706, "step": 1460 }, { "epoch": 0.7840776049506607, "grad_norm": 6.477388224027011, "learning_rate": 9.227870209296395e-07, "logits/chosen": -0.020612351596355438, "logits/rejected": 0.06191522628068924, "logps/chosen": -1.3425668478012085, "logps/rejected": -1.4560648202896118, "loss": 1.716, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3425668478012085, "rewards/margins": 0.11349793523550034, "rewards/rejected": -1.4560648202896118, "step": 1465 }, { "epoch": 0.7867536377320622, "grad_norm": 8.658190452030363, "learning_rate": 9.219535392058728e-07, "logits/chosen": -0.133492112159729, "logits/rejected": -0.10479389131069183, "logps/chosen": -1.3073797225952148, "logps/rejected": -1.38153874874115, "loss": 1.6891, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3073797225952148, "rewards/margins": 0.0741589292883873, "rewards/rejected": -1.38153874874115, "step": 1470 }, { "epoch": 0.7894296705134638, "grad_norm": 5.762694343210087, "learning_rate": 9.211159636870181e-07, "logits/chosen": -0.11618445813655853, "logits/rejected": 0.019445959478616714, "logps/chosen": -1.2836791276931763, "logps/rejected": -1.4270191192626953, "loss": 1.6231, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2836791276931763, "rewards/margins": 0.14333993196487427, "rewards/rejected": -1.4270191192626953, "step": 1475 }, { "epoch": 0.7921057032948654, "grad_norm": 7.3912636159559835, "learning_rate": 9.202743024992367e-07, "logits/chosen": -0.05757608264684677, "logits/rejected": 0.027957985177636147, "logps/chosen": -1.270872950553894, "logps/rejected": -1.4571425914764404, "loss": 1.6426, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.270872950553894, "rewards/margins": 0.18626955151557922, "rewards/rejected": -1.4571425914764404, "step": 1480 }, { "epoch": 0.7947817360762669, "grad_norm": 7.677644725215643, "learning_rate": 9.194285638083293e-07, "logits/chosen": -0.013660090044140816, "logits/rejected": 0.11817152798175812, "logps/chosen": -1.3152987957000732, "logps/rejected": -1.453066349029541, "loss": 1.6646, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3152987957000732, "rewards/margins": 0.13776752352714539, "rewards/rejected": -1.453066349029541, "step": 1485 }, { "epoch": 0.7974577688576685, "grad_norm": 10.387890790192381, "learning_rate": 9.185787558196562e-07, "logits/chosen": -0.0556030198931694, "logits/rejected": 0.036634549498558044, "logps/chosen": -1.2904164791107178, "logps/rejected": -1.3281852006912231, "loss": 1.6845, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.2904164791107178, "rewards/margins": 0.037768661975860596, "rewards/rejected": -1.3281852006912231, "step": 1490 }, { "epoch": 0.8001338016390701, "grad_norm": 9.354322301421194, "learning_rate": 9.177248867780583e-07, "logits/chosen": -0.06285317242145538, "logits/rejected": 0.03233514353632927, "logps/chosen": -1.3816922903060913, "logps/rejected": -1.4063470363616943, "loss": 1.7503, "rewards/accuracies": 0.5, "rewards/chosen": -1.3816922903060913, "rewards/margins": 0.024654611945152283, "rewards/rejected": -1.4063470363616943, "step": 1495 }, { "epoch": 0.8028098344204716, "grad_norm": 9.378817585928564, "learning_rate": 9.168669649677769e-07, "logits/chosen": -0.08374687284231186, "logits/rejected": 0.006220565643161535, "logps/chosen": -1.2730871438980103, "logps/rejected": -1.4184569120407104, "loss": 1.6471, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2730871438980103, "rewards/margins": 0.14536967873573303, "rewards/rejected": -1.4184569120407104, "step": 1500 }, { "epoch": 0.8054858672018732, "grad_norm": 9.036846103442565, "learning_rate": 9.16004998712373e-07, "logits/chosen": -0.0019377961289137602, "logits/rejected": 0.04818776249885559, "logps/chosen": -1.227207899093628, "logps/rejected": -1.445753812789917, "loss": 1.5801, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.227207899093628, "rewards/margins": 0.21854594349861145, "rewards/rejected": -1.445753812789917, "step": 1505 }, { "epoch": 0.8081618999832748, "grad_norm": 6.099930304574893, "learning_rate": 9.151389963746472e-07, "logits/chosen": -0.08609070628881454, "logits/rejected": 0.14547133445739746, "logps/chosen": -1.322655200958252, "logps/rejected": -1.4444901943206787, "loss": 1.6681, "rewards/accuracies": 0.5625, "rewards/chosen": -1.322655200958252, "rewards/margins": 0.12183494865894318, "rewards/rejected": -1.4444901943206787, "step": 1510 }, { "epoch": 0.8108379327646764, "grad_norm": 6.268135673470169, "learning_rate": 9.142689663565577e-07, "logits/chosen": -0.008268142119050026, "logits/rejected": 0.053254224359989166, "logps/chosen": -1.2595961093902588, "logps/rejected": -1.3987139463424683, "loss": 1.6466, "rewards/accuracies": 0.5, "rewards/chosen": -1.2595961093902588, "rewards/margins": 0.139118030667305, "rewards/rejected": -1.3987139463424683, "step": 1515 }, { "epoch": 0.8135139655460779, "grad_norm": 8.021966955793015, "learning_rate": 9.133949170991397e-07, "logits/chosen": -0.03498731926083565, "logits/rejected": 0.030461037531495094, "logps/chosen": -1.2982255220413208, "logps/rejected": -1.4231431484222412, "loss": 1.6557, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2982255220413208, "rewards/margins": 0.1249176487326622, "rewards/rejected": -1.4231431484222412, "step": 1520 }, { "epoch": 0.8161899983274795, "grad_norm": 7.341712205930606, "learning_rate": 9.125168570824231e-07, "logits/chosen": -0.061010561883449554, "logits/rejected": 0.0892036184668541, "logps/chosen": -1.2900292873382568, "logps/rejected": -1.3733108043670654, "loss": 1.6893, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2900292873382568, "rewards/margins": 0.08328154683113098, "rewards/rejected": -1.3733108043670654, "step": 1525 }, { "epoch": 0.8188660311088811, "grad_norm": 9.797618530298605, "learning_rate": 9.116347948253496e-07, "logits/chosen": -0.07041227072477341, "logits/rejected": 0.012045316398143768, "logps/chosen": -1.3219391107559204, "logps/rejected": -1.4183783531188965, "loss": 1.6705, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3219391107559204, "rewards/margins": 0.09643938392400742, "rewards/rejected": -1.4183783531188965, "step": 1530 }, { "epoch": 0.8215420638902826, "grad_norm": 9.662443664040012, "learning_rate": 9.107487388856916e-07, "logits/chosen": -0.10239896923303604, "logits/rejected": 0.0421513170003891, "logps/chosen": -1.2454922199249268, "logps/rejected": -1.4122923612594604, "loss": 1.6068, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2454922199249268, "rewards/margins": 0.16680024564266205, "rewards/rejected": -1.4122923612594604, "step": 1535 }, { "epoch": 0.8242180966716842, "grad_norm": 8.787552287478555, "learning_rate": 9.098586978599673e-07, "logits/chosen": -0.03155619651079178, "logits/rejected": 0.10633011907339096, "logps/chosen": -1.273524522781372, "logps/rejected": -1.489218831062317, "loss": 1.6086, "rewards/accuracies": 0.59375, "rewards/chosen": -1.273524522781372, "rewards/margins": 0.2156943827867508, "rewards/rejected": -1.489218831062317, "step": 1540 }, { "epoch": 0.8268941294530858, "grad_norm": 7.029763697711535, "learning_rate": 9.089646803833588e-07, "logits/chosen": -0.01791234128177166, "logits/rejected": 0.12071450799703598, "logps/chosen": -1.2818536758422852, "logps/rejected": -1.3673170804977417, "loss": 1.6724, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2818536758422852, "rewards/margins": 0.0854635089635849, "rewards/rejected": -1.3673170804977417, "step": 1545 }, { "epoch": 0.8295701622344873, "grad_norm": 8.037816116604773, "learning_rate": 9.080666951296276e-07, "logits/chosen": -0.1481930911540985, "logits/rejected": 0.08756627142429352, "logps/chosen": -1.3378673791885376, "logps/rejected": -1.4493402242660522, "loss": 1.7021, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3378673791885376, "rewards/margins": 0.11147283017635345, "rewards/rejected": -1.4493402242660522, "step": 1550 }, { "epoch": 0.8322461950158889, "grad_norm": 4.843415681313186, "learning_rate": 9.071647508110305e-07, "logits/chosen": -0.11743097007274628, "logits/rejected": 0.09829775989055634, "logps/chosen": -1.3695207834243774, "logps/rejected": -1.5013995170593262, "loss": 1.7087, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3695207834243774, "rewards/margins": 0.1318788379430771, "rewards/rejected": -1.5013995170593262, "step": 1555 }, { "epoch": 0.8349222277972905, "grad_norm": 7.42188862659024, "learning_rate": 9.062588561782354e-07, "logits/chosen": -0.01619536615908146, "logits/rejected": 0.039568670094013214, "logps/chosen": -1.3216458559036255, "logps/rejected": -1.4633032083511353, "loss": 1.6597, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3216458559036255, "rewards/margins": 0.14165739715099335, "rewards/rejected": -1.4633032083511353, "step": 1560 }, { "epoch": 0.8375982605786921, "grad_norm": 6.469587220819448, "learning_rate": 9.053490200202358e-07, "logits/chosen": -0.016276555135846138, "logits/rejected": 0.06772230565547943, "logps/chosen": -1.3213629722595215, "logps/rejected": -1.4255964756011963, "loss": 1.687, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3213629722595215, "rewards/margins": 0.10423363745212555, "rewards/rejected": -1.4255964756011963, "step": 1565 }, { "epoch": 0.8402742933600936, "grad_norm": 7.895324039537961, "learning_rate": 9.044352511642661e-07, "logits/chosen": 0.025027606636285782, "logits/rejected": 0.029301051050424576, "logps/chosen": -1.219848871231079, "logps/rejected": -1.3602640628814697, "loss": 1.6118, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.219848871231079, "rewards/margins": 0.14041535556316376, "rewards/rejected": -1.3602640628814697, "step": 1570 }, { "epoch": 0.8429503261414952, "grad_norm": 8.004495331877422, "learning_rate": 9.03517558475716e-07, "logits/chosen": -0.043523408472537994, "logits/rejected": 0.042294926941394806, "logps/chosen": -1.2938941717147827, "logps/rejected": -1.3521257638931274, "loss": 1.6673, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2938941717147827, "rewards/margins": 0.058231599628925323, "rewards/rejected": -1.3521257638931274, "step": 1575 }, { "epoch": 0.8456263589228968, "grad_norm": 6.624735016894455, "learning_rate": 9.025959508580436e-07, "logits/chosen": 0.030850157141685486, "logits/rejected": 0.24259766936302185, "logps/chosen": -1.286088228225708, "logps/rejected": -1.421175241470337, "loss": 1.6249, "rewards/accuracies": 0.5625, "rewards/chosen": -1.286088228225708, "rewards/margins": 0.13508695363998413, "rewards/rejected": -1.421175241470337, "step": 1580 }, { "epoch": 0.8483023917042983, "grad_norm": 6.966093077574928, "learning_rate": 9.016704372526905e-07, "logits/chosen": -0.01990427076816559, "logits/rejected": 0.11765620857477188, "logps/chosen": -1.2193325757980347, "logps/rejected": -1.467350721359253, "loss": 1.5908, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2193325757980347, "rewards/margins": 0.24801822006702423, "rewards/rejected": -1.467350721359253, "step": 1585 }, { "epoch": 0.8509784244856999, "grad_norm": 7.458677237507632, "learning_rate": 9.007410266389934e-07, "logits/chosen": -0.07480472326278687, "logits/rejected": 0.004531972110271454, "logps/chosen": -1.2420049905776978, "logps/rejected": -1.3440052270889282, "loss": 1.6286, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2420049905776978, "rewards/margins": 0.10200023651123047, "rewards/rejected": -1.3440052270889282, "step": 1590 }, { "epoch": 0.8536544572671015, "grad_norm": 9.279232253571516, "learning_rate": 8.998077280340981e-07, "logits/chosen": -0.017351821064949036, "logits/rejected": 0.05029697343707085, "logps/chosen": -1.3600581884384155, "logps/rejected": -1.3599737882614136, "loss": 1.7314, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.3600581884384155, "rewards/margins": -8.444488048553467e-05, "rewards/rejected": -1.3599737882614136, "step": 1595 }, { "epoch": 0.8563304900485031, "grad_norm": 6.8897238465074215, "learning_rate": 8.988705504928722e-07, "logits/chosen": -0.13428761065006256, "logits/rejected": 0.0201809611171484, "logps/chosen": -1.3263137340545654, "logps/rejected": -1.500237226486206, "loss": 1.6569, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3263137340545654, "rewards/margins": 0.17392337322235107, "rewards/rejected": -1.500237226486206, "step": 1600 }, { "epoch": 0.8563304900485031, "eval_logits/chosen": 0.18784047663211823, "eval_logits/rejected": 0.2617562711238861, "eval_logps/chosen": -1.3093810081481934, "eval_logps/rejected": -1.4585061073303223, "eval_loss": 1.6597652435302734, "eval_rewards/accuracies": 0.5593471527099609, "eval_rewards/chosen": -1.3093810081481934, "eval_rewards/margins": 0.14912521839141846, "eval_rewards/rejected": -1.4585061073303223, "eval_runtime": 40.2695, "eval_samples_per_second": 33.4, "eval_steps_per_second": 8.369, "step": 1600 }, { "epoch": 0.8590065228299046, "grad_norm": 5.203434488845097, "learning_rate": 8.979295031078157e-07, "logits/chosen": -0.12461533397436142, "logits/rejected": 0.062472231686115265, "logps/chosen": -1.2665091753005981, "logps/rejected": -1.4833637475967407, "loss": 1.6001, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2665091753005981, "rewards/margins": 0.21685440838336945, "rewards/rejected": -1.4833637475967407, "step": 1605 }, { "epoch": 0.8616825556113062, "grad_norm": 5.831112519526011, "learning_rate": 8.969845950089751e-07, "logits/chosen": -0.1279621422290802, "logits/rejected": 0.02661943808197975, "logps/chosen": -1.2545493841171265, "logps/rejected": -1.4126601219177246, "loss": 1.6204, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2545493841171265, "rewards/margins": 0.1581106185913086, "rewards/rejected": -1.4126601219177246, "step": 1610 }, { "epoch": 0.8643585883927078, "grad_norm": 7.223114448938468, "learning_rate": 8.960358353638526e-07, "logits/chosen": -0.0668589398264885, "logits/rejected": 0.021599432453513145, "logps/chosen": -1.3225297927856445, "logps/rejected": -1.5004407167434692, "loss": 1.6714, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3225297927856445, "rewards/margins": 0.17791098356246948, "rewards/rejected": -1.5004407167434692, "step": 1615 }, { "epoch": 0.8670346211741093, "grad_norm": 6.852298379018131, "learning_rate": 8.950832333773184e-07, "logits/chosen": -0.015650823712348938, "logits/rejected": 0.10268108546733856, "logps/chosen": -1.19065260887146, "logps/rejected": -1.4471497535705566, "loss": 1.554, "rewards/accuracies": 0.59375, "rewards/chosen": -1.19065260887146, "rewards/margins": 0.2564970850944519, "rewards/rejected": -1.4471497535705566, "step": 1620 }, { "epoch": 0.869710653955511, "grad_norm": 8.250958293931829, "learning_rate": 8.941267982915213e-07, "logits/chosen": 0.06653941422700882, "logits/rejected": 0.11267844587564468, "logps/chosen": -1.3536145687103271, "logps/rejected": -1.4748936891555786, "loss": 1.7003, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3536145687103271, "rewards/margins": 0.12127907574176788, "rewards/rejected": -1.4748936891555786, "step": 1625 }, { "epoch": 0.8723866867369126, "grad_norm": 7.516947224333936, "learning_rate": 8.931665393857983e-07, "logits/chosen": 0.02407361939549446, "logits/rejected": 0.16263404488563538, "logps/chosen": -1.311852216720581, "logps/rejected": -1.3916914463043213, "loss": 1.6763, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.311852216720581, "rewards/margins": 0.07983918488025665, "rewards/rejected": -1.3916914463043213, "step": 1630 }, { "epoch": 0.875062719518314, "grad_norm": 11.051621336233081, "learning_rate": 8.922024659765861e-07, "logits/chosen": -0.08223044127225876, "logits/rejected": 0.019241278991103172, "logps/chosen": -1.2253283262252808, "logps/rejected": -1.378430724143982, "loss": 1.6116, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2253283262252808, "rewards/margins": 0.15310248732566833, "rewards/rejected": -1.378430724143982, "step": 1635 }, { "epoch": 0.8777387522997157, "grad_norm": 6.939402757279656, "learning_rate": 8.912345874173288e-07, "logits/chosen": -0.073940709233284, "logits/rejected": 0.01882219687104225, "logps/chosen": -1.2232308387756348, "logps/rejected": -1.3898383378982544, "loss": 1.6271, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2232308387756348, "rewards/margins": 0.1666075587272644, "rewards/rejected": -1.3898383378982544, "step": 1640 }, { "epoch": 0.8804147850811173, "grad_norm": 10.12377883475872, "learning_rate": 8.902629130983885e-07, "logits/chosen": -0.04563382267951965, "logits/rejected": -0.0019987802952528, "logps/chosen": -1.2227944135665894, "logps/rejected": -1.366692304611206, "loss": 1.6104, "rewards/accuracies": 0.625, "rewards/chosen": -1.2227944135665894, "rewards/margins": 0.1438979208469391, "rewards/rejected": -1.366692304611206, "step": 1645 }, { "epoch": 0.8830908178625189, "grad_norm": 11.195083737144992, "learning_rate": 8.892874524469537e-07, "logits/chosen": 0.01719006523489952, "logits/rejected": 0.07226898521184921, "logps/chosen": -1.2502089738845825, "logps/rejected": -1.4320770502090454, "loss": 1.6281, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2502089738845825, "rewards/margins": 0.18186809122562408, "rewards/rejected": -1.4320770502090454, "step": 1650 }, { "epoch": 0.8857668506439204, "grad_norm": 10.210347433071322, "learning_rate": 8.883082149269478e-07, "logits/chosen": -0.10980989784002304, "logits/rejected": -0.023154709488153458, "logps/chosen": -1.3029086589813232, "logps/rejected": -1.3899312019348145, "loss": 1.7001, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3029086589813232, "rewards/margins": 0.08702269196510315, "rewards/rejected": -1.3899312019348145, "step": 1655 }, { "epoch": 0.888442883425322, "grad_norm": 6.521466025549045, "learning_rate": 8.873252100389377e-07, "logits/chosen": -0.021430622786283493, "logits/rejected": -0.02687462605535984, "logps/chosen": -1.2348066568374634, "logps/rejected": -1.4026113748550415, "loss": 1.6231, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2348066568374634, "rewards/margins": 0.167804554104805, "rewards/rejected": -1.4026113748550415, "step": 1660 }, { "epoch": 0.8911189162067236, "grad_norm": 6.715722948427614, "learning_rate": 8.863384473200411e-07, "logits/chosen": -0.07429514080286026, "logits/rejected": -0.014555975794792175, "logps/chosen": -1.3120548725128174, "logps/rejected": -1.4360677003860474, "loss": 1.6682, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3120548725128174, "rewards/margins": 0.1240130066871643, "rewards/rejected": -1.4360677003860474, "step": 1665 }, { "epoch": 0.8937949489881251, "grad_norm": 6.370698833816654, "learning_rate": 8.853479363438342e-07, "logits/chosen": -0.030076956376433372, "logits/rejected": 0.10825265944004059, "logps/chosen": -1.3282768726348877, "logps/rejected": -1.3937517404556274, "loss": 1.7328, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -1.3282768726348877, "rewards/margins": 0.0654749944806099, "rewards/rejected": -1.3937517404556274, "step": 1670 }, { "epoch": 0.8964709817695267, "grad_norm": 5.216425362326687, "learning_rate": 8.843536867202588e-07, "logits/chosen": -0.03181172162294388, "logits/rejected": 0.1464720368385315, "logps/chosen": -1.3180739879608154, "logps/rejected": -1.5350233316421509, "loss": 1.6808, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3180739879608154, "rewards/margins": 0.21694926917552948, "rewards/rejected": -1.5350233316421509, "step": 1675 }, { "epoch": 0.8991470145509283, "grad_norm": 7.1767344877196395, "learning_rate": 8.833557080955292e-07, "logits/chosen": -0.12470666319131851, "logits/rejected": -0.026866847649216652, "logps/chosen": -1.346660852432251, "logps/rejected": -1.456504225730896, "loss": 1.6877, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.346660852432251, "rewards/margins": 0.10984358936548233, "rewards/rejected": -1.456504225730896, "step": 1680 }, { "epoch": 0.9018230473323299, "grad_norm": 8.837202279598182, "learning_rate": 8.823540101520381e-07, "logits/chosen": -0.11130370944738388, "logits/rejected": 0.0831967443227768, "logps/chosen": -1.2970563173294067, "logps/rejected": -1.4199358224868774, "loss": 1.6593, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2970563173294067, "rewards/margins": 0.12287943065166473, "rewards/rejected": -1.4199358224868774, "step": 1685 }, { "epoch": 0.9044990801137314, "grad_norm": 7.101092718565959, "learning_rate": 8.813486026082637e-07, "logits/chosen": -0.09423160552978516, "logits/rejected": 0.0658324807882309, "logps/chosen": -1.2349854707717896, "logps/rejected": -1.4053113460540771, "loss": 1.5942, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2349854707717896, "rewards/margins": 0.1703261435031891, "rewards/rejected": -1.4053113460540771, "step": 1690 }, { "epoch": 0.907175112895133, "grad_norm": 11.380726305489514, "learning_rate": 8.803394952186742e-07, "logits/chosen": -0.22291898727416992, "logits/rejected": -0.10640467703342438, "logps/chosen": -1.3459924459457397, "logps/rejected": -1.4340465068817139, "loss": 1.6934, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3459924459457397, "rewards/margins": 0.08805397152900696, "rewards/rejected": -1.4340465068817139, "step": 1695 }, { "epoch": 0.9098511456765346, "grad_norm": 8.462265361008011, "learning_rate": 8.793266977736342e-07, "logits/chosen": -0.07550237327814102, "logits/rejected": -0.108060821890831, "logps/chosen": -1.3348376750946045, "logps/rejected": -1.3659391403198242, "loss": 1.7038, "rewards/accuracies": 0.5, "rewards/chosen": -1.3348376750946045, "rewards/margins": 0.031101513653993607, "rewards/rejected": -1.3659391403198242, "step": 1700 }, { "epoch": 0.9125271784579361, "grad_norm": 8.429614512956558, "learning_rate": 8.783102200993085e-07, "logits/chosen": -0.05549921840429306, "logits/rejected": 0.06219468265771866, "logps/chosen": -1.3043372631072998, "logps/rejected": -1.3859697580337524, "loss": 1.6764, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3043372631072998, "rewards/margins": 0.08163245767354965, "rewards/rejected": -1.3859697580337524, "step": 1705 }, { "epoch": 0.9152032112393377, "grad_norm": 7.277943305755828, "learning_rate": 8.772900720575683e-07, "logits/chosen": -0.08484108000993729, "logits/rejected": -0.018476814031600952, "logps/chosen": -1.2428280115127563, "logps/rejected": -1.3736469745635986, "loss": 1.6092, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2428280115127563, "rewards/margins": 0.13081900775432587, "rewards/rejected": -1.3736469745635986, "step": 1710 }, { "epoch": 0.9178792440207393, "grad_norm": 6.586317878710247, "learning_rate": 8.762662635458944e-07, "logits/chosen": -0.05393046885728836, "logits/rejected": 0.13396093249320984, "logps/chosen": -1.341209053993225, "logps/rejected": -1.4041177034378052, "loss": 1.7144, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.341209053993225, "rewards/margins": 0.06290876865386963, "rewards/rejected": -1.4041177034378052, "step": 1715 }, { "epoch": 0.9205552768021408, "grad_norm": 6.562643912935402, "learning_rate": 8.752388044972811e-07, "logits/chosen": -0.05539485067129135, "logits/rejected": -0.0037333101499825716, "logps/chosen": -1.1808788776397705, "logps/rejected": -1.4093412160873413, "loss": 1.5732, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1808788776397705, "rewards/margins": 0.22846241295337677, "rewards/rejected": -1.4093412160873413, "step": 1720 }, { "epoch": 0.9232313095835424, "grad_norm": 6.781443857743248, "learning_rate": 8.74207704880141e-07, "logits/chosen": -0.09590543806552887, "logits/rejected": -0.013217410072684288, "logps/chosen": -1.315438985824585, "logps/rejected": -1.5088715553283691, "loss": 1.6445, "rewards/accuracies": 0.53125, "rewards/chosen": -1.315438985824585, "rewards/margins": 0.19343265891075134, "rewards/rejected": -1.5088715553283691, "step": 1725 }, { "epoch": 0.925907342364944, "grad_norm": 9.628500492354721, "learning_rate": 8.731729746982068e-07, "logits/chosen": -0.014389969408512115, "logits/rejected": 0.02294905297458172, "logps/chosen": -1.2849242687225342, "logps/rejected": -1.3661396503448486, "loss": 1.6693, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2849242687225342, "rewards/margins": 0.08121520280838013, "rewards/rejected": -1.3661396503448486, "step": 1730 }, { "epoch": 0.9285833751463456, "grad_norm": 7.225324232180578, "learning_rate": 8.721346239904355e-07, "logits/chosen": -0.1296091377735138, "logits/rejected": -0.0007158033549785614, "logps/chosen": -1.219684362411499, "logps/rejected": -1.5547126531600952, "loss": 1.5572, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.219684362411499, "rewards/margins": 0.33502835035324097, "rewards/rejected": -1.5547126531600952, "step": 1735 }, { "epoch": 0.9312594079277471, "grad_norm": 6.684078242296239, "learning_rate": 8.710926628309101e-07, "logits/chosen": -0.10116715729236603, "logits/rejected": 0.00490850955247879, "logps/chosen": -1.2695972919464111, "logps/rejected": -1.3845787048339844, "loss": 1.6351, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2695972919464111, "rewards/margins": 0.11498130857944489, "rewards/rejected": -1.3845787048339844, "step": 1740 }, { "epoch": 0.9339354407091487, "grad_norm": 6.928267081500885, "learning_rate": 8.700471013287424e-07, "logits/chosen": -0.04399807006120682, "logits/rejected": -0.02890801429748535, "logps/chosen": -1.2690376043319702, "logps/rejected": -1.3916754722595215, "loss": 1.6372, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2690376043319702, "rewards/margins": 0.12263789027929306, "rewards/rejected": -1.3916754722595215, "step": 1745 }, { "epoch": 0.9366114734905503, "grad_norm": 10.873837266637638, "learning_rate": 8.689979496279746e-07, "logits/chosen": -0.09588148444890976, "logits/rejected": -0.04271275922656059, "logps/chosen": -1.267203450202942, "logps/rejected": -1.4820235967636108, "loss": 1.6022, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.267203450202942, "rewards/margins": 0.21482017636299133, "rewards/rejected": -1.4820235967636108, "step": 1750 }, { "epoch": 0.9392875062719518, "grad_norm": 5.799402693567249, "learning_rate": 8.679452179074811e-07, "logits/chosen": -0.10049313306808472, "logits/rejected": -0.03091563656926155, "logps/chosen": -1.2542965412139893, "logps/rejected": -1.3590319156646729, "loss": 1.6401, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2542965412139893, "rewards/margins": 0.1047353744506836, "rewards/rejected": -1.3590319156646729, "step": 1755 }, { "epoch": 0.9419635390533534, "grad_norm": 6.570245639505618, "learning_rate": 8.668889163808698e-07, "logits/chosen": -0.04846472293138504, "logits/rejected": 0.04894987493753433, "logps/chosen": -1.2414864301681519, "logps/rejected": -1.3920495510101318, "loss": 1.6123, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2414864301681519, "rewards/margins": 0.15056300163269043, "rewards/rejected": -1.3920495510101318, "step": 1760 }, { "epoch": 0.944639571834755, "grad_norm": 6.5262857057583945, "learning_rate": 8.658290552963827e-07, "logits/chosen": -0.07380203902721405, "logits/rejected": -0.06247757002711296, "logps/chosen": -1.2712262868881226, "logps/rejected": -1.428794503211975, "loss": 1.6511, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2712262868881226, "rewards/margins": 0.15756818652153015, "rewards/rejected": -1.428794503211975, "step": 1765 }, { "epoch": 0.9473156046161565, "grad_norm": 5.698089935456351, "learning_rate": 8.647656449367966e-07, "logits/chosen": -0.07224474102258682, "logits/rejected": 0.056889694184064865, "logps/chosen": -1.2968194484710693, "logps/rejected": -1.3564350605010986, "loss": 1.6635, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2968194484710693, "rewards/margins": 0.059615641832351685, "rewards/rejected": -1.3564350605010986, "step": 1770 }, { "epoch": 0.9499916373975581, "grad_norm": 8.571788954671964, "learning_rate": 8.636986956193235e-07, "logits/chosen": -0.10607154667377472, "logits/rejected": -0.044310472905635834, "logps/chosen": -1.2398325204849243, "logps/rejected": -1.3684263229370117, "loss": 1.6144, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2398325204849243, "rewards/margins": 0.12859384715557098, "rewards/rejected": -1.3684263229370117, "step": 1775 }, { "epoch": 0.9526676701789597, "grad_norm": 6.323104023838116, "learning_rate": 8.626282176955104e-07, "logits/chosen": -0.10209281742572784, "logits/rejected": -0.0055378819815814495, "logps/chosen": -1.2686989307403564, "logps/rejected": -1.4309422969818115, "loss": 1.6258, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2686989307403564, "rewards/margins": 0.16224347054958344, "rewards/rejected": -1.4309422969818115, "step": 1780 }, { "epoch": 0.9553437029603613, "grad_norm": 7.238973153918211, "learning_rate": 8.615542215511389e-07, "logits/chosen": -0.0030011646449565887, "logits/rejected": 0.05824938416481018, "logps/chosen": -1.2141025066375732, "logps/rejected": -1.3099324703216553, "loss": 1.6379, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2141025066375732, "rewards/margins": 0.09582982957363129, "rewards/rejected": -1.3099324703216553, "step": 1785 }, { "epoch": 0.9580197357417628, "grad_norm": 7.513938150549295, "learning_rate": 8.604767176061241e-07, "logits/chosen": -0.01581764593720436, "logits/rejected": 0.01219199039041996, "logps/chosen": -1.2912495136260986, "logps/rejected": -1.4009100198745728, "loss": 1.6528, "rewards/accuracies": 0.5, "rewards/chosen": -1.2912495136260986, "rewards/margins": 0.10966048389673233, "rewards/rejected": -1.4009100198745728, "step": 1790 }, { "epoch": 0.9606957685231644, "grad_norm": 7.930001609958226, "learning_rate": 8.593957163144141e-07, "logits/chosen": -0.10950282961130142, "logits/rejected": 0.006719866301864386, "logps/chosen": -1.2600187063217163, "logps/rejected": -1.407183289527893, "loss": 1.628, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2600187063217163, "rewards/margins": 0.1471645087003708, "rewards/rejected": -1.407183289527893, "step": 1795 }, { "epoch": 0.963371801304566, "grad_norm": 5.2855450798311345, "learning_rate": 8.58311228163888e-07, "logits/chosen": -0.048804882913827896, "logits/rejected": 0.008072657510638237, "logps/chosen": -1.2580320835113525, "logps/rejected": -1.348109245300293, "loss": 1.6488, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2580320835113525, "rewards/margins": 0.09007705003023148, "rewards/rejected": -1.348109245300293, "step": 1800 }, { "epoch": 0.9660478340859675, "grad_norm": 5.961444422853368, "learning_rate": 8.57223263676255e-07, "logits/chosen": -0.17057205736637115, "logits/rejected": -0.06418205797672272, "logps/chosen": -1.2190537452697754, "logps/rejected": -1.4326839447021484, "loss": 1.5672, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2190537452697754, "rewards/margins": 0.21363027393817902, "rewards/rejected": -1.4326839447021484, "step": 1805 }, { "epoch": 0.9687238668673691, "grad_norm": 7.644020895661337, "learning_rate": 8.561318334069511e-07, "logits/chosen": -0.07568531483411789, "logits/rejected": 0.04486411437392235, "logps/chosen": -1.2447948455810547, "logps/rejected": -1.3525478839874268, "loss": 1.6348, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2447948455810547, "rewards/margins": 0.10775311291217804, "rewards/rejected": -1.3525478839874268, "step": 1810 }, { "epoch": 0.9713998996487707, "grad_norm": 6.165154965788141, "learning_rate": 8.550369479450375e-07, "logits/chosen": -0.08772541582584381, "logits/rejected": 0.012555623427033424, "logps/chosen": -1.2673677206039429, "logps/rejected": -1.3656182289123535, "loss": 1.6303, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2673677206039429, "rewards/margins": 0.09825052320957184, "rewards/rejected": -1.3656182289123535, "step": 1815 }, { "epoch": 0.9740759324301723, "grad_norm": 7.377835800463116, "learning_rate": 8.539386179130977e-07, "logits/chosen": -0.0693771094083786, "logits/rejected": -0.01106178667396307, "logps/chosen": -1.25356125831604, "logps/rejected": -1.3627688884735107, "loss": 1.6323, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.25356125831604, "rewards/margins": 0.10920742899179459, "rewards/rejected": -1.3627688884735107, "step": 1820 }, { "epoch": 0.9767519652115738, "grad_norm": 7.934772543113003, "learning_rate": 8.528368539671347e-07, "logits/chosen": -0.11911240965127945, "logits/rejected": -0.001425326569005847, "logps/chosen": -1.2460615634918213, "logps/rejected": -1.41835618019104, "loss": 1.6204, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2460615634918213, "rewards/margins": 0.1722945123910904, "rewards/rejected": -1.41835618019104, "step": 1825 }, { "epoch": 0.9794279979929754, "grad_norm": 6.007705828704846, "learning_rate": 8.51731666796467e-07, "logits/chosen": 0.026783233508467674, "logits/rejected": 0.04644298553466797, "logps/chosen": -1.3235788345336914, "logps/rejected": -1.358957052230835, "loss": 1.7024, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3235788345336914, "rewards/margins": 0.03537829965353012, "rewards/rejected": -1.358957052230835, "step": 1830 }, { "epoch": 0.982104030774377, "grad_norm": 7.568868917842844, "learning_rate": 8.506230671236254e-07, "logits/chosen": -0.10422410070896149, "logits/rejected": -0.06238463521003723, "logps/chosen": -1.2823209762573242, "logps/rejected": -1.3079733848571777, "loss": 1.6726, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2823209762573242, "rewards/margins": 0.02565241977572441, "rewards/rejected": -1.3079733848571777, "step": 1835 }, { "epoch": 0.9847800635557785, "grad_norm": 6.762765506914262, "learning_rate": 8.495110657042488e-07, "logits/chosen": -0.013921190984547138, "logits/rejected": 0.07133986055850983, "logps/chosen": -1.278881311416626, "logps/rejected": -1.4670178890228271, "loss": 1.6263, "rewards/accuracies": 0.5625, "rewards/chosen": -1.278881311416626, "rewards/margins": 0.18813663721084595, "rewards/rejected": -1.4670178890228271, "step": 1840 }, { "epoch": 0.9874560963371801, "grad_norm": 9.237192200041854, "learning_rate": 8.483956733269799e-07, "logits/chosen": -0.09804896265268326, "logits/rejected": -0.021493665874004364, "logps/chosen": -1.3015546798706055, "logps/rejected": -1.3693385124206543, "loss": 1.6763, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.3015546798706055, "rewards/margins": 0.06778378784656525, "rewards/rejected": -1.3693385124206543, "step": 1845 }, { "epoch": 0.9901321291185817, "grad_norm": 6.283239648580291, "learning_rate": 8.472769008133602e-07, "logits/chosen": -0.19651444256305695, "logits/rejected": -0.09052984416484833, "logps/chosen": -1.3121778964996338, "logps/rejected": -1.3153350353240967, "loss": 1.692, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.3121778964996338, "rewards/margins": 0.003157214028760791, "rewards/rejected": -1.3153350353240967, "step": 1850 }, { "epoch": 0.9928081618999832, "grad_norm": 6.419885578451689, "learning_rate": 8.461547590177259e-07, "logits/chosen": -0.0700322762131691, "logits/rejected": 0.012923069298267365, "logps/chosen": -1.2328972816467285, "logps/rejected": -1.3694627285003662, "loss": 1.6283, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2328972816467285, "rewards/margins": 0.13656553626060486, "rewards/rejected": -1.3694627285003662, "step": 1855 }, { "epoch": 0.9954841946813848, "grad_norm": 7.90071898604423, "learning_rate": 8.450292588271014e-07, "logits/chosen": -0.06717083603143692, "logits/rejected": -0.01019772794097662, "logps/chosen": -1.3151053190231323, "logps/rejected": -1.4005153179168701, "loss": 1.6626, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3151053190231323, "rewards/margins": 0.08540989458560944, "rewards/rejected": -1.4005153179168701, "step": 1860 }, { "epoch": 0.9981602274627864, "grad_norm": 6.849933687630706, "learning_rate": 8.439004111610945e-07, "logits/chosen": -0.07957644760608673, "logits/rejected": -0.012797405943274498, "logps/chosen": -1.1850707530975342, "logps/rejected": -1.4084447622299194, "loss": 1.5734, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1850707530975342, "rewards/margins": 0.22337403893470764, "rewards/rejected": -1.4084447622299194, "step": 1865 }, { "epoch": 1.000836260244188, "grad_norm": 7.127300215535271, "learning_rate": 8.427682269717901e-07, "logits/chosen": -0.12439294904470444, "logits/rejected": 0.008108256384730339, "logps/chosen": -1.329679012298584, "logps/rejected": -1.378037452697754, "loss": 1.6945, "rewards/accuracies": 0.46875, "rewards/chosen": -1.329679012298584, "rewards/margins": 0.04835844039916992, "rewards/rejected": -1.378037452697754, "step": 1870 }, { "epoch": 1.0035122930255895, "grad_norm": 5.767377885020671, "learning_rate": 8.416327172436446e-07, "logits/chosen": -0.1611921340227127, "logits/rejected": -0.04769011586904526, "logps/chosen": -1.3013145923614502, "logps/rejected": -1.3885616064071655, "loss": 1.6582, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3013145923614502, "rewards/margins": 0.0872470885515213, "rewards/rejected": -1.3885616064071655, "step": 1875 }, { "epoch": 1.0061883258069912, "grad_norm": 7.775122590944508, "learning_rate": 8.404938929933778e-07, "logits/chosen": -0.04842232167720795, "logits/rejected": 0.10100684314966202, "logps/chosen": -1.2551026344299316, "logps/rejected": -1.5187945365905762, "loss": 1.6163, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2551026344299316, "rewards/margins": 0.2636919319629669, "rewards/rejected": -1.5187945365905762, "step": 1880 }, { "epoch": 1.0088643585883927, "grad_norm": 7.56258171453072, "learning_rate": 8.39351765269868e-07, "logits/chosen": -0.08237516134977341, "logits/rejected": -0.016337528824806213, "logps/chosen": -1.212425947189331, "logps/rejected": -1.3873404264450073, "loss": 1.5975, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.212425947189331, "rewards/margins": 0.17491456866264343, "rewards/rejected": -1.3873404264450073, "step": 1885 }, { "epoch": 1.0115403913697942, "grad_norm": 6.432102596180142, "learning_rate": 8.382063451540431e-07, "logits/chosen": -0.08515187352895737, "logits/rejected": 0.08776731789112091, "logps/chosen": -1.2676560878753662, "logps/rejected": -1.41463303565979, "loss": 1.6642, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2676560878753662, "rewards/margins": 0.146977037191391, "rewards/rejected": -1.41463303565979, "step": 1890 }, { "epoch": 1.014216424151196, "grad_norm": 7.346269675416436, "learning_rate": 8.370576437587742e-07, "logits/chosen": -0.03652515262365341, "logits/rejected": 0.013286001980304718, "logps/chosen": -1.2494899034500122, "logps/rejected": -1.3886620998382568, "loss": 1.6114, "rewards/accuracies": 0.625, "rewards/chosen": -1.2494899034500122, "rewards/margins": 0.13917216658592224, "rewards/rejected": -1.3886620998382568, "step": 1895 }, { "epoch": 1.0168924569325974, "grad_norm": 7.157047781782498, "learning_rate": 8.359056722287674e-07, "logits/chosen": -0.144033744931221, "logits/rejected": 0.07644566893577576, "logps/chosen": -1.2893970012664795, "logps/rejected": -1.3897775411605835, "loss": 1.6615, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2893970012664795, "rewards/margins": 0.10038057714700699, "rewards/rejected": -1.3897775411605835, "step": 1900 }, { "epoch": 1.019568489713999, "grad_norm": 6.635519431216746, "learning_rate": 8.347504417404553e-07, "logits/chosen": -0.07440408319234848, "logits/rejected": 0.04772048071026802, "logps/chosen": -1.3085484504699707, "logps/rejected": -1.4066262245178223, "loss": 1.6716, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3085484504699707, "rewards/margins": 0.09807775914669037, "rewards/rejected": -1.4066262245178223, "step": 1905 }, { "epoch": 1.0222445224954007, "grad_norm": 7.46897956636028, "learning_rate": 8.335919635018893e-07, "logits/chosen": -0.1354013830423355, "logits/rejected": -0.029260188341140747, "logps/chosen": -1.272361397743225, "logps/rejected": -1.4245942831039429, "loss": 1.646, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.272361397743225, "rewards/margins": 0.1522330790758133, "rewards/rejected": -1.4245942831039429, "step": 1910 }, { "epoch": 1.0249205552768021, "grad_norm": 5.021632127702659, "learning_rate": 8.324302487526303e-07, "logits/chosen": -0.0943860188126564, "logits/rejected": -0.033584389835596085, "logps/chosen": -1.237389326095581, "logps/rejected": -1.335540533065796, "loss": 1.6227, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.237389326095581, "rewards/margins": 0.09815128147602081, "rewards/rejected": -1.335540533065796, "step": 1915 }, { "epoch": 1.0275965880582036, "grad_norm": 6.477177263733326, "learning_rate": 8.312653087636398e-07, "logits/chosen": -0.11647417396306992, "logits/rejected": -0.0515231117606163, "logps/chosen": -1.1594212055206299, "logps/rejected": -1.3588650226593018, "loss": 1.5432, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1594212055206299, "rewards/margins": 0.19944405555725098, "rewards/rejected": -1.3588650226593018, "step": 1920 }, { "epoch": 1.0302726208396054, "grad_norm": 7.798364232564433, "learning_rate": 8.300971548371711e-07, "logits/chosen": -0.23913152515888214, "logits/rejected": -0.065396249294281, "logps/chosen": -1.3392633199691772, "logps/rejected": -1.415614128112793, "loss": 1.6877, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3392633199691772, "rewards/margins": 0.07635075598955154, "rewards/rejected": -1.415614128112793, "step": 1925 }, { "epoch": 1.0329486536210069, "grad_norm": 6.679655849773325, "learning_rate": 8.289257983066582e-07, "logits/chosen": -0.16242703795433044, "logits/rejected": -0.0466943196952343, "logps/chosen": -1.2052810192108154, "logps/rejected": -1.3960282802581787, "loss": 1.5734, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2052810192108154, "rewards/margins": 0.19074717164039612, "rewards/rejected": -1.3960282802581787, "step": 1930 }, { "epoch": 1.0356246864024083, "grad_norm": 6.569896272550323, "learning_rate": 8.277512505366077e-07, "logits/chosen": -0.17269453406333923, "logits/rejected": -0.024066952988505363, "logps/chosen": -1.2725989818572998, "logps/rejected": -1.4294071197509766, "loss": 1.6212, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2725989818572998, "rewards/margins": 0.15680822730064392, "rewards/rejected": -1.4294071197509766, "step": 1935 }, { "epoch": 1.03830071918381, "grad_norm": 7.3011255358224805, "learning_rate": 8.265735229224868e-07, "logits/chosen": -0.07453256845474243, "logits/rejected": 0.019622694700956345, "logps/chosen": -1.277066946029663, "logps/rejected": -1.4361835718154907, "loss": 1.6421, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.277066946029663, "rewards/margins": 0.15911665558815002, "rewards/rejected": -1.4361835718154907, "step": 1940 }, { "epoch": 1.0409767519652116, "grad_norm": 7.008864997055822, "learning_rate": 8.253926268906144e-07, "logits/chosen": -0.18078279495239258, "logits/rejected": -0.037943147122859955, "logps/chosen": -1.2641736268997192, "logps/rejected": -1.4041342735290527, "loss": 1.6017, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2641736268997192, "rewards/margins": 0.1399606615304947, "rewards/rejected": -1.4041342735290527, "step": 1945 }, { "epoch": 1.043652784746613, "grad_norm": 6.158357754291056, "learning_rate": 8.242085738980487e-07, "logits/chosen": -0.10943479835987091, "logits/rejected": 0.0673285573720932, "logps/chosen": -1.3101437091827393, "logps/rejected": -1.4342368841171265, "loss": 1.6474, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3101437091827393, "rewards/margins": 0.12409321218729019, "rewards/rejected": -1.4342368841171265, "step": 1950 }, { "epoch": 1.0463288175280148, "grad_norm": 7.083286067561728, "learning_rate": 8.230213754324772e-07, "logits/chosen": -0.05732116848230362, "logits/rejected": 0.0017935142386704683, "logps/chosen": -1.2026760578155518, "logps/rejected": -1.4104280471801758, "loss": 1.5549, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2026760578155518, "rewards/margins": 0.20775175094604492, "rewards/rejected": -1.4104280471801758, "step": 1955 }, { "epoch": 1.0490048503094163, "grad_norm": 6.817524204988108, "learning_rate": 8.218310430121045e-07, "logits/chosen": -0.14348937571048737, "logits/rejected": -0.12298564612865448, "logps/chosen": -1.26016104221344, "logps/rejected": -1.3855254650115967, "loss": 1.6371, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.26016104221344, "rewards/margins": 0.12536440789699554, "rewards/rejected": -1.3855254650115967, "step": 1960 }, { "epoch": 1.051680883090818, "grad_norm": 7.530792490915361, "learning_rate": 8.20637588185541e-07, "logits/chosen": -0.04563091695308685, "logits/rejected": 0.011556325480341911, "logps/chosen": -1.2070486545562744, "logps/rejected": -1.4804352521896362, "loss": 1.5382, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2070486545562744, "rewards/margins": 0.2733866274356842, "rewards/rejected": -1.4804352521896362, "step": 1965 }, { "epoch": 1.0543569158722195, "grad_norm": 8.37658162909406, "learning_rate": 8.194410225316906e-07, "logits/chosen": -0.10722582042217255, "logits/rejected": 0.012183678336441517, "logps/chosen": -1.271067500114441, "logps/rejected": -1.4413100481033325, "loss": 1.6331, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.271067500114441, "rewards/margins": 0.1702425330877304, "rewards/rejected": -1.4413100481033325, "step": 1970 }, { "epoch": 1.057032948653621, "grad_norm": 7.464246482195032, "learning_rate": 8.182413576596385e-07, "logits/chosen": -0.059704482555389404, "logits/rejected": 0.018035439774394035, "logps/chosen": -1.2023093700408936, "logps/rejected": -1.3934221267700195, "loss": 1.567, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2023093700408936, "rewards/margins": 0.19111260771751404, "rewards/rejected": -1.3934221267700195, "step": 1975 }, { "epoch": 1.0597089814350227, "grad_norm": 7.578940543456657, "learning_rate": 8.170386052085389e-07, "logits/chosen": -0.016566650941967964, "logits/rejected": 0.09114401787519455, "logps/chosen": -1.275981068611145, "logps/rejected": -1.4322946071624756, "loss": 1.6423, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.275981068611145, "rewards/margins": 0.15631352365016937, "rewards/rejected": -1.4322946071624756, "step": 1980 }, { "epoch": 1.0623850142164242, "grad_norm": 7.700307637421773, "learning_rate": 8.158327768475008e-07, "logits/chosen": -0.08740787953138351, "logits/rejected": 0.04793049767613411, "logps/chosen": -1.298185110092163, "logps/rejected": -1.3765360116958618, "loss": 1.6752, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.298185110092163, "rewards/margins": 0.07835081964731216, "rewards/rejected": -1.3765360116958618, "step": 1985 }, { "epoch": 1.0650610469978257, "grad_norm": 7.481636127449876, "learning_rate": 8.146238842754767e-07, "logits/chosen": -0.11278965324163437, "logits/rejected": -0.03838568925857544, "logps/chosen": -1.3026186227798462, "logps/rejected": -1.4424206018447876, "loss": 1.6447, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3026186227798462, "rewards/margins": 0.13980190455913544, "rewards/rejected": -1.4424206018447876, "step": 1990 }, { "epoch": 1.0677370797792274, "grad_norm": 7.134876372667975, "learning_rate": 8.134119392211476e-07, "logits/chosen": -0.012486773543059826, "logits/rejected": 0.11474494636058807, "logps/chosen": -1.2178418636322021, "logps/rejected": -1.470642328262329, "loss": 1.5763, "rewards/accuracies": 0.625, "rewards/chosen": -1.2178418636322021, "rewards/margins": 0.2528002858161926, "rewards/rejected": -1.470642328262329, "step": 1995 }, { "epoch": 1.0704131125606289, "grad_norm": 11.370675419727878, "learning_rate": 8.121969534428094e-07, "logits/chosen": -0.11129634082317352, "logits/rejected": 0.017330098897218704, "logps/chosen": -1.301973581314087, "logps/rejected": -1.3558018207550049, "loss": 1.7111, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -1.301973581314087, "rewards/margins": 0.05382819101214409, "rewards/rejected": -1.3558018207550049, "step": 2000 }, { "epoch": 1.0704131125606289, "eval_logits/chosen": 0.15607893466949463, "eval_logits/rejected": 0.22898563742637634, "eval_logps/chosen": -1.3001796007156372, "eval_logps/rejected": -1.4570097923278809, "eval_loss": 1.6547960042953491, "eval_rewards/accuracies": 0.5652819275856018, "eval_rewards/chosen": -1.3001796007156372, "eval_rewards/margins": 0.1568300873041153, "eval_rewards/rejected": -1.4570097923278809, "eval_runtime": 40.3494, "eval_samples_per_second": 33.334, "eval_steps_per_second": 8.352, "step": 2000 }, { "epoch": 1.0730891453420304, "grad_norm": 8.28139468174092, "learning_rate": 8.109789387282599e-07, "logits/chosen": -0.07982680201530457, "logits/rejected": -0.008802761323750019, "logps/chosen": -1.3038156032562256, "logps/rejected": -1.3784399032592773, "loss": 1.6776, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.3038156032562256, "rewards/margins": 0.07462429255247116, "rewards/rejected": -1.3784399032592773, "step": 2005 }, { "epoch": 1.075765178123432, "grad_norm": 9.331971218931889, "learning_rate": 8.097579068946827e-07, "logits/chosen": -0.04008934646844864, "logits/rejected": 0.04494168609380722, "logps/chosen": -1.2365713119506836, "logps/rejected": -1.3746750354766846, "loss": 1.6245, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2365713119506836, "rewards/margins": 0.13810376822948456, "rewards/rejected": -1.3746750354766846, "step": 2010 }, { "epoch": 1.0784412109048336, "grad_norm": 6.722924785559778, "learning_rate": 8.085338697885344e-07, "logits/chosen": -0.09236650913953781, "logits/rejected": 0.04134202376008034, "logps/chosen": -1.2058128118515015, "logps/rejected": -1.3785531520843506, "loss": 1.5911, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2058128118515015, "rewards/margins": 0.17274019122123718, "rewards/rejected": -1.3785531520843506, "step": 2015 }, { "epoch": 1.081117243686235, "grad_norm": 7.269682600004079, "learning_rate": 8.073068392854282e-07, "logits/chosen": -0.11385257542133331, "logits/rejected": 0.02374332770705223, "logps/chosen": -1.3075402975082397, "logps/rejected": -1.4661028385162354, "loss": 1.6516, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3075402975082397, "rewards/margins": 0.15856240689754486, "rewards/rejected": -1.4661028385162354, "step": 2020 }, { "epoch": 1.0837932764676368, "grad_norm": 7.28315693106649, "learning_rate": 8.060768272900193e-07, "logits/chosen": -0.034406136721372604, "logits/rejected": 0.0682801604270935, "logps/chosen": -1.2798516750335693, "logps/rejected": -1.4447687864303589, "loss": 1.6364, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2798516750335693, "rewards/margins": 0.16491718590259552, "rewards/rejected": -1.4447687864303589, "step": 2025 }, { "epoch": 1.0864693092490383, "grad_norm": 6.330620780904875, "learning_rate": 8.0484384573589e-07, "logits/chosen": -0.11100749671459198, "logits/rejected": -0.0917942225933075, "logps/chosen": -1.224623680114746, "logps/rejected": -1.3809386491775513, "loss": 1.6082, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.224623680114746, "rewards/margins": 0.15631495416164398, "rewards/rejected": -1.3809386491775513, "step": 2030 }, { "epoch": 1.0891453420304398, "grad_norm": 6.963132320526369, "learning_rate": 8.03607906585432e-07, "logits/chosen": -0.1358167678117752, "logits/rejected": 0.003392726182937622, "logps/chosen": -1.2268271446228027, "logps/rejected": -1.369664192199707, "loss": 1.5988, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2268271446228027, "rewards/margins": 0.14283707737922668, "rewards/rejected": -1.369664192199707, "step": 2035 }, { "epoch": 1.0918213748118415, "grad_norm": 8.895581081854896, "learning_rate": 8.023690218297329e-07, "logits/chosen": -0.1667802482843399, "logits/rejected": -0.11317966133356094, "logps/chosen": -1.244923710823059, "logps/rejected": -1.3557780981063843, "loss": 1.6213, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.244923710823059, "rewards/margins": 0.11085452139377594, "rewards/rejected": -1.3557780981063843, "step": 2040 }, { "epoch": 1.094497407593243, "grad_norm": 10.612181651266052, "learning_rate": 8.01127203488458e-07, "logits/chosen": -0.033369529992341995, "logits/rejected": 0.00568109517917037, "logps/chosen": -1.2427600622177124, "logps/rejected": -1.4127317667007446, "loss": 1.6157, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2427600622177124, "rewards/margins": 0.16997160017490387, "rewards/rejected": -1.4127317667007446, "step": 2045 }, { "epoch": 1.0971734403746445, "grad_norm": 8.355761392932418, "learning_rate": 7.998824636097339e-07, "logits/chosen": -0.1496758908033371, "logits/rejected": -0.055368490517139435, "logps/chosen": -1.2817127704620361, "logps/rejected": -1.3591550588607788, "loss": 1.6769, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2817127704620361, "rewards/margins": 0.07744228094816208, "rewards/rejected": -1.3591550588607788, "step": 2050 }, { "epoch": 1.0998494731560462, "grad_norm": 8.360987746496702, "learning_rate": 7.986348142700328e-07, "logits/chosen": -0.06102384254336357, "logits/rejected": 0.03928098827600479, "logps/chosen": -1.25723397731781, "logps/rejected": -1.3665558099746704, "loss": 1.6262, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.25723397731781, "rewards/margins": 0.10932165384292603, "rewards/rejected": -1.3665558099746704, "step": 2055 }, { "epoch": 1.1025255059374477, "grad_norm": 8.116749314927926, "learning_rate": 7.973842675740539e-07, "logits/chosen": -0.06331825256347656, "logits/rejected": -0.025836337357759476, "logps/chosen": -1.283013939857483, "logps/rejected": -1.4757953882217407, "loss": 1.6254, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.283013939857483, "rewards/margins": 0.1927814930677414, "rewards/rejected": -1.4757953882217407, "step": 2060 }, { "epoch": 1.1052015387188494, "grad_norm": 7.775977833971184, "learning_rate": 7.961308356546066e-07, "logits/chosen": -0.09919768571853638, "logits/rejected": 0.0063897669315338135, "logps/chosen": -1.2619423866271973, "logps/rejected": -1.3673630952835083, "loss": 1.6427, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2619423866271973, "rewards/margins": 0.10542067140340805, "rewards/rejected": -1.3673630952835083, "step": 2065 }, { "epoch": 1.107877571500251, "grad_norm": 8.689507448407964, "learning_rate": 7.948745306724931e-07, "logits/chosen": -0.12621857225894928, "logits/rejected": -0.0004501536604948342, "logps/chosen": -1.2156966924667358, "logps/rejected": -1.4456615447998047, "loss": 1.5808, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2156966924667358, "rewards/margins": 0.22996488213539124, "rewards/rejected": -1.4456615447998047, "step": 2070 }, { "epoch": 1.1105536042816524, "grad_norm": 7.854038469429008, "learning_rate": 7.936153648163897e-07, "logits/chosen": -0.13651198148727417, "logits/rejected": -0.053521834313869476, "logps/chosen": -1.2578051090240479, "logps/rejected": -1.453272819519043, "loss": 1.6072, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2578051090240479, "rewards/margins": 0.19546756148338318, "rewards/rejected": -1.453272819519043, "step": 2075 }, { "epoch": 1.1132296370630541, "grad_norm": 5.696613158096192, "learning_rate": 7.92353350302729e-07, "logits/chosen": -0.18901076912879944, "logits/rejected": -0.05768461897969246, "logps/chosen": -1.1884305477142334, "logps/rejected": -1.4610515832901, "loss": 1.5212, "rewards/accuracies": 0.625, "rewards/chosen": -1.1884305477142334, "rewards/margins": 0.2726210355758667, "rewards/rejected": -1.4610515832901, "step": 2080 }, { "epoch": 1.1159056698444556, "grad_norm": 11.533281987271195, "learning_rate": 7.910884993755816e-07, "logits/chosen": -0.14586128294467926, "logits/rejected": -0.04839394986629486, "logps/chosen": -1.2378690242767334, "logps/rejected": -1.3928966522216797, "loss": 1.6117, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2378690242767334, "rewards/margins": 0.15502753853797913, "rewards/rejected": -1.3928966522216797, "step": 2085 }, { "epoch": 1.118581702625857, "grad_norm": 8.700654931151222, "learning_rate": 7.898208243065367e-07, "logits/chosen": -0.21394655108451843, "logits/rejected": -0.2073047161102295, "logps/chosen": -1.2208964824676514, "logps/rejected": -1.3645246028900146, "loss": 1.6088, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2208964824676514, "rewards/margins": 0.14362801611423492, "rewards/rejected": -1.3645246028900146, "step": 2090 }, { "epoch": 1.1212577354072588, "grad_norm": 4.874928851192373, "learning_rate": 7.88550337394583e-07, "logits/chosen": -0.12973353266716003, "logits/rejected": -0.014638014137744904, "logps/chosen": -1.368638038635254, "logps/rejected": -1.481255292892456, "loss": 1.7235, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.368638038635254, "rewards/margins": 0.11261717975139618, "rewards/rejected": -1.481255292892456, "step": 2095 }, { "epoch": 1.1239337681886603, "grad_norm": 8.98215076654875, "learning_rate": 7.872770509659905e-07, "logits/chosen": -0.07033531367778778, "logits/rejected": -0.03589242324233055, "logps/chosen": -1.3756530284881592, "logps/rejected": -1.4444859027862549, "loss": 1.7215, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3756530284881592, "rewards/margins": 0.06883300095796585, "rewards/rejected": -1.4444859027862549, "step": 2100 }, { "epoch": 1.1266098009700618, "grad_norm": 7.2182694130911145, "learning_rate": 7.860009773741896e-07, "logits/chosen": -0.04326310753822327, "logits/rejected": 0.051185887306928635, "logps/chosen": -1.3004047870635986, "logps/rejected": -1.425777792930603, "loss": 1.6538, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3004047870635986, "rewards/margins": 0.12537285685539246, "rewards/rejected": -1.425777792930603, "step": 2105 }, { "epoch": 1.1292858337514635, "grad_norm": 8.322581432863783, "learning_rate": 7.84722128999652e-07, "logits/chosen": -0.1491938829421997, "logits/rejected": -0.009843626990914345, "logps/chosen": -1.2436976432800293, "logps/rejected": -1.5600513219833374, "loss": 1.5737, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2436976432800293, "rewards/margins": 0.31635379791259766, "rewards/rejected": -1.5600513219833374, "step": 2110 }, { "epoch": 1.131961866532865, "grad_norm": 7.902528377969128, "learning_rate": 7.834405182497699e-07, "logits/chosen": -0.03493956848978996, "logits/rejected": 0.010392325930297375, "logps/chosen": -1.248529314994812, "logps/rejected": -1.376386046409607, "loss": 1.6132, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.248529314994812, "rewards/margins": 0.12785673141479492, "rewards/rejected": -1.376386046409607, "step": 2115 }, { "epoch": 1.1346378993142665, "grad_norm": 8.606169225182978, "learning_rate": 7.821561575587368e-07, "logits/chosen": -0.15363888442516327, "logits/rejected": -0.12825623154640198, "logps/chosen": -1.2925937175750732, "logps/rejected": -1.417511224746704, "loss": 1.6548, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2925937175750732, "rewards/margins": 0.12491756677627563, "rewards/rejected": -1.417511224746704, "step": 2120 }, { "epoch": 1.1373139320956682, "grad_norm": 6.966458401018398, "learning_rate": 7.808690593874254e-07, "logits/chosen": -0.10021853446960449, "logits/rejected": -0.054100703448057175, "logps/chosen": -1.2008590698242188, "logps/rejected": -1.4212961196899414, "loss": 1.5527, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2008590698242188, "rewards/margins": 0.2204369753599167, "rewards/rejected": -1.4212961196899414, "step": 2125 }, { "epoch": 1.1399899648770697, "grad_norm": 8.922400788653567, "learning_rate": 7.79579236223268e-07, "logits/chosen": -0.067740797996521, "logits/rejected": 0.1286747008562088, "logps/chosen": -1.2899023294448853, "logps/rejected": -1.4537795782089233, "loss": 1.6364, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2899023294448853, "rewards/margins": 0.1638774424791336, "rewards/rejected": -1.4537795782089233, "step": 2130 }, { "epoch": 1.1426659976584714, "grad_norm": 6.447264543939956, "learning_rate": 7.782867005801346e-07, "logits/chosen": -0.07721833139657974, "logits/rejected": 0.0497472882270813, "logps/chosen": -1.2961310148239136, "logps/rejected": -1.446725845336914, "loss": 1.6488, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2961310148239136, "rewards/margins": 0.1505947858095169, "rewards/rejected": -1.446725845336914, "step": 2135 }, { "epoch": 1.145342030439873, "grad_norm": 7.918047521721078, "learning_rate": 7.769914649982117e-07, "logits/chosen": -0.1346278041601181, "logits/rejected": -0.022326147183775902, "logps/chosen": -1.2685073614120483, "logps/rejected": -1.3912893533706665, "loss": 1.6397, "rewards/accuracies": 0.5, "rewards/chosen": -1.2685073614120483, "rewards/margins": 0.12278201431035995, "rewards/rejected": -1.3912893533706665, "step": 2140 }, { "epoch": 1.1480180632212744, "grad_norm": 7.802641619948372, "learning_rate": 7.756935420438803e-07, "logits/chosen": -0.10416577756404877, "logits/rejected": -0.04111819341778755, "logps/chosen": -1.1541697978973389, "logps/rejected": -1.391998052597046, "loss": 1.5353, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.1541697978973389, "rewards/margins": 0.2378282994031906, "rewards/rejected": -1.391998052597046, "step": 2145 }, { "epoch": 1.1506940960026761, "grad_norm": 5.455496619518153, "learning_rate": 7.743929443095951e-07, "logits/chosen": -0.15382114052772522, "logits/rejected": -0.1144246906042099, "logps/chosen": -1.3197849988937378, "logps/rejected": -1.4354331493377686, "loss": 1.6704, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3197849988937378, "rewards/margins": 0.11564798653125763, "rewards/rejected": -1.4354331493377686, "step": 2150 }, { "epoch": 1.1533701287840776, "grad_norm": 6.590797479707265, "learning_rate": 7.730896844137609e-07, "logits/chosen": -0.07442931830883026, "logits/rejected": -0.023158835247159004, "logps/chosen": -1.30672287940979, "logps/rejected": -1.4486448764801025, "loss": 1.6368, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.30672287940979, "rewards/margins": 0.14192210137844086, "rewards/rejected": -1.4486448764801025, "step": 2155 }, { "epoch": 1.1560461615654791, "grad_norm": 8.71553571729924, "learning_rate": 7.717837750006106e-07, "logits/chosen": -0.16045820713043213, "logits/rejected": -0.06852944940328598, "logps/chosen": -1.2256841659545898, "logps/rejected": -1.4260807037353516, "loss": 1.593, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2256841659545898, "rewards/margins": 0.20039649307727814, "rewards/rejected": -1.4260807037353516, "step": 2160 }, { "epoch": 1.1587221943468808, "grad_norm": 7.010858616243839, "learning_rate": 7.704752287400832e-07, "logits/chosen": -0.13092279434204102, "logits/rejected": 0.020101290196180344, "logps/chosen": -1.2686822414398193, "logps/rejected": -1.4782660007476807, "loss": 1.6078, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2686822414398193, "rewards/margins": 0.2095838040113449, "rewards/rejected": -1.4782660007476807, "step": 2165 }, { "epoch": 1.1613982271282823, "grad_norm": 5.457383139654073, "learning_rate": 7.691640583277004e-07, "logits/chosen": -0.1236150711774826, "logits/rejected": 0.0086661446839571, "logps/chosen": -1.2275776863098145, "logps/rejected": -1.46425199508667, "loss": 1.5892, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2275776863098145, "rewards/margins": 0.23667418956756592, "rewards/rejected": -1.46425199508667, "step": 2170 }, { "epoch": 1.1640742599096838, "grad_norm": 6.136029451743161, "learning_rate": 7.678502764844433e-07, "logits/chosen": -0.12522700428962708, "logits/rejected": 0.016509711742401123, "logps/chosen": -1.3003489971160889, "logps/rejected": -1.386040210723877, "loss": 1.6649, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3003489971160889, "rewards/margins": 0.08569102734327316, "rewards/rejected": -1.386040210723877, "step": 2175 }, { "epoch": 1.1667502926910855, "grad_norm": 6.081976073123297, "learning_rate": 7.665338959566288e-07, "logits/chosen": -0.12515996396541595, "logits/rejected": -0.06724338978528976, "logps/chosen": -1.2342045307159424, "logps/rejected": -1.4075241088867188, "loss": 1.597, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2342045307159424, "rewards/margins": 0.17331981658935547, "rewards/rejected": -1.4075241088867188, "step": 2180 }, { "epoch": 1.169426325472487, "grad_norm": 8.431364114046154, "learning_rate": 7.652149295157868e-07, "logits/chosen": -0.05369647592306137, "logits/rejected": 0.05816897749900818, "logps/chosen": -1.265222430229187, "logps/rejected": -1.3789777755737305, "loss": 1.62, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.265222430229187, "rewards/margins": 0.11375533044338226, "rewards/rejected": -1.3789777755737305, "step": 2185 }, { "epoch": 1.1721023582538885, "grad_norm": 6.915142843947729, "learning_rate": 7.638933899585354e-07, "logits/chosen": 0.02304958924651146, "logits/rejected": 0.06372608244419098, "logps/chosen": -1.2534650564193726, "logps/rejected": -1.3916088342666626, "loss": 1.6261, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2534650564193726, "rewards/margins": 0.13814368844032288, "rewards/rejected": -1.3916088342666626, "step": 2190 }, { "epoch": 1.1747783910352902, "grad_norm": 8.9950441140299, "learning_rate": 7.625692901064573e-07, "logits/chosen": -0.040085647255182266, "logits/rejected": 0.0382084995508194, "logps/chosen": -1.2309796810150146, "logps/rejected": -1.4908417463302612, "loss": 1.5839, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2309796810150146, "rewards/margins": 0.2598620057106018, "rewards/rejected": -1.4908417463302612, "step": 2195 }, { "epoch": 1.1774544238166917, "grad_norm": 7.528852900763409, "learning_rate": 7.61242642805975e-07, "logits/chosen": -0.14972008764743805, "logits/rejected": -0.15985627472400665, "logps/chosen": -1.2626259326934814, "logps/rejected": -1.4132511615753174, "loss": 1.6235, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2626259326934814, "rewards/margins": 0.15062522888183594, "rewards/rejected": -1.4132511615753174, "step": 2200 }, { "epoch": 1.1801304565980932, "grad_norm": 7.013133418167818, "learning_rate": 7.599134609282266e-07, "logits/chosen": -0.14835958182811737, "logits/rejected": 0.0036022409331053495, "logps/chosen": -1.199286699295044, "logps/rejected": -1.3756859302520752, "loss": 1.5803, "rewards/accuracies": 0.625, "rewards/chosen": -1.199286699295044, "rewards/margins": 0.17639943957328796, "rewards/rejected": -1.3756859302520752, "step": 2205 }, { "epoch": 1.182806489379495, "grad_norm": 8.446358358562367, "learning_rate": 7.585817573689402e-07, "logits/chosen": -0.2204298973083496, "logits/rejected": -0.1284269541501999, "logps/chosen": -1.1430498361587524, "logps/rejected": -1.395293951034546, "loss": 1.5234, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1430498361587524, "rewards/margins": 0.2522440552711487, "rewards/rejected": -1.395293951034546, "step": 2210 }, { "epoch": 1.1854825221608964, "grad_norm": 7.8713142227663635, "learning_rate": 7.572475450483098e-07, "logits/chosen": -0.16724872589111328, "logits/rejected": -0.10567456483840942, "logps/chosen": -1.3644531965255737, "logps/rejected": -1.530957818031311, "loss": 1.6859, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3644531965255737, "rewards/margins": 0.1665046066045761, "rewards/rejected": -1.530957818031311, "step": 2215 }, { "epoch": 1.188158554942298, "grad_norm": 6.236928008931055, "learning_rate": 7.559108369108689e-07, "logits/chosen": -0.22434313595294952, "logits/rejected": -0.12110327184200287, "logps/chosen": -1.2065577507019043, "logps/rejected": -1.3525969982147217, "loss": 1.5948, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2065577507019043, "rewards/margins": 0.14603909850120544, "rewards/rejected": -1.3525969982147217, "step": 2220 }, { "epoch": 1.1908345877236997, "grad_norm": 8.20692761506631, "learning_rate": 7.54571645925366e-07, "logits/chosen": -0.18120639026165009, "logits/rejected": -0.02124299295246601, "logps/chosen": -1.2259933948516846, "logps/rejected": -1.461756706237793, "loss": 1.5631, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2259933948516846, "rewards/margins": 0.2357633411884308, "rewards/rejected": -1.461756706237793, "step": 2225 }, { "epoch": 1.1935106205051011, "grad_norm": 13.378601723429425, "learning_rate": 7.532299850846378e-07, "logits/chosen": -0.23142382502555847, "logits/rejected": -0.1267291009426117, "logps/chosen": -1.2854366302490234, "logps/rejected": -1.5767123699188232, "loss": 1.634, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2854366302490234, "rewards/margins": 0.29127568006515503, "rewards/rejected": -1.5767123699188232, "step": 2230 }, { "epoch": 1.1961866532865026, "grad_norm": 9.090806417248851, "learning_rate": 7.518858674054838e-07, "logits/chosen": -0.17949996888637543, "logits/rejected": -0.02718859352171421, "logps/chosen": -1.2322248220443726, "logps/rejected": -1.4807246923446655, "loss": 1.5928, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2322248220443726, "rewards/margins": 0.24849990010261536, "rewards/rejected": -1.4807246923446655, "step": 2235 }, { "epoch": 1.1988626860679044, "grad_norm": 8.873698696072829, "learning_rate": 7.505393059285394e-07, "logits/chosen": -0.18178102374076843, "logits/rejected": -0.06476934254169464, "logps/chosen": -1.2484729290008545, "logps/rejected": -1.45114266872406, "loss": 1.5865, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2484729290008545, "rewards/margins": 0.2026699334383011, "rewards/rejected": -1.45114266872406, "step": 2240 }, { "epoch": 1.2015387188493059, "grad_norm": 9.381368374566106, "learning_rate": 7.491903137181501e-07, "logits/chosen": -0.14490970969200134, "logits/rejected": -0.10534831136465073, "logps/chosen": -1.2206541299819946, "logps/rejected": -1.4163111448287964, "loss": 1.6053, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2206541299819946, "rewards/margins": 0.19565710425376892, "rewards/rejected": -1.4163111448287964, "step": 2245 }, { "epoch": 1.2042147516307076, "grad_norm": 7.520181553218833, "learning_rate": 7.478389038622441e-07, "logits/chosen": -0.10299573838710785, "logits/rejected": -0.0947377160191536, "logps/chosen": -1.1875255107879639, "logps/rejected": -1.4461685419082642, "loss": 1.5358, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1875255107879639, "rewards/margins": 0.2586430609226227, "rewards/rejected": -1.4461685419082642, "step": 2250 }, { "epoch": 1.206890784412109, "grad_norm": 7.006260078296691, "learning_rate": 7.46485089472206e-07, "logits/chosen": -0.15663310885429382, "logits/rejected": -0.06570927798748016, "logps/chosen": -1.3104556798934937, "logps/rejected": -1.3670212030410767, "loss": 1.6854, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3104556798934937, "rewards/margins": 0.05656537413597107, "rewards/rejected": -1.3670212030410767, "step": 2255 }, { "epoch": 1.2095668171935106, "grad_norm": 7.18732291044755, "learning_rate": 7.451288836827487e-07, "logits/chosen": -0.09750930964946747, "logits/rejected": -0.11318697780370712, "logps/chosen": -1.2687238454818726, "logps/rejected": -1.3918168544769287, "loss": 1.6193, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2687238454818726, "rewards/margins": 0.12309300899505615, "rewards/rejected": -1.3918168544769287, "step": 2260 }, { "epoch": 1.2122428499749123, "grad_norm": 8.435818849093998, "learning_rate": 7.437702996517869e-07, "logits/chosen": -0.1856766641139984, "logits/rejected": -0.10571583360433578, "logps/chosen": -1.3002653121948242, "logps/rejected": -1.4226324558258057, "loss": 1.6478, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3002653121948242, "rewards/margins": 0.12236703932285309, "rewards/rejected": -1.4226324558258057, "step": 2265 }, { "epoch": 1.2149188827563138, "grad_norm": 8.438577399522528, "learning_rate": 7.424093505603087e-07, "logits/chosen": -0.275261253118515, "logits/rejected": -0.14463934302330017, "logps/chosen": -1.239253044128418, "logps/rejected": -1.4394651651382446, "loss": 1.5969, "rewards/accuracies": 0.59375, "rewards/chosen": -1.239253044128418, "rewards/margins": 0.20021216571331024, "rewards/rejected": -1.4394651651382446, "step": 2270 }, { "epoch": 1.2175949155377153, "grad_norm": 8.260528732786215, "learning_rate": 7.410460496122482e-07, "logits/chosen": -0.1685386598110199, "logits/rejected": -0.0866832584142685, "logps/chosen": -1.2267197370529175, "logps/rejected": -1.4667956829071045, "loss": 1.5713, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2267197370529175, "rewards/margins": 0.24007587134838104, "rewards/rejected": -1.4667956829071045, "step": 2275 }, { "epoch": 1.220270948319117, "grad_norm": 10.35057514954781, "learning_rate": 7.396804100343572e-07, "logits/chosen": -0.21187300980091095, "logits/rejected": -0.09693863242864609, "logps/chosen": -1.1686277389526367, "logps/rejected": -1.3477180004119873, "loss": 1.566, "rewards/accuracies": 0.53125, "rewards/chosen": -1.1686277389526367, "rewards/margins": 0.17909027636051178, "rewards/rejected": -1.3477180004119873, "step": 2280 }, { "epoch": 1.2229469811005185, "grad_norm": 7.877628525181929, "learning_rate": 7.383124450760768e-07, "logits/chosen": -0.15671268105506897, "logits/rejected": 0.0005567491170950234, "logps/chosen": -1.2920244932174683, "logps/rejected": -1.444446086883545, "loss": 1.6564, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2920244932174683, "rewards/margins": 0.1524215042591095, "rewards/rejected": -1.444446086883545, "step": 2285 }, { "epoch": 1.22562301388192, "grad_norm": 9.659680567006552, "learning_rate": 7.369421680094091e-07, "logits/chosen": -0.2353818118572235, "logits/rejected": -0.11391669511795044, "logps/chosen": -1.156830906867981, "logps/rejected": -1.3382948637008667, "loss": 1.563, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.156830906867981, "rewards/margins": 0.18146398663520813, "rewards/rejected": -1.3382948637008667, "step": 2290 }, { "epoch": 1.2282990466633217, "grad_norm": 7.933801521830124, "learning_rate": 7.355695921287881e-07, "logits/chosen": -0.18124429881572723, "logits/rejected": -0.10757305473089218, "logps/chosen": -1.2433009147644043, "logps/rejected": -1.4033946990966797, "loss": 1.6102, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2433009147644043, "rewards/margins": 0.16009381413459778, "rewards/rejected": -1.4033946990966797, "step": 2295 }, { "epoch": 1.2309750794447232, "grad_norm": 8.86717143935084, "learning_rate": 7.341947307509513e-07, "logits/chosen": -0.15329954028129578, "logits/rejected": -0.05239754170179367, "logps/chosen": -1.260248064994812, "logps/rejected": -1.3629111051559448, "loss": 1.6576, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.260248064994812, "rewards/margins": 0.10266293585300446, "rewards/rejected": -1.3629111051559448, "step": 2300 }, { "epoch": 1.233651112226125, "grad_norm": 7.549256858237088, "learning_rate": 7.328175972148094e-07, "logits/chosen": -0.2044523060321808, "logits/rejected": -0.09235817193984985, "logps/chosen": -1.3756862878799438, "logps/rejected": -1.4864490032196045, "loss": 1.717, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3756862878799438, "rewards/margins": 0.11076265573501587, "rewards/rejected": -1.4864490032196045, "step": 2305 }, { "epoch": 1.2363271450075264, "grad_norm": 11.536576447700092, "learning_rate": 7.314382048813185e-07, "logits/chosen": -0.1483876258134842, "logits/rejected": 0.06599108129739761, "logps/chosen": -1.2939670085906982, "logps/rejected": -1.4761382341384888, "loss": 1.6266, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2939670085906982, "rewards/margins": 0.18217109143733978, "rewards/rejected": -1.4761382341384888, "step": 2310 }, { "epoch": 1.2390031777889279, "grad_norm": 9.100206977273533, "learning_rate": 7.300565671333486e-07, "logits/chosen": -0.14136271178722382, "logits/rejected": -0.005953514017164707, "logps/chosen": -1.271116852760315, "logps/rejected": -1.4551541805267334, "loss": 1.6098, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.271116852760315, "rewards/margins": 0.1840374618768692, "rewards/rejected": -1.4551541805267334, "step": 2315 }, { "epoch": 1.2416792105703296, "grad_norm": 7.434798313930728, "learning_rate": 7.286726973755554e-07, "logits/chosen": -0.06562301516532898, "logits/rejected": -0.05574440956115723, "logps/chosen": -1.278440237045288, "logps/rejected": -1.4780454635620117, "loss": 1.6292, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.278440237045288, "rewards/margins": 0.19960518181324005, "rewards/rejected": -1.4780454635620117, "step": 2320 }, { "epoch": 1.244355243351731, "grad_norm": 10.144557930338843, "learning_rate": 7.272866090342493e-07, "logits/chosen": -0.022041847929358482, "logits/rejected": 0.035709965974092484, "logps/chosen": -1.3029844760894775, "logps/rejected": -1.5329054594039917, "loss": 1.642, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3029844760894775, "rewards/margins": 0.22992126643657684, "rewards/rejected": -1.5329054594039917, "step": 2325 }, { "epoch": 1.2470312761331326, "grad_norm": 7.631646145563696, "learning_rate": 7.258983155572656e-07, "logits/chosen": -0.20035552978515625, "logits/rejected": -0.11055605113506317, "logps/chosen": -1.258976697921753, "logps/rejected": -1.4475109577178955, "loss": 1.6193, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.258976697921753, "rewards/margins": 0.18853434920310974, "rewards/rejected": -1.4475109577178955, "step": 2330 }, { "epoch": 1.2497073089145343, "grad_norm": 8.96296132267378, "learning_rate": 7.245078304138335e-07, "logits/chosen": -0.07240384817123413, "logits/rejected": -0.03183193877339363, "logps/chosen": -1.2547897100448608, "logps/rejected": -1.4930450916290283, "loss": 1.5914, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2547897100448608, "rewards/margins": 0.23825562000274658, "rewards/rejected": -1.4930450916290283, "step": 2335 }, { "epoch": 1.2523833416959358, "grad_norm": 6.011723470561662, "learning_rate": 7.231151670944462e-07, "logits/chosen": -0.22535963356494904, "logits/rejected": -0.07960195094347, "logps/chosen": -1.270557165145874, "logps/rejected": -1.4250514507293701, "loss": 1.6538, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.270557165145874, "rewards/margins": 0.15449431538581848, "rewards/rejected": -1.4250514507293701, "step": 2340 }, { "epoch": 1.2550593744773373, "grad_norm": 7.806406519370909, "learning_rate": 7.217203391107291e-07, "logits/chosen": -0.15927395224571228, "logits/rejected": -0.03005305863916874, "logps/chosen": -1.2619166374206543, "logps/rejected": -1.440375566482544, "loss": 1.6392, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2619166374206543, "rewards/margins": 0.17845895886421204, "rewards/rejected": -1.440375566482544, "step": 2345 }, { "epoch": 1.257735407258739, "grad_norm": 6.854262667921416, "learning_rate": 7.203233599953096e-07, "logits/chosen": -0.16050459444522858, "logits/rejected": -0.05893082544207573, "logps/chosen": -1.2983603477478027, "logps/rejected": -1.4060739278793335, "loss": 1.6549, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2983603477478027, "rewards/margins": 0.10771362483501434, "rewards/rejected": -1.4060739278793335, "step": 2350 }, { "epoch": 1.2604114400401405, "grad_norm": 8.601731390879266, "learning_rate": 7.189242433016852e-07, "logits/chosen": -0.10667651891708374, "logits/rejected": 0.005719064734876156, "logps/chosen": -1.1957563161849976, "logps/rejected": -1.4306471347808838, "loss": 1.5816, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1957563161849976, "rewards/margins": 0.23489081859588623, "rewards/rejected": -1.4306471347808838, "step": 2355 }, { "epoch": 1.263087472821542, "grad_norm": 7.030941860681588, "learning_rate": 7.17523002604092e-07, "logits/chosen": -0.0977768674492836, "logits/rejected": -0.0004381418111734092, "logps/chosen": -1.2323318719863892, "logps/rejected": -1.5626273155212402, "loss": 1.577, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2323318719863892, "rewards/margins": 0.3302956223487854, "rewards/rejected": -1.5626273155212402, "step": 2360 }, { "epoch": 1.2657635056029437, "grad_norm": 5.815185432245015, "learning_rate": 7.161196514973734e-07, "logits/chosen": -0.08281068503856659, "logits/rejected": 0.02186240255832672, "logps/chosen": -1.2621674537658691, "logps/rejected": -1.4554924964904785, "loss": 1.5928, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2621674537658691, "rewards/margins": 0.19332513213157654, "rewards/rejected": -1.4554924964904785, "step": 2365 }, { "epoch": 1.2684395383843452, "grad_norm": 8.139447127312923, "learning_rate": 7.147142035968483e-07, "logits/chosen": -0.07579265534877777, "logits/rejected": 0.01429951936006546, "logps/chosen": -1.2227771282196045, "logps/rejected": -1.4302802085876465, "loss": 1.5711, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2227771282196045, "rewards/margins": 0.20750324428081512, "rewards/rejected": -1.4302802085876465, "step": 2370 }, { "epoch": 1.2711155711657467, "grad_norm": 8.398987175821437, "learning_rate": 7.133066725381781e-07, "logits/chosen": -0.21982577443122864, "logits/rejected": -0.0813729465007782, "logps/chosen": -1.175386667251587, "logps/rejected": -1.2754274606704712, "loss": 1.5729, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.175386667251587, "rewards/margins": 0.10004077106714249, "rewards/rejected": -1.2754274606704712, "step": 2375 }, { "epoch": 1.2737916039471484, "grad_norm": 8.845243873719252, "learning_rate": 7.118970719772354e-07, "logits/chosen": -0.21248643100261688, "logits/rejected": -0.029846271499991417, "logps/chosen": -1.2926652431488037, "logps/rejected": -1.4724124670028687, "loss": 1.6462, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2926652431488037, "rewards/margins": 0.17974717915058136, "rewards/rejected": -1.4724124670028687, "step": 2380 }, { "epoch": 1.27646763672855, "grad_norm": 8.305037106630097, "learning_rate": 7.104854155899711e-07, "logits/chosen": -0.0735519751906395, "logits/rejected": 0.012401674874126911, "logps/chosen": -1.2600033283233643, "logps/rejected": -1.42123544216156, "loss": 1.639, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2600033283233643, "rewards/margins": 0.16123203933238983, "rewards/rejected": -1.42123544216156, "step": 2385 }, { "epoch": 1.2791436695099514, "grad_norm": 6.713897767599404, "learning_rate": 7.090717170722817e-07, "logits/chosen": -0.10170115530490875, "logits/rejected": -0.053350962698459625, "logps/chosen": -1.2383091449737549, "logps/rejected": -1.4422080516815186, "loss": 1.5887, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2383091449737549, "rewards/margins": 0.20389895141124725, "rewards/rejected": -1.4422080516815186, "step": 2390 }, { "epoch": 1.2818197022913531, "grad_norm": 10.19938062205549, "learning_rate": 7.076559901398762e-07, "logits/chosen": -0.259472519159317, "logits/rejected": -0.16848035156726837, "logps/chosen": -1.1729857921600342, "logps/rejected": -1.378514051437378, "loss": 1.5701, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1729857921600342, "rewards/margins": 0.20552822947502136, "rewards/rejected": -1.378514051437378, "step": 2395 }, { "epoch": 1.2844957350727546, "grad_norm": 9.317184801020808, "learning_rate": 7.062382485281436e-07, "logits/chosen": -0.09926240146160126, "logits/rejected": -0.007072200533002615, "logps/chosen": -1.2322889566421509, "logps/rejected": -1.4107961654663086, "loss": 1.6123, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2322889566421509, "rewards/margins": 0.17850707471370697, "rewards/rejected": -1.4107961654663086, "step": 2400 }, { "epoch": 1.2844957350727546, "eval_logits/chosen": 0.1950339674949646, "eval_logits/rejected": 0.27288389205932617, "eval_logps/chosen": -1.3029353618621826, "eval_logps/rejected": -1.4740725755691528, "eval_loss": 1.652165174484253, "eval_rewards/accuracies": 0.5675074458122253, "eval_rewards/chosen": -1.3029353618621826, "eval_rewards/margins": 0.17113718390464783, "eval_rewards/rejected": -1.4740725755691528, "eval_runtime": 40.6885, "eval_samples_per_second": 33.056, "eval_steps_per_second": 8.282, "step": 2400 }, { "epoch": 1.287171767854156, "grad_norm": 7.870523096019233, "learning_rate": 7.048185059920193e-07, "logits/chosen": -0.11690235137939453, "logits/rejected": 0.014140060171484947, "logps/chosen": -1.2653751373291016, "logps/rejected": -1.4744374752044678, "loss": 1.6085, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2653751373291016, "rewards/margins": 0.20906221866607666, "rewards/rejected": -1.4744374752044678, "step": 2405 }, { "epoch": 1.2898478006355578, "grad_norm": 5.774238616199731, "learning_rate": 7.033967763058516e-07, "logits/chosen": -0.25922006368637085, "logits/rejected": -0.08819141983985901, "logps/chosen": -1.236149549484253, "logps/rejected": -1.3286923170089722, "loss": 1.6408, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.236149549484253, "rewards/margins": 0.09254281967878342, "rewards/rejected": -1.3286923170089722, "step": 2410 }, { "epoch": 1.2925238334169593, "grad_norm": 6.588861507194845, "learning_rate": 7.019730732632681e-07, "logits/chosen": -0.08580441772937775, "logits/rejected": -0.0363566055893898, "logps/chosen": -1.1683250665664673, "logps/rejected": -1.4436743259429932, "loss": 1.5217, "rewards/accuracies": 0.625, "rewards/chosen": -1.1683250665664673, "rewards/margins": 0.2753491699695587, "rewards/rejected": -1.4436743259429932, "step": 2415 }, { "epoch": 1.2951998661983608, "grad_norm": 8.127051455874462, "learning_rate": 7.005474106770418e-07, "logits/chosen": -0.2234508991241455, "logits/rejected": -0.12078411877155304, "logps/chosen": -1.2782351970672607, "logps/rejected": -1.5385297536849976, "loss": 1.5969, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2782351970672607, "rewards/margins": 0.26029473543167114, "rewards/rejected": -1.5385297536849976, "step": 2420 }, { "epoch": 1.2978758989797625, "grad_norm": 8.007284967592312, "learning_rate": 6.991198023789577e-07, "logits/chosen": -0.11372745037078857, "logits/rejected": -0.048462964594364166, "logps/chosen": -1.2203960418701172, "logps/rejected": -1.3983186483383179, "loss": 1.5918, "rewards/accuracies": 0.625, "rewards/chosen": -1.2203960418701172, "rewards/margins": 0.1779225468635559, "rewards/rejected": -1.3983186483383179, "step": 2425 }, { "epoch": 1.300551931761164, "grad_norm": 8.464943925768683, "learning_rate": 6.976902622196776e-07, "logits/chosen": -0.11578784137964249, "logits/rejected": -0.08799702674150467, "logps/chosen": -1.3550939559936523, "logps/rejected": -1.5142543315887451, "loss": 1.6842, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3550939559936523, "rewards/margins": 0.1591602861881256, "rewards/rejected": -1.5142543315887451, "step": 2430 }, { "epoch": 1.3032279645425655, "grad_norm": 4.964557907334137, "learning_rate": 6.962588040686064e-07, "logits/chosen": -0.0680847316980362, "logits/rejected": 0.029392996802926064, "logps/chosen": -1.229901909828186, "logps/rejected": -1.377708077430725, "loss": 1.6048, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.229901909828186, "rewards/margins": 0.14780613780021667, "rewards/rejected": -1.377708077430725, "step": 2435 }, { "epoch": 1.3059039973239672, "grad_norm": 9.173356236493861, "learning_rate": 6.948254418137573e-07, "logits/chosen": -0.22466211020946503, "logits/rejected": -0.14043515920639038, "logps/chosen": -1.208801507949829, "logps/rejected": -1.401143193244934, "loss": 1.5828, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.208801507949829, "rewards/margins": 0.19234168529510498, "rewards/rejected": -1.401143193244934, "step": 2440 }, { "epoch": 1.3085800301053687, "grad_norm": 8.615344938569784, "learning_rate": 6.933901893616174e-07, "logits/chosen": -0.16824540495872498, "logits/rejected": -0.04846107214689255, "logps/chosen": -1.2559640407562256, "logps/rejected": -1.3949588537216187, "loss": 1.6227, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2559640407562256, "rewards/margins": 0.13899469375610352, "rewards/rejected": -1.3949588537216187, "step": 2445 }, { "epoch": 1.3112560628867704, "grad_norm": 6.534082551436406, "learning_rate": 6.919530606370121e-07, "logits/chosen": -0.16765077412128448, "logits/rejected": -0.043173328042030334, "logps/chosen": -1.2167965173721313, "logps/rejected": -1.4837311506271362, "loss": 1.5455, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2167965173721313, "rewards/margins": 0.2669346332550049, "rewards/rejected": -1.4837311506271362, "step": 2450 }, { "epoch": 1.313932095668172, "grad_norm": 5.162323626655467, "learning_rate": 6.905140695829706e-07, "logits/chosen": -0.18577834963798523, "logits/rejected": -0.01245189644396305, "logps/chosen": -1.3045991659164429, "logps/rejected": -1.4210084676742554, "loss": 1.6483, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3045991659164429, "rewards/margins": 0.11640922725200653, "rewards/rejected": -1.4210084676742554, "step": 2455 }, { "epoch": 1.3166081284495736, "grad_norm": 8.779305794595118, "learning_rate": 6.890732301605904e-07, "logits/chosen": -0.1385190486907959, "logits/rejected": -0.06480330973863602, "logps/chosen": -1.2952321767807007, "logps/rejected": -1.3900924921035767, "loss": 1.6514, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2952321767807007, "rewards/margins": 0.09486036002635956, "rewards/rejected": -1.3900924921035767, "step": 2460 }, { "epoch": 1.3192841612309751, "grad_norm": 6.879124186375867, "learning_rate": 6.876305563489021e-07, "logits/chosen": -0.1268390715122223, "logits/rejected": -0.07661868631839752, "logps/chosen": -1.2554209232330322, "logps/rejected": -1.4585278034210205, "loss": 1.6044, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2554209232330322, "rewards/margins": 0.20310668647289276, "rewards/rejected": -1.4585278034210205, "step": 2465 }, { "epoch": 1.3219601940123766, "grad_norm": 11.279106206347876, "learning_rate": 6.861860621447331e-07, "logits/chosen": -0.2476646453142166, "logits/rejected": -0.14627087116241455, "logps/chosen": -1.2362347841262817, "logps/rejected": -1.330241084098816, "loss": 1.6221, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2362347841262817, "rewards/margins": 0.09400641173124313, "rewards/rejected": -1.330241084098816, "step": 2470 }, { "epoch": 1.3246362267937783, "grad_norm": 6.601011703249647, "learning_rate": 6.847397615625725e-07, "logits/chosen": -0.17375004291534424, "logits/rejected": -0.11787985265254974, "logps/chosen": -1.2520265579223633, "logps/rejected": -1.459517240524292, "loss": 1.6086, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2520265579223633, "rewards/margins": 0.20749051868915558, "rewards/rejected": -1.459517240524292, "step": 2475 }, { "epoch": 1.3273122595751798, "grad_norm": 6.32496552850165, "learning_rate": 6.83291668634435e-07, "logits/chosen": -0.2697427272796631, "logits/rejected": -0.1335819959640503, "logps/chosen": -1.2836506366729736, "logps/rejected": -1.494109869003296, "loss": 1.6118, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2836506366729736, "rewards/margins": 0.21045920252799988, "rewards/rejected": -1.494109869003296, "step": 2480 }, { "epoch": 1.3299882923565813, "grad_norm": 7.993260328336925, "learning_rate": 6.818417974097246e-07, "logits/chosen": -0.11526402086019516, "logits/rejected": 0.03870851546525955, "logps/chosen": -1.2829654216766357, "logps/rejected": -1.495354413986206, "loss": 1.6461, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2829654216766357, "rewards/margins": 0.2123890370130539, "rewards/rejected": -1.495354413986206, "step": 2485 }, { "epoch": 1.332664325137983, "grad_norm": 7.319826411914482, "learning_rate": 6.803901619550981e-07, "logits/chosen": -0.210830956697464, "logits/rejected": -0.1711214929819107, "logps/chosen": -1.2724640369415283, "logps/rejected": -1.5099023580551147, "loss": 1.6111, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2724640369415283, "rewards/margins": 0.23743829131126404, "rewards/rejected": -1.5099023580551147, "step": 2490 }, { "epoch": 1.3353403579193845, "grad_norm": 7.088857844116769, "learning_rate": 6.789367763543292e-07, "logits/chosen": -0.12234888225793839, "logits/rejected": -0.12407497316598892, "logps/chosen": -1.2546168565750122, "logps/rejected": -1.4399218559265137, "loss": 1.6088, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2546168565750122, "rewards/margins": 0.18530511856079102, "rewards/rejected": -1.4399218559265137, "step": 2495 }, { "epoch": 1.338016390700786, "grad_norm": 7.917698914494699, "learning_rate": 6.774816547081714e-07, "logits/chosen": -0.10988526046276093, "logits/rejected": 0.012320229783654213, "logps/chosen": -1.2048064470291138, "logps/rejected": -1.39044189453125, "loss": 1.5679, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2048064470291138, "rewards/margins": 0.18563534319400787, "rewards/rejected": -1.39044189453125, "step": 2500 }, { "epoch": 1.3406924234821878, "grad_norm": 7.108609404496919, "learning_rate": 6.760248111342211e-07, "logits/chosen": -0.12231286615133286, "logits/rejected": 0.017514532431960106, "logps/chosen": -1.2140834331512451, "logps/rejected": -1.38956618309021, "loss": 1.575, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2140834331512451, "rewards/margins": 0.1754826456308365, "rewards/rejected": -1.38956618309021, "step": 2505 }, { "epoch": 1.3433684562635893, "grad_norm": 10.213421753778464, "learning_rate": 6.745662597667813e-07, "logits/chosen": -0.19724589586257935, "logits/rejected": -0.09277534484863281, "logps/chosen": -1.2082345485687256, "logps/rejected": -1.4114068746566772, "loss": 1.5714, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2082345485687256, "rewards/margins": 0.2031722366809845, "rewards/rejected": -1.4114068746566772, "step": 2510 }, { "epoch": 1.3460444890449907, "grad_norm": 6.255396425592991, "learning_rate": 6.731060147567236e-07, "logits/chosen": -0.12620656192302704, "logits/rejected": -0.03647005930542946, "logps/chosen": -1.291902780532837, "logps/rejected": -1.401580572128296, "loss": 1.6676, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.291902780532837, "rewards/margins": 0.10967768728733063, "rewards/rejected": -1.401580572128296, "step": 2515 }, { "epoch": 1.3487205218263925, "grad_norm": 5.758726271944098, "learning_rate": 6.716440902713515e-07, "logits/chosen": -0.22279579937458038, "logits/rejected": -0.16483859717845917, "logps/chosen": -1.271669626235962, "logps/rejected": -1.426734209060669, "loss": 1.6181, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.271669626235962, "rewards/margins": 0.1550646275281906, "rewards/rejected": -1.426734209060669, "step": 2520 }, { "epoch": 1.351396554607794, "grad_norm": 10.562665707587549, "learning_rate": 6.701805004942627e-07, "logits/chosen": -0.16978059709072113, "logits/rejected": -0.11235268414020538, "logps/chosen": -1.2969720363616943, "logps/rejected": -1.4508230686187744, "loss": 1.635, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2969720363616943, "rewards/margins": 0.1538509726524353, "rewards/rejected": -1.4508230686187744, "step": 2525 }, { "epoch": 1.3540725873891954, "grad_norm": 8.606204546975732, "learning_rate": 6.687152596252119e-07, "logits/chosen": -0.20366983115673065, "logits/rejected": -0.1693367063999176, "logps/chosen": -1.2421363592147827, "logps/rejected": -1.4101701974868774, "loss": 1.6156, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2421363592147827, "rewards/margins": 0.16803371906280518, "rewards/rejected": -1.4101701974868774, "step": 2530 }, { "epoch": 1.3567486201705972, "grad_norm": 6.1125997832319525, "learning_rate": 6.672483818799722e-07, "logits/chosen": -0.25032004714012146, "logits/rejected": -0.14668934047222137, "logps/chosen": -1.2616934776306152, "logps/rejected": -1.4530704021453857, "loss": 1.6074, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2616934776306152, "rewards/margins": 0.19137685000896454, "rewards/rejected": -1.4530704021453857, "step": 2535 }, { "epoch": 1.3594246529519987, "grad_norm": 9.518342328508894, "learning_rate": 6.657798814901978e-07, "logits/chosen": -0.16530966758728027, "logits/rejected": -0.037598997354507446, "logps/chosen": -1.3490521907806396, "logps/rejected": -1.4344947338104248, "loss": 1.6997, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3490521907806396, "rewards/margins": 0.0854424387216568, "rewards/rejected": -1.4344947338104248, "step": 2540 }, { "epoch": 1.3621006857334002, "grad_norm": 6.611910319793207, "learning_rate": 6.643097727032863e-07, "logits/chosen": -0.12749221920967102, "logits/rejected": 0.005467116832733154, "logps/chosen": -1.2553026676177979, "logps/rejected": -1.4536714553833008, "loss": 1.6155, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2553026676177979, "rewards/margins": 0.1983688771724701, "rewards/rejected": -1.4536714553833008, "step": 2545 }, { "epoch": 1.3647767185148019, "grad_norm": 6.999183872004218, "learning_rate": 6.628380697822392e-07, "logits/chosen": -0.15842144191265106, "logits/rejected": -0.03417439013719559, "logps/chosen": -1.2414013147354126, "logps/rejected": -1.3254547119140625, "loss": 1.6356, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2414013147354126, "rewards/margins": 0.08405350893735886, "rewards/rejected": -1.3254547119140625, "step": 2550 }, { "epoch": 1.3674527512962034, "grad_norm": 9.23276822558898, "learning_rate": 6.61364787005525e-07, "logits/chosen": -0.1209961324930191, "logits/rejected": -0.06522391736507416, "logps/chosen": -1.1821562051773071, "logps/rejected": -1.4737653732299805, "loss": 1.5397, "rewards/accuracies": 0.625, "rewards/chosen": -1.1821562051773071, "rewards/margins": 0.2916093170642853, "rewards/rejected": -1.4737653732299805, "step": 2555 }, { "epoch": 1.3701287840776049, "grad_norm": 7.419043651798917, "learning_rate": 6.598899386669395e-07, "logits/chosen": -0.11700377613306046, "logits/rejected": -0.018622055649757385, "logps/chosen": -1.2295806407928467, "logps/rejected": -1.4211266040802002, "loss": 1.5895, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2295806407928467, "rewards/margins": 0.19154588878154755, "rewards/rejected": -1.4211266040802002, "step": 2560 }, { "epoch": 1.3728048168590066, "grad_norm": 8.739827529012569, "learning_rate": 6.584135390754679e-07, "logits/chosen": -0.13102322816848755, "logits/rejected": -0.0399283766746521, "logps/chosen": -1.2161195278167725, "logps/rejected": -1.4188120365142822, "loss": 1.5784, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2161195278167725, "rewards/margins": 0.2026924192905426, "rewards/rejected": -1.4188120365142822, "step": 2565 }, { "epoch": 1.375480849640408, "grad_norm": 5.28190719308519, "learning_rate": 6.569356025551454e-07, "logits/chosen": -0.07822327315807343, "logits/rejected": -0.028475111350417137, "logps/chosen": -1.2194154262542725, "logps/rejected": -1.4345922470092773, "loss": 1.5667, "rewards/accuracies": 0.625, "rewards/chosen": -1.2194154262542725, "rewards/margins": 0.21517670154571533, "rewards/rejected": -1.4345922470092773, "step": 2570 }, { "epoch": 1.3781568824218096, "grad_norm": 9.348428251126803, "learning_rate": 6.554561434449186e-07, "logits/chosen": -0.222854346036911, "logits/rejected": -0.1074608564376831, "logps/chosen": -1.2142174243927002, "logps/rejected": -1.444159984588623, "loss": 1.5712, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2142174243927002, "rewards/margins": 0.22994264960289001, "rewards/rejected": -1.444159984588623, "step": 2575 }, { "epoch": 1.3808329152032113, "grad_norm": 10.517077556395618, "learning_rate": 6.539751760985063e-07, "logits/chosen": -0.12920764088630676, "logits/rejected": -0.07156732678413391, "logps/chosen": -1.314318299293518, "logps/rejected": -1.4463088512420654, "loss": 1.6786, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.314318299293518, "rewards/margins": 0.13199035823345184, "rewards/rejected": -1.4463088512420654, "step": 2580 }, { "epoch": 1.3835089479846128, "grad_norm": 7.614676517974458, "learning_rate": 6.524927148842602e-07, "logits/chosen": -0.053183816373348236, "logits/rejected": 0.08111951500177383, "logps/chosen": -1.1608166694641113, "logps/rejected": -1.417043685913086, "loss": 1.544, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1608166694641113, "rewards/margins": 0.2562270164489746, "rewards/rejected": -1.417043685913086, "step": 2585 }, { "epoch": 1.3861849807660143, "grad_norm": 8.537552692374557, "learning_rate": 6.510087741850254e-07, "logits/chosen": -0.1578807532787323, "logits/rejected": -0.03583088144659996, "logps/chosen": -1.2086454629898071, "logps/rejected": -1.418277382850647, "loss": 1.5913, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2086454629898071, "rewards/margins": 0.20963187515735626, "rewards/rejected": -1.418277382850647, "step": 2590 }, { "epoch": 1.388861013547416, "grad_norm": 7.919291466775935, "learning_rate": 6.495233683980012e-07, "logits/chosen": -0.11748051643371582, "logits/rejected": -0.0915294662117958, "logps/chosen": -1.2231186628341675, "logps/rejected": -1.4085519313812256, "loss": 1.5786, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2231186628341675, "rewards/margins": 0.18543313443660736, "rewards/rejected": -1.4085519313812256, "step": 2595 }, { "epoch": 1.3915370463288175, "grad_norm": 9.665461917485846, "learning_rate": 6.480365119346011e-07, "logits/chosen": -0.04559020325541496, "logits/rejected": 0.05628150701522827, "logps/chosen": -1.2490603923797607, "logps/rejected": -1.374648094177246, "loss": 1.624, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2490603923797607, "rewards/margins": 0.12558765709400177, "rewards/rejected": -1.374648094177246, "step": 2600 }, { "epoch": 1.394213079110219, "grad_norm": 8.237334325510016, "learning_rate": 6.465482192203129e-07, "logits/chosen": -0.06613899767398834, "logits/rejected": -0.03729189559817314, "logps/chosen": -1.2390214204788208, "logps/rejected": -1.3908802270889282, "loss": 1.6125, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2390214204788208, "rewards/margins": 0.15185879170894623, "rewards/rejected": -1.3908802270889282, "step": 2605 }, { "epoch": 1.3968891118916207, "grad_norm": 9.157276767854839, "learning_rate": 6.45058504694559e-07, "logits/chosen": -0.04596097022294998, "logits/rejected": 0.015534261241555214, "logps/chosen": -1.2925312519073486, "logps/rejected": -1.402549386024475, "loss": 1.6832, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2925312519073486, "rewards/margins": 0.11001831293106079, "rewards/rejected": -1.402549386024475, "step": 2610 }, { "epoch": 1.3995651446730222, "grad_norm": 11.201303785821795, "learning_rate": 6.435673828105564e-07, "logits/chosen": -0.13810259103775024, "logits/rejected": -0.02151239663362503, "logps/chosen": -1.2091667652130127, "logps/rejected": -1.4765537977218628, "loss": 1.5709, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2091667652130127, "rewards/margins": 0.2673870921134949, "rewards/rejected": -1.4765537977218628, "step": 2615 }, { "epoch": 1.402241177454424, "grad_norm": 9.141856468133408, "learning_rate": 6.420748680351763e-07, "logits/chosen": -0.14668779075145721, "logits/rejected": -0.15653502941131592, "logps/chosen": -1.327497124671936, "logps/rejected": -1.4073745012283325, "loss": 1.7246, "rewards/accuracies": 0.44999998807907104, "rewards/chosen": -1.327497124671936, "rewards/margins": 0.07987751811742783, "rewards/rejected": -1.4073745012283325, "step": 2620 }, { "epoch": 1.4049172102358254, "grad_norm": 8.592315234668606, "learning_rate": 6.405809748488032e-07, "logits/chosen": -0.09723938256502151, "logits/rejected": 0.02080477401614189, "logps/chosen": -1.2495981454849243, "logps/rejected": -1.438523292541504, "loss": 1.5908, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2495981454849243, "rewards/margins": 0.18892532587051392, "rewards/rejected": -1.438523292541504, "step": 2625 }, { "epoch": 1.4075932430172269, "grad_norm": 8.364445063548647, "learning_rate": 6.390857177451956e-07, "logits/chosen": -0.2485821545124054, "logits/rejected": -0.07332058995962143, "logps/chosen": -1.3173432350158691, "logps/rejected": -1.4042927026748657, "loss": 1.6781, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3173432350158691, "rewards/margins": 0.08694945275783539, "rewards/rejected": -1.4042927026748657, "step": 2630 }, { "epoch": 1.4102692757986286, "grad_norm": 10.206806565461111, "learning_rate": 6.375891112313445e-07, "logits/chosen": -0.12817701697349548, "logits/rejected": -0.06272152066230774, "logps/chosen": -1.2560198307037354, "logps/rejected": -1.421544075012207, "loss": 1.5929, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2560198307037354, "rewards/margins": 0.1655244678258896, "rewards/rejected": -1.421544075012207, "step": 2635 }, { "epoch": 1.41294530858003, "grad_norm": 6.646996513211527, "learning_rate": 6.360911698273326e-07, "logits/chosen": -0.08035455644130707, "logits/rejected": -0.02586877904832363, "logps/chosen": -1.3121914863586426, "logps/rejected": -1.4780056476593018, "loss": 1.6552, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3121914863586426, "rewards/margins": 0.1658143550157547, "rewards/rejected": -1.4780056476593018, "step": 2640 }, { "epoch": 1.4156213413614318, "grad_norm": 8.205700226774075, "learning_rate": 6.345919080661944e-07, "logits/chosen": -0.1574331670999527, "logits/rejected": -0.09802021086215973, "logps/chosen": -1.2450119256973267, "logps/rejected": -1.4981917142868042, "loss": 1.5773, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2450119256973267, "rewards/margins": 0.2531798481941223, "rewards/rejected": -1.4981917142868042, "step": 2645 }, { "epoch": 1.4182973741428333, "grad_norm": 6.74614958982811, "learning_rate": 6.330913404937737e-07, "logits/chosen": -0.20685835182666779, "logits/rejected": -0.08519494533538818, "logps/chosen": -1.2531399726867676, "logps/rejected": -1.6164621114730835, "loss": 1.6031, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2531399726867676, "rewards/margins": 0.3633221983909607, "rewards/rejected": -1.6164621114730835, "step": 2650 }, { "epoch": 1.4209734069242348, "grad_norm": 8.830012908631538, "learning_rate": 6.315894816685838e-07, "logits/chosen": -0.11997779458761215, "logits/rejected": 0.0006568536045961082, "logps/chosen": -1.158200740814209, "logps/rejected": -1.3409008979797363, "loss": 1.545, "rewards/accuracies": 0.5625, "rewards/chosen": -1.158200740814209, "rewards/margins": 0.1827000081539154, "rewards/rejected": -1.3409008979797363, "step": 2655 }, { "epoch": 1.4236494397056365, "grad_norm": 8.649731098796604, "learning_rate": 6.300863461616657e-07, "logits/chosen": -0.07503405958414078, "logits/rejected": -0.04890364408493042, "logps/chosen": -1.1313199996948242, "logps/rejected": -1.3792779445648193, "loss": 1.5135, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1313199996948242, "rewards/margins": 0.24795803427696228, "rewards/rejected": -1.3792779445648193, "step": 2660 }, { "epoch": 1.426325472487038, "grad_norm": 7.4933448100013695, "learning_rate": 6.285819485564465e-07, "logits/chosen": -0.18574872612953186, "logits/rejected": -0.09579996019601822, "logps/chosen": -1.2786818742752075, "logps/rejected": -1.4751293659210205, "loss": 1.6158, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2786818742752075, "rewards/margins": 0.1964474618434906, "rewards/rejected": -1.4751293659210205, "step": 2665 }, { "epoch": 1.4290015052684395, "grad_norm": 8.864208288115423, "learning_rate": 6.270763034485986e-07, "logits/chosen": -0.09423649311065674, "logits/rejected": -0.02666972018778324, "logps/chosen": -1.3697813749313354, "logps/rejected": -1.4894285202026367, "loss": 1.7199, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.3697813749313354, "rewards/margins": 0.1196470707654953, "rewards/rejected": -1.4894285202026367, "step": 2670 }, { "epoch": 1.4316775380498412, "grad_norm": 9.997330006904571, "learning_rate": 6.255694254458972e-07, "logits/chosen": -0.13650649785995483, "logits/rejected": -0.005107997450977564, "logps/chosen": -1.2864766120910645, "logps/rejected": -1.4099823236465454, "loss": 1.6435, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2864766120910645, "rewards/margins": 0.12350551784038544, "rewards/rejected": -1.4099823236465454, "step": 2675 }, { "epoch": 1.4343535708312427, "grad_norm": 9.257757138585266, "learning_rate": 6.240613291680795e-07, "logits/chosen": -0.19174785912036896, "logits/rejected": -0.057482410222291946, "logps/chosen": -1.2712281942367554, "logps/rejected": -1.4052618741989136, "loss": 1.642, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2712281942367554, "rewards/margins": 0.13403376936912537, "rewards/rejected": -1.4052618741989136, "step": 2680 }, { "epoch": 1.4370296036126442, "grad_norm": 9.12705530288166, "learning_rate": 6.225520292467021e-07, "logits/chosen": -0.13390594720840454, "logits/rejected": 0.035317469388246536, "logps/chosen": -1.26984703540802, "logps/rejected": -1.3742210865020752, "loss": 1.6372, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.26984703540802, "rewards/margins": 0.10437413305044174, "rewards/rejected": -1.3742210865020752, "step": 2685 }, { "epoch": 1.439705636394046, "grad_norm": 10.132403967111959, "learning_rate": 6.210415403249993e-07, "logits/chosen": -0.27369385957717896, "logits/rejected": -0.082986980676651, "logps/chosen": -1.2638477087020874, "logps/rejected": -1.5354796648025513, "loss": 1.6034, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2638477087020874, "rewards/margins": 0.27163180708885193, "rewards/rejected": -1.5354796648025513, "step": 2690 }, { "epoch": 1.4423816691754474, "grad_norm": 6.391142485683344, "learning_rate": 6.195298770577415e-07, "logits/chosen": -0.14904870092868805, "logits/rejected": -0.13162747025489807, "logps/chosen": -1.2408156394958496, "logps/rejected": -1.454119324684143, "loss": 1.5791, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2408156394958496, "rewards/margins": 0.21330364048480988, "rewards/rejected": -1.454119324684143, "step": 2695 }, { "epoch": 1.445057701956849, "grad_norm": 7.727114200402468, "learning_rate": 6.180170541110923e-07, "logits/chosen": -0.18305456638336182, "logits/rejected": -0.02566959336400032, "logps/chosen": -1.2899725437164307, "logps/rejected": -1.4568368196487427, "loss": 1.6393, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2899725437164307, "rewards/margins": 0.16686417162418365, "rewards/rejected": -1.4568368196487427, "step": 2700 }, { "epoch": 1.4477337347382506, "grad_norm": 5.701958769249109, "learning_rate": 6.165030861624663e-07, "logits/chosen": -0.19134236872196198, "logits/rejected": -0.025602400302886963, "logps/chosen": -1.1669811010360718, "logps/rejected": -1.4505869150161743, "loss": 1.5034, "rewards/accuracies": 0.625, "rewards/chosen": -1.1669811010360718, "rewards/margins": 0.2836058437824249, "rewards/rejected": -1.4505869150161743, "step": 2705 }, { "epoch": 1.4504097675196521, "grad_norm": 7.643404876619234, "learning_rate": 6.149879879003876e-07, "logits/chosen": -0.10536251217126846, "logits/rejected": -0.08227679133415222, "logps/chosen": -1.2573299407958984, "logps/rejected": -1.473145604133606, "loss": 1.5871, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2573299407958984, "rewards/margins": 0.21581566333770752, "rewards/rejected": -1.473145604133606, "step": 2710 }, { "epoch": 1.4530858003010536, "grad_norm": 7.684216778403979, "learning_rate": 6.13471774024346e-07, "logits/chosen": -0.23534438014030457, "logits/rejected": -0.16144892573356628, "logps/chosen": -1.1998146772384644, "logps/rejected": -1.3855111598968506, "loss": 1.5734, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.1998146772384644, "rewards/margins": 0.18569636344909668, "rewards/rejected": -1.3855111598968506, "step": 2715 }, { "epoch": 1.4557618330824553, "grad_norm": 7.474698672929521, "learning_rate": 6.119544592446551e-07, "logits/chosen": -0.18152351677417755, "logits/rejected": -0.08897839486598969, "logps/chosen": -1.2277088165283203, "logps/rejected": -1.3388454914093018, "loss": 1.6011, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2277088165283203, "rewards/margins": 0.11113659292459488, "rewards/rejected": -1.3388454914093018, "step": 2720 }, { "epoch": 1.4584378658638568, "grad_norm": 9.308831518397065, "learning_rate": 6.104360582823096e-07, "logits/chosen": -0.1577589064836502, "logits/rejected": -0.06443767249584198, "logps/chosen": -1.2389967441558838, "logps/rejected": -1.4576349258422852, "loss": 1.6043, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2389967441558838, "rewards/margins": 0.21863818168640137, "rewards/rejected": -1.4576349258422852, "step": 2725 }, { "epoch": 1.4611138986452583, "grad_norm": 8.364378865528284, "learning_rate": 6.089165858688423e-07, "logits/chosen": -0.19063672423362732, "logits/rejected": -0.055591851472854614, "logps/chosen": -1.239959955215454, "logps/rejected": -1.4646373987197876, "loss": 1.6223, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.239959955215454, "rewards/margins": 0.22467747330665588, "rewards/rejected": -1.4646373987197876, "step": 2730 }, { "epoch": 1.46378993142666, "grad_norm": 7.2502968807995165, "learning_rate": 6.073960567461811e-07, "logits/chosen": -0.15855631232261658, "logits/rejected": 0.005321502685546875, "logps/chosen": -1.154059648513794, "logps/rejected": -1.4797316789627075, "loss": 1.5238, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.154059648513794, "rewards/margins": 0.32567209005355835, "rewards/rejected": -1.4797316789627075, "step": 2735 }, { "epoch": 1.4664659642080615, "grad_norm": 9.284400020312043, "learning_rate": 6.058744856665065e-07, "logits/chosen": -0.1830758899450302, "logits/rejected": -0.09102828800678253, "logps/chosen": -1.192286491394043, "logps/rejected": -1.4547383785247803, "loss": 1.5673, "rewards/accuracies": 0.59375, "rewards/chosen": -1.192286491394043, "rewards/margins": 0.2624519467353821, "rewards/rejected": -1.4547383785247803, "step": 2740 }, { "epoch": 1.469141996989463, "grad_norm": 7.236813918750393, "learning_rate": 6.043518873921074e-07, "logits/chosen": -0.1663316786289215, "logits/rejected": -0.07601561397314072, "logps/chosen": -1.2092053890228271, "logps/rejected": -1.3376567363739014, "loss": 1.5912, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2092053890228271, "rewards/margins": 0.12845143675804138, "rewards/rejected": -1.3376567363739014, "step": 2745 }, { "epoch": 1.4718180297708647, "grad_norm": 7.953164084744921, "learning_rate": 6.028282766952393e-07, "logits/chosen": -0.13207189738750458, "logits/rejected": -0.05418176203966141, "logps/chosen": -1.2845780849456787, "logps/rejected": -1.52390718460083, "loss": 1.6049, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2845780849456787, "rewards/margins": 0.2393290102481842, "rewards/rejected": -1.52390718460083, "step": 2750 }, { "epoch": 1.4744940625522662, "grad_norm": 9.956779025794944, "learning_rate": 6.013036683579798e-07, "logits/chosen": -0.08539600670337677, "logits/rejected": 0.03946538269519806, "logps/chosen": -1.2316306829452515, "logps/rejected": -1.3691304922103882, "loss": 1.5971, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2316306829452515, "rewards/margins": 0.13749974966049194, "rewards/rejected": -1.3691304922103882, "step": 2755 }, { "epoch": 1.4771700953336677, "grad_norm": 7.9294848838307015, "learning_rate": 5.997780771720854e-07, "logits/chosen": -0.22258487343788147, "logits/rejected": -0.08026249706745148, "logps/chosen": -1.2677894830703735, "logps/rejected": -1.5182039737701416, "loss": 1.6203, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2677894830703735, "rewards/margins": 0.25041455030441284, "rewards/rejected": -1.5182039737701416, "step": 2760 }, { "epoch": 1.4798461281150694, "grad_norm": 8.259625818318181, "learning_rate": 5.982515179388486e-07, "logits/chosen": -0.12650857865810394, "logits/rejected": -0.011118963360786438, "logps/chosen": -1.188811182975769, "logps/rejected": -1.3890987634658813, "loss": 1.5566, "rewards/accuracies": 0.625, "rewards/chosen": -1.188811182975769, "rewards/margins": 0.20028741657733917, "rewards/rejected": -1.3890987634658813, "step": 2765 }, { "epoch": 1.482522160896471, "grad_norm": 7.953137807396655, "learning_rate": 5.967240054689541e-07, "logits/chosen": -0.16903971135616302, "logits/rejected": -0.11257338523864746, "logps/chosen": -1.218370795249939, "logps/rejected": -1.2844295501708984, "loss": 1.6292, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.218370795249939, "rewards/margins": 0.06605866551399231, "rewards/rejected": -1.2844295501708984, "step": 2770 }, { "epoch": 1.4851981936778724, "grad_norm": 5.5455741309036615, "learning_rate": 5.951955545823342e-07, "logits/chosen": -0.14832541346549988, "logits/rejected": -0.13577041029930115, "logps/chosen": -1.1799458265304565, "logps/rejected": -1.3889585733413696, "loss": 1.5523, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1799458265304565, "rewards/margins": 0.20901262760162354, "rewards/rejected": -1.3889585733413696, "step": 2775 }, { "epoch": 1.4878742264592741, "grad_norm": 5.398542683932957, "learning_rate": 5.936661801080263e-07, "logits/chosen": -0.13961201906204224, "logits/rejected": -0.0569029338657856, "logps/chosen": -1.3484827280044556, "logps/rejected": -1.4868760108947754, "loss": 1.6867, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.3484827280044556, "rewards/margins": 0.1383933126926422, "rewards/rejected": -1.4868760108947754, "step": 2780 }, { "epoch": 1.4905502592406756, "grad_norm": 7.595821185890207, "learning_rate": 5.92135896884028e-07, "logits/chosen": -0.1922890841960907, "logits/rejected": -0.0873006209731102, "logps/chosen": -1.3084876537322998, "logps/rejected": -1.4598156213760376, "loss": 1.6707, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3084876537322998, "rewards/margins": 0.1513279229402542, "rewards/rejected": -1.4598156213760376, "step": 2785 }, { "epoch": 1.4932262920220774, "grad_norm": 11.569421426806633, "learning_rate": 5.906047197571541e-07, "logits/chosen": -0.11923675239086151, "logits/rejected": -0.13315056264400482, "logps/chosen": -1.2030599117279053, "logps/rejected": -1.418312907218933, "loss": 1.5777, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2030599117279053, "rewards/margins": 0.21525295078754425, "rewards/rejected": -1.418312907218933, "step": 2790 }, { "epoch": 1.4959023248034788, "grad_norm": 7.132297547112176, "learning_rate": 5.890726635828919e-07, "logits/chosen": -0.02672731503844261, "logits/rejected": -0.013998660258948803, "logps/chosen": -1.1751325130462646, "logps/rejected": -1.3113696575164795, "loss": 1.6077, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.1751325130462646, "rewards/margins": 0.13623718917369843, "rewards/rejected": -1.3113696575164795, "step": 2795 }, { "epoch": 1.4985783575848803, "grad_norm": 7.321205535658056, "learning_rate": 5.875397432252569e-07, "logits/chosen": -0.20826506614685059, "logits/rejected": -0.13590756058692932, "logps/chosen": -1.319227933883667, "logps/rejected": -1.5354816913604736, "loss": 1.6687, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.319227933883667, "rewards/margins": 0.21625368297100067, "rewards/rejected": -1.5354816913604736, "step": 2800 }, { "epoch": 1.4985783575848803, "eval_logits/chosen": 0.10514389723539352, "eval_logits/rejected": 0.17535388469696045, "eval_logps/chosen": -1.299961805343628, "eval_logps/rejected": -1.4737409353256226, "eval_loss": 1.6487802267074585, "eval_rewards/accuracies": 0.5697329640388489, "eval_rewards/chosen": -1.299961805343628, "eval_rewards/margins": 0.1737791746854782, "eval_rewards/rejected": -1.4737409353256226, "eval_runtime": 40.7375, "eval_samples_per_second": 33.016, "eval_steps_per_second": 8.272, "step": 2800 }, { "epoch": 1.5012543903662818, "grad_norm": 5.001064596005135, "learning_rate": 5.860059735566491e-07, "logits/chosen": -0.2607018053531647, "logits/rejected": -0.1325281709432602, "logps/chosen": -1.1559540033340454, "logps/rejected": -1.3943657875061035, "loss": 1.531, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1559540033340454, "rewards/margins": 0.23841187357902527, "rewards/rejected": -1.3943657875061035, "step": 2805 }, { "epoch": 1.5039304231476835, "grad_norm": 12.124841820595227, "learning_rate": 5.844713694577087e-07, "logits/chosen": -0.16709429025650024, "logits/rejected": -0.15542718768119812, "logps/chosen": -1.2861963510513306, "logps/rejected": -1.4874119758605957, "loss": 1.6393, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2861963510513306, "rewards/margins": 0.20121553540229797, "rewards/rejected": -1.4874119758605957, "step": 2810 }, { "epoch": 1.5066064559290853, "grad_norm": 7.814378815191505, "learning_rate": 5.829359458171714e-07, "logits/chosen": -0.11754331737756729, "logits/rejected": -0.010437699034810066, "logps/chosen": -1.280771255493164, "logps/rejected": -1.4639467000961304, "loss": 1.612, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.280771255493164, "rewards/margins": 0.1831754595041275, "rewards/rejected": -1.4639467000961304, "step": 2815 }, { "epoch": 1.5092824887104868, "grad_norm": 6.59949921623073, "learning_rate": 5.81399717531724e-07, "logits/chosen": -0.11054511368274689, "logits/rejected": 0.01979086734354496, "logps/chosen": -1.2499945163726807, "logps/rejected": -1.3047266006469727, "loss": 1.6465, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2499945163726807, "rewards/margins": 0.05473214387893677, "rewards/rejected": -1.3047266006469727, "step": 2820 }, { "epoch": 1.5119585214918883, "grad_norm": 7.0873516438434265, "learning_rate": 5.798626995058602e-07, "logits/chosen": -0.1889365315437317, "logits/rejected": -0.0532827265560627, "logps/chosen": -1.3098100423812866, "logps/rejected": -1.563604474067688, "loss": 1.6327, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3098100423812866, "rewards/margins": 0.2537943720817566, "rewards/rejected": -1.563604474067688, "step": 2825 }, { "epoch": 1.51463455427329, "grad_norm": 7.1073016737017705, "learning_rate": 5.783249066517354e-07, "logits/chosen": -0.1716947853565216, "logits/rejected": -0.04569972679018974, "logps/chosen": -1.309134602546692, "logps/rejected": -1.3283313512802124, "loss": 1.7066, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -1.309134602546692, "rewards/margins": 0.019196750596165657, "rewards/rejected": -1.3283313512802124, "step": 2830 }, { "epoch": 1.5173105870546915, "grad_norm": 9.573012875878954, "learning_rate": 5.767863538890228e-07, "logits/chosen": -0.16613741219043732, "logits/rejected": -0.031196704134345055, "logps/chosen": -1.2335695028305054, "logps/rejected": -1.385694980621338, "loss": 1.6026, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2335695028305054, "rewards/margins": 0.15212538838386536, "rewards/rejected": -1.385694980621338, "step": 2835 }, { "epoch": 1.519986619836093, "grad_norm": 8.765137445446333, "learning_rate": 5.75247056144768e-07, "logits/chosen": -0.14558474719524384, "logits/rejected": -0.054780591279268265, "logps/chosen": -1.2625131607055664, "logps/rejected": -1.4063299894332886, "loss": 1.6113, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2625131607055664, "rewards/margins": 0.14381691813468933, "rewards/rejected": -1.4063299894332886, "step": 2840 }, { "epoch": 1.5226626526174947, "grad_norm": 7.848634771449643, "learning_rate": 5.737070283532444e-07, "logits/chosen": -0.13879218697547913, "logits/rejected": -0.07548417150974274, "logps/chosen": -1.2000709772109985, "logps/rejected": -1.4368489980697632, "loss": 1.5687, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2000709772109985, "rewards/margins": 0.23677799105644226, "rewards/rejected": -1.4368489980697632, "step": 2845 }, { "epoch": 1.5253386853988962, "grad_norm": 9.429682825684978, "learning_rate": 5.721662854558084e-07, "logits/chosen": -0.18743689358234406, "logits/rejected": -0.1311352550983429, "logps/chosen": -1.2835838794708252, "logps/rejected": -1.4209805727005005, "loss": 1.6482, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -1.2835838794708252, "rewards/margins": 0.13739685714244843, "rewards/rejected": -1.4209805727005005, "step": 2850 }, { "epoch": 1.5280147181802977, "grad_norm": 6.648982314070762, "learning_rate": 5.706248424007545e-07, "logits/chosen": -0.2049282044172287, "logits/rejected": -0.07677256315946579, "logps/chosen": -1.3301162719726562, "logps/rejected": -1.5274568796157837, "loss": 1.6551, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3301162719726562, "rewards/margins": 0.1973404586315155, "rewards/rejected": -1.5274568796157837, "step": 2855 }, { "epoch": 1.5306907509616994, "grad_norm": 8.789640159262424, "learning_rate": 5.690827141431699e-07, "logits/chosen": -0.2506573796272278, "logits/rejected": -0.11205937713384628, "logps/chosen": -1.218569040298462, "logps/rejected": -1.350964903831482, "loss": 1.5879, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.218569040298462, "rewards/margins": 0.13239574432373047, "rewards/rejected": -1.350964903831482, "step": 2860 }, { "epoch": 1.5333667837431009, "grad_norm": 7.910421340948889, "learning_rate": 5.675399156447897e-07, "logits/chosen": -0.299457848072052, "logits/rejected": -0.19384565949440002, "logps/chosen": -1.2221951484680176, "logps/rejected": -1.4229544401168823, "loss": 1.6032, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2221951484680176, "rewards/margins": 0.20075933635234833, "rewards/rejected": -1.4229544401168823, "step": 2865 }, { "epoch": 1.5360428165245024, "grad_norm": 9.374208664023072, "learning_rate": 5.659964618738515e-07, "logits/chosen": -0.1947907656431198, "logits/rejected": -0.1025640144944191, "logps/chosen": -1.2786771059036255, "logps/rejected": -1.3355814218521118, "loss": 1.6788, "rewards/accuracies": 0.46875, "rewards/chosen": -1.2786771059036255, "rewards/margins": 0.05690423399209976, "rewards/rejected": -1.3355814218521118, "step": 2870 }, { "epoch": 1.538718849305904, "grad_norm": 8.115924306598316, "learning_rate": 5.644523678049509e-07, "logits/chosen": -0.1951065957546234, "logits/rejected": -0.1314360648393631, "logps/chosen": -1.2781469821929932, "logps/rejected": -1.3914257287979126, "loss": 1.6343, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2781469821929932, "rewards/margins": 0.11327870935201645, "rewards/rejected": -1.3914257287979126, "step": 2875 }, { "epoch": 1.5413948820873056, "grad_norm": 8.892601728738727, "learning_rate": 5.629076484188952e-07, "logits/chosen": -0.07616610080003738, "logits/rejected": 0.016429344192147255, "logps/chosen": -1.2245382070541382, "logps/rejected": -1.440077781677246, "loss": 1.6103, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2245382070541382, "rewards/margins": 0.21553964912891388, "rewards/rejected": -1.440077781677246, "step": 2880 }, { "epoch": 1.544070914868707, "grad_norm": 8.237630831845575, "learning_rate": 5.613623187025587e-07, "logits/chosen": -0.15724192559719086, "logits/rejected": -0.05733314901590347, "logps/chosen": -1.2409113645553589, "logps/rejected": -1.400888442993164, "loss": 1.6074, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2409113645553589, "rewards/margins": 0.1599770486354828, "rewards/rejected": -1.400888442993164, "step": 2885 }, { "epoch": 1.5467469476501088, "grad_norm": 7.162494464066858, "learning_rate": 5.598163936487369e-07, "logits/chosen": -0.2565734386444092, "logits/rejected": -0.10567812621593475, "logps/chosen": -1.2417627573013306, "logps/rejected": -1.4008761644363403, "loss": 1.6039, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2417627573013306, "rewards/margins": 0.15911336243152618, "rewards/rejected": -1.4008761644363403, "step": 2890 }, { "epoch": 1.5494229804315103, "grad_norm": 7.713818723912763, "learning_rate": 5.582698882560017e-07, "logits/chosen": -0.24017128348350525, "logits/rejected": -0.13441559672355652, "logps/chosen": -1.1790987253189087, "logps/rejected": -1.3518292903900146, "loss": 1.5616, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1790987253189087, "rewards/margins": 0.17273077368736267, "rewards/rejected": -1.3518292903900146, "step": 2895 }, { "epoch": 1.5520990132129118, "grad_norm": 7.728791758359173, "learning_rate": 5.567228175285549e-07, "logits/chosen": -0.14355513453483582, "logits/rejected": -0.06344401091337204, "logps/chosen": -1.2629728317260742, "logps/rejected": -1.4144147634506226, "loss": 1.6173, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2629728317260742, "rewards/margins": 0.15144187211990356, "rewards/rejected": -1.4144147634506226, "step": 2900 }, { "epoch": 1.5547750459943135, "grad_norm": 6.08821972601008, "learning_rate": 5.551751964760838e-07, "logits/chosen": -0.09214814752340317, "logits/rejected": -0.068619005382061, "logps/chosen": -1.22005295753479, "logps/rejected": -1.4123097658157349, "loss": 1.5815, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.22005295753479, "rewards/margins": 0.19225691258907318, "rewards/rejected": -1.4123097658157349, "step": 2905 }, { "epoch": 1.557451078775715, "grad_norm": 8.049352936261435, "learning_rate": 5.536270401136145e-07, "logits/chosen": -0.15920159220695496, "logits/rejected": -0.1046360582113266, "logps/chosen": -1.2048557996749878, "logps/rejected": -1.360347032546997, "loss": 1.5893, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2048557996749878, "rewards/margins": 0.15549114346504211, "rewards/rejected": -1.360347032546997, "step": 2910 }, { "epoch": 1.5601271115571165, "grad_norm": 9.62263310263903, "learning_rate": 5.520783634613667e-07, "logits/chosen": -0.10347437858581543, "logits/rejected": 0.02849293313920498, "logps/chosen": -1.2959470748901367, "logps/rejected": -1.4177794456481934, "loss": 1.6532, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2959470748901367, "rewards/margins": 0.12183227390050888, "rewards/rejected": -1.4177794456481934, "step": 2915 }, { "epoch": 1.5628031443385182, "grad_norm": 5.957591983364518, "learning_rate": 5.505291815446082e-07, "logits/chosen": -0.09513330459594727, "logits/rejected": -0.00013175010099075735, "logps/chosen": -1.2686645984649658, "logps/rejected": -1.4559599161148071, "loss": 1.6123, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2686645984649658, "rewards/margins": 0.18729540705680847, "rewards/rejected": -1.4559599161148071, "step": 2920 }, { "epoch": 1.5654791771199197, "grad_norm": 8.750103612028086, "learning_rate": 5.489795093935089e-07, "logits/chosen": -0.10519526898860931, "logits/rejected": -0.07157649099826813, "logps/chosen": -1.1874984502792358, "logps/rejected": -1.4648293256759644, "loss": 1.5577, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1874984502792358, "rewards/margins": 0.2773308753967285, "rewards/rejected": -1.4648293256759644, "step": 2925 }, { "epoch": 1.5681552099013212, "grad_norm": 10.277760462873392, "learning_rate": 5.474293620429946e-07, "logits/chosen": -0.22603020071983337, "logits/rejected": -0.09562810510396957, "logps/chosen": -1.2209093570709229, "logps/rejected": -1.4970707893371582, "loss": 1.5691, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2209093570709229, "rewards/margins": 0.276161253452301, "rewards/rejected": -1.4970707893371582, "step": 2930 }, { "epoch": 1.570831242682723, "grad_norm": 7.6856069730391345, "learning_rate": 5.458787545326018e-07, "logits/chosen": -0.18429319560527802, "logits/rejected": -0.06572946906089783, "logps/chosen": -1.2605324983596802, "logps/rejected": -1.4396363496780396, "loss": 1.6223, "rewards/accuracies": 0.5, "rewards/chosen": -1.2605324983596802, "rewards/margins": 0.17910373210906982, "rewards/rejected": -1.4396363496780396, "step": 2935 }, { "epoch": 1.5735072754641244, "grad_norm": 6.231542616149614, "learning_rate": 5.443277019063311e-07, "logits/chosen": -0.18334154784679413, "logits/rejected": -0.044742949306964874, "logps/chosen": -1.2537963390350342, "logps/rejected": -1.4843844175338745, "loss": 1.6136, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2537963390350342, "rewards/margins": 0.23058822751045227, "rewards/rejected": -1.4843844175338745, "step": 2940 }, { "epoch": 1.5761833082455259, "grad_norm": 9.884251728039047, "learning_rate": 5.427762192125023e-07, "logits/chosen": -0.2172185480594635, "logits/rejected": -0.10349267721176147, "logps/chosen": -1.2288261651992798, "logps/rejected": -1.3491674661636353, "loss": 1.6158, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2288261651992798, "rewards/margins": 0.12034142017364502, "rewards/rejected": -1.3491674661636353, "step": 2945 }, { "epoch": 1.5788593410269276, "grad_norm": 9.984560793832488, "learning_rate": 5.41224321503607e-07, "logits/chosen": -0.08993600308895111, "logits/rejected": 0.1004088744521141, "logps/chosen": -1.198878526687622, "logps/rejected": -1.4119997024536133, "loss": 1.5484, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.198878526687622, "rewards/margins": 0.21312110126018524, "rewards/rejected": -1.4119997024536133, "step": 2950 }, { "epoch": 1.5815353738083293, "grad_norm": 7.432126220058146, "learning_rate": 5.396720238361637e-07, "logits/chosen": -0.10413719713687897, "logits/rejected": -0.02635631524026394, "logps/chosen": -1.2171152830123901, "logps/rejected": -1.4342138767242432, "loss": 1.5748, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2171152830123901, "rewards/margins": 0.21709854900836945, "rewards/rejected": -1.4342138767242432, "step": 2955 }, { "epoch": 1.5842114065897306, "grad_norm": 8.632437936611266, "learning_rate": 5.381193412705711e-07, "logits/chosen": -0.21219193935394287, "logits/rejected": -0.11716421693563461, "logps/chosen": -1.2380011081695557, "logps/rejected": -1.4030927419662476, "loss": 1.5898, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2380011081695557, "rewards/margins": 0.16509170830249786, "rewards/rejected": -1.4030927419662476, "step": 2960 }, { "epoch": 1.5868874393711323, "grad_norm": 11.416226627501322, "learning_rate": 5.365662888709622e-07, "logits/chosen": -0.16225700080394745, "logits/rejected": -0.09003710001707077, "logps/chosen": -1.2026822566986084, "logps/rejected": -1.3658852577209473, "loss": 1.58, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2026822566986084, "rewards/margins": 0.1632029116153717, "rewards/rejected": -1.3658852577209473, "step": 2965 }, { "epoch": 1.589563472152534, "grad_norm": 11.522106023505476, "learning_rate": 5.350128817050585e-07, "logits/chosen": -0.16338010132312775, "logits/rejected": -0.026224765926599503, "logps/chosen": -1.2979391813278198, "logps/rejected": -1.40617036819458, "loss": 1.6787, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.2979391813278198, "rewards/margins": 0.10823094844818115, "rewards/rejected": -1.40617036819458, "step": 2970 }, { "epoch": 1.5922395049339353, "grad_norm": 8.561026141051741, "learning_rate": 5.334591348440229e-07, "logits/chosen": -0.12392149120569229, "logits/rejected": -0.01384222786873579, "logps/chosen": -1.263944149017334, "logps/rejected": -1.5268821716308594, "loss": 1.5979, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.263944149017334, "rewards/margins": 0.2629380524158478, "rewards/rejected": -1.5268821716308594, "step": 2975 }, { "epoch": 1.594915537715337, "grad_norm": 10.232686401494488, "learning_rate": 5.319050633623141e-07, "logits/chosen": -0.19973398745059967, "logits/rejected": -0.07975804805755615, "logps/chosen": -1.299760103225708, "logps/rejected": -1.4754284620285034, "loss": 1.6607, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.299760103225708, "rewards/margins": 0.17566823959350586, "rewards/rejected": -1.4754284620285034, "step": 2980 }, { "epoch": 1.5975915704967387, "grad_norm": 5.869578591150815, "learning_rate": 5.303506823375409e-07, "logits/chosen": -0.20419225096702576, "logits/rejected": -0.06428646296262741, "logps/chosen": -1.3240875005722046, "logps/rejected": -1.422776699066162, "loss": 1.6718, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3240875005722046, "rewards/margins": 0.0986892357468605, "rewards/rejected": -1.422776699066162, "step": 2985 }, { "epoch": 1.60026760327814, "grad_norm": 8.404848854739129, "learning_rate": 5.287960068503143e-07, "logits/chosen": -0.1343335062265396, "logits/rejected": 0.011005854234099388, "logps/chosen": -1.2011334896087646, "logps/rejected": -1.452879548072815, "loss": 1.5661, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2011334896087646, "rewards/margins": 0.25174611806869507, "rewards/rejected": -1.452879548072815, "step": 2990 }, { "epoch": 1.6029436360595417, "grad_norm": 8.63111795683355, "learning_rate": 5.272410519841032e-07, "logits/chosen": -0.15247437357902527, "logits/rejected": -0.07777571678161621, "logps/chosen": -1.3044708967208862, "logps/rejected": -1.5415220260620117, "loss": 1.6299, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3044708967208862, "rewards/margins": 0.23705117404460907, "rewards/rejected": -1.5415220260620117, "step": 2995 }, { "epoch": 1.6056196688409434, "grad_norm": 6.9809615671792775, "learning_rate": 5.256858328250861e-07, "logits/chosen": -0.18980927765369415, "logits/rejected": -0.09071429073810577, "logps/chosen": -1.306738018989563, "logps/rejected": -1.4723536968231201, "loss": 1.6838, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.306738018989563, "rewards/margins": 0.16561564803123474, "rewards/rejected": -1.4723536968231201, "step": 3000 }, { "epoch": 1.608295701622345, "grad_norm": 7.845303840240266, "learning_rate": 5.241303644620063e-07, "logits/chosen": -0.23705609142780304, "logits/rejected": -0.1214757114648819, "logps/chosen": -1.2059553861618042, "logps/rejected": -1.3896453380584717, "loss": 1.5884, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2059553861618042, "rewards/margins": 0.18369005620479584, "rewards/rejected": -1.3896453380584717, "step": 3005 }, { "epoch": 1.6109717344037464, "grad_norm": 8.923428009458794, "learning_rate": 5.225746619860248e-07, "logits/chosen": -0.21868646144866943, "logits/rejected": -0.1172574907541275, "logps/chosen": -1.2465583086013794, "logps/rejected": -1.411179542541504, "loss": 1.636, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2465583086013794, "rewards/margins": 0.16462130844593048, "rewards/rejected": -1.411179542541504, "step": 3010 }, { "epoch": 1.6136477671851481, "grad_norm": 8.58111075001241, "learning_rate": 5.210187404905735e-07, "logits/chosen": -0.07596341520547867, "logits/rejected": -0.01020109374076128, "logps/chosen": -1.241079568862915, "logps/rejected": -1.4651001691818237, "loss": 1.5836, "rewards/accuracies": 0.625, "rewards/chosen": -1.241079568862915, "rewards/margins": 0.22402064502239227, "rewards/rejected": -1.4651001691818237, "step": 3015 }, { "epoch": 1.6163237999665496, "grad_norm": 7.440191528787235, "learning_rate": 5.194626150712098e-07, "logits/chosen": -0.21312424540519714, "logits/rejected": -0.08387692272663116, "logps/chosen": -1.241636037826538, "logps/rejected": -1.3631786108016968, "loss": 1.618, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.241636037826538, "rewards/margins": 0.12154249846935272, "rewards/rejected": -1.3631786108016968, "step": 3020 }, { "epoch": 1.6189998327479511, "grad_norm": 8.20125824115334, "learning_rate": 5.179063008254695e-07, "logits/chosen": -0.15682704746723175, "logits/rejected": -0.04331011325120926, "logps/chosen": -1.2104800939559937, "logps/rejected": -1.360103964805603, "loss": 1.597, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2104800939559937, "rewards/margins": 0.149623841047287, "rewards/rejected": -1.360103964805603, "step": 3025 }, { "epoch": 1.6216758655293528, "grad_norm": 5.562431256517953, "learning_rate": 5.163498128527199e-07, "logits/chosen": -0.13923481106758118, "logits/rejected": -0.024134326726198196, "logps/chosen": -1.2932157516479492, "logps/rejected": -1.400172472000122, "loss": 1.6373, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2932157516479492, "rewards/margins": 0.10695650428533554, "rewards/rejected": -1.400172472000122, "step": 3030 }, { "epoch": 1.6243518983107543, "grad_norm": 8.08617825073629, "learning_rate": 5.147931662540144e-07, "logits/chosen": -0.03716077283024788, "logits/rejected": 0.04836971312761307, "logps/chosen": -1.2672970294952393, "logps/rejected": -1.349109411239624, "loss": 1.6496, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2672970294952393, "rewards/margins": 0.08181246370077133, "rewards/rejected": -1.349109411239624, "step": 3035 }, { "epoch": 1.6270279310921558, "grad_norm": 11.86487644872994, "learning_rate": 5.132363761319449e-07, "logits/chosen": -0.15009820461273193, "logits/rejected": -0.10209157317876816, "logps/chosen": -1.1850813627243042, "logps/rejected": -1.4225887060165405, "loss": 1.5398, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1850813627243042, "rewards/margins": 0.23750725388526917, "rewards/rejected": -1.4225887060165405, "step": 3040 }, { "epoch": 1.6297039638735575, "grad_norm": 14.321647563947106, "learning_rate": 5.116794575904962e-07, "logits/chosen": -0.1324605941772461, "logits/rejected": -0.0472918264567852, "logps/chosen": -1.217961072921753, "logps/rejected": -1.329044222831726, "loss": 1.6154, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.217961072921753, "rewards/margins": 0.11108307540416718, "rewards/rejected": -1.329044222831726, "step": 3045 }, { "epoch": 1.632379996654959, "grad_norm": 9.063482796097663, "learning_rate": 5.101224257348987e-07, "logits/chosen": -0.18629930913448334, "logits/rejected": -0.05383709818124771, "logps/chosen": -1.2813564538955688, "logps/rejected": -1.503534197807312, "loss": 1.6073, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2813564538955688, "rewards/margins": 0.2221778929233551, "rewards/rejected": -1.503534197807312, "step": 3050 }, { "epoch": 1.6350560294363605, "grad_norm": 6.12253360363484, "learning_rate": 5.085652956714823e-07, "logits/chosen": -0.19061146676540375, "logits/rejected": -0.07919798791408539, "logps/chosen": -1.217474341392517, "logps/rejected": -1.438800573348999, "loss": 1.5872, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.217474341392517, "rewards/margins": 0.22132618725299835, "rewards/rejected": -1.438800573348999, "step": 3055 }, { "epoch": 1.6377320622177622, "grad_norm": 6.7458196371965755, "learning_rate": 5.070080825075298e-07, "logits/chosen": -0.19373759627342224, "logits/rejected": -0.05019166320562363, "logps/chosen": -1.2642765045166016, "logps/rejected": -1.4489099979400635, "loss": 1.6179, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2642765045166016, "rewards/margins": 0.18463341891765594, "rewards/rejected": -1.4489099979400635, "step": 3060 }, { "epoch": 1.6404080949991637, "grad_norm": 7.8307477062379895, "learning_rate": 5.0545080135113e-07, "logits/chosen": -0.09949559718370438, "logits/rejected": -0.0767357349395752, "logps/chosen": -1.2530882358551025, "logps/rejected": -1.5171091556549072, "loss": 1.5936, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2530882358551025, "rewards/margins": 0.264021098613739, "rewards/rejected": -1.5171091556549072, "step": 3065 }, { "epoch": 1.6430841277805652, "grad_norm": 6.924831831509434, "learning_rate": 5.038934673110316e-07, "logits/chosen": -0.2290758341550827, "logits/rejected": -0.13408730924129486, "logps/chosen": -1.249206781387329, "logps/rejected": -1.4324674606323242, "loss": 1.6044, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.249206781387329, "rewards/margins": 0.18326060473918915, "rewards/rejected": -1.4324674606323242, "step": 3070 }, { "epoch": 1.645760160561967, "grad_norm": 6.550753239242128, "learning_rate": 5.023360954964963e-07, "logits/chosen": -0.25031578540802, "logits/rejected": -0.20224837958812714, "logps/chosen": -1.1969693899154663, "logps/rejected": -1.402588963508606, "loss": 1.5611, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1969693899154663, "rewards/margins": 0.20561976730823517, "rewards/rejected": -1.402588963508606, "step": 3075 }, { "epoch": 1.6484361933433684, "grad_norm": 8.22013975849115, "learning_rate": 5.007787010171524e-07, "logits/chosen": -0.2754722237586975, "logits/rejected": -0.11621172726154327, "logps/chosen": -1.1683692932128906, "logps/rejected": -1.361311912536621, "loss": 1.5563, "rewards/accuracies": 0.59375, "rewards/chosen": -1.1683692932128906, "rewards/margins": 0.19294250011444092, "rewards/rejected": -1.361311912536621, "step": 3080 }, { "epoch": 1.65111222612477, "grad_norm": 7.9245890872821905, "learning_rate": 4.992212989828477e-07, "logits/chosen": -0.11269180476665497, "logits/rejected": -0.11005918681621552, "logps/chosen": -1.184718370437622, "logps/rejected": -1.3857439756393433, "loss": 1.5516, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.184718370437622, "rewards/margins": 0.20102588832378387, "rewards/rejected": -1.3857439756393433, "step": 3085 }, { "epoch": 1.6537882589061716, "grad_norm": 7.910985905319788, "learning_rate": 4.976639045035036e-07, "logits/chosen": -0.0895986557006836, "logits/rejected": -0.034634217619895935, "logps/chosen": -1.2269132137298584, "logps/rejected": -1.3635425567626953, "loss": 1.6229, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2269132137298584, "rewards/margins": 0.1366293728351593, "rewards/rejected": -1.3635425567626953, "step": 3090 }, { "epoch": 1.6564642916875731, "grad_norm": 8.989701022473724, "learning_rate": 4.961065326889683e-07, "logits/chosen": -0.16246844828128815, "logits/rejected": -0.04997162148356438, "logps/chosen": -1.2454721927642822, "logps/rejected": -1.412967562675476, "loss": 1.6076, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2454721927642822, "rewards/margins": 0.16749544441699982, "rewards/rejected": -1.412967562675476, "step": 3095 }, { "epoch": 1.6591403244689746, "grad_norm": 9.286518315162434, "learning_rate": 4.9454919864887e-07, "logits/chosen": -0.3038082718849182, "logits/rejected": -0.1936473548412323, "logps/chosen": -1.2898128032684326, "logps/rejected": -1.420832633972168, "loss": 1.6559, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2898128032684326, "rewards/margins": 0.1310197114944458, "rewards/rejected": -1.420832633972168, "step": 3100 }, { "epoch": 1.6618163572503764, "grad_norm": 9.867318872639597, "learning_rate": 4.929919174924701e-07, "logits/chosen": -0.2150929719209671, "logits/rejected": -0.06844820082187653, "logps/chosen": -1.2798480987548828, "logps/rejected": -1.4365907907485962, "loss": 1.6296, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2798480987548828, "rewards/margins": 0.15674252808094025, "rewards/rejected": -1.4365907907485962, "step": 3105 }, { "epoch": 1.6644923900317778, "grad_norm": 7.123668267524398, "learning_rate": 4.914347043285177e-07, "logits/chosen": -0.15136371552944183, "logits/rejected": -0.0637313649058342, "logps/chosen": -1.2542929649353027, "logps/rejected": -1.4589463472366333, "loss": 1.6048, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2542929649353027, "rewards/margins": 0.2046533077955246, "rewards/rejected": -1.4589463472366333, "step": 3110 }, { "epoch": 1.6671684228131793, "grad_norm": 5.689036503731929, "learning_rate": 4.898775742651013e-07, "logits/chosen": -0.09614956378936768, "logits/rejected": -0.038172896951436996, "logps/chosen": -1.260207176208496, "logps/rejected": -1.4991014003753662, "loss": 1.594, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.260207176208496, "rewards/margins": 0.2388942688703537, "rewards/rejected": -1.4991014003753662, "step": 3115 }, { "epoch": 1.669844455594581, "grad_norm": 5.621420712895871, "learning_rate": 4.883205424095037e-07, "logits/chosen": -0.22079035639762878, "logits/rejected": -0.10798345506191254, "logps/chosen": -1.3159030675888062, "logps/rejected": -1.482990026473999, "loss": 1.6713, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.3159030675888062, "rewards/margins": 0.16708692908287048, "rewards/rejected": -1.482990026473999, "step": 3120 }, { "epoch": 1.6725204883759828, "grad_norm": 7.30915810696171, "learning_rate": 4.86763623868055e-07, "logits/chosen": -0.130017951130867, "logits/rejected": -0.050635069608688354, "logps/chosen": -1.3146250247955322, "logps/rejected": -1.5301706790924072, "loss": 1.6536, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3146250247955322, "rewards/margins": 0.21554553508758545, "rewards/rejected": -1.5301706790924072, "step": 3125 }, { "epoch": 1.675196521157384, "grad_norm": 7.061782489077253, "learning_rate": 4.852068337459856e-07, "logits/chosen": -0.1131129041314125, "logits/rejected": -0.011948710307478905, "logps/chosen": -1.3172558546066284, "logps/rejected": -1.4849779605865479, "loss": 1.6484, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3172558546066284, "rewards/margins": 0.16772189736366272, "rewards/rejected": -1.4849779605865479, "step": 3130 }, { "epoch": 1.6778725539387858, "grad_norm": 7.380763411129059, "learning_rate": 4.8365018714728e-07, "logits/chosen": -0.09311554580926895, "logits/rejected": -0.06127345561981201, "logps/chosen": -1.3208004236221313, "logps/rejected": -1.4602913856506348, "loss": 1.6796, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3208004236221313, "rewards/margins": 0.13949090242385864, "rewards/rejected": -1.4602913856506348, "step": 3135 }, { "epoch": 1.6805485867201875, "grad_norm": 5.702850842533426, "learning_rate": 4.820936991745304e-07, "logits/chosen": -0.34844082593917847, "logits/rejected": -0.21953165531158447, "logps/chosen": -1.1999485492706299, "logps/rejected": -1.3034613132476807, "loss": 1.5953, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.1999485492706299, "rewards/margins": 0.10351266711950302, "rewards/rejected": -1.3034613132476807, "step": 3140 }, { "epoch": 1.6832246195015887, "grad_norm": 8.342386540198653, "learning_rate": 4.8053738492879e-07, "logits/chosen": -0.1601831614971161, "logits/rejected": -0.055739402770996094, "logps/chosen": -1.242384910583496, "logps/rejected": -1.351478934288025, "loss": 1.621, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.242384910583496, "rewards/margins": 0.10909397900104523, "rewards/rejected": -1.351478934288025, "step": 3145 }, { "epoch": 1.6859006522829905, "grad_norm": 7.9266129658506905, "learning_rate": 4.789812595094265e-07, "logits/chosen": -0.2788141369819641, "logits/rejected": -0.1777099072933197, "logps/chosen": -1.3118698596954346, "logps/rejected": -1.4719301462173462, "loss": 1.6398, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.3118698596954346, "rewards/margins": 0.16006025671958923, "rewards/rejected": -1.4719301462173462, "step": 3150 }, { "epoch": 1.6885766850643922, "grad_norm": 11.687494404795508, "learning_rate": 4.774253380139752e-07, "logits/chosen": -0.2808185815811157, "logits/rejected": -0.1869959831237793, "logps/chosen": -1.194966435432434, "logps/rejected": -1.4038288593292236, "loss": 1.5647, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.194966435432434, "rewards/margins": 0.20886237919330597, "rewards/rejected": -1.4038288593292236, "step": 3155 }, { "epoch": 1.6912527178457935, "grad_norm": 8.339394315233186, "learning_rate": 4.758696355379936e-07, "logits/chosen": -0.24925696849822998, "logits/rejected": -0.22909533977508545, "logps/chosen": -1.2405601739883423, "logps/rejected": -1.4713505506515503, "loss": 1.5969, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2405601739883423, "rewards/margins": 0.2307904064655304, "rewards/rejected": -1.4713505506515503, "step": 3160 }, { "epoch": 1.6939287506271952, "grad_norm": 6.232115794794186, "learning_rate": 4.743141671749138e-07, "logits/chosen": -0.26409196853637695, "logits/rejected": -0.1972818374633789, "logps/chosen": -1.281771183013916, "logps/rejected": -1.4099771976470947, "loss": 1.6426, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.281771183013916, "rewards/margins": 0.12820599973201752, "rewards/rejected": -1.4099771976470947, "step": 3165 }, { "epoch": 1.6966047834085969, "grad_norm": 6.145214171888279, "learning_rate": 4.727589480158968e-07, "logits/chosen": -0.22552287578582764, "logits/rejected": -0.1499921828508377, "logps/chosen": -1.2544889450073242, "logps/rejected": -1.4211806058883667, "loss": 1.6312, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2544889450073242, "rewards/margins": 0.1666916310787201, "rewards/rejected": -1.4211806058883667, "step": 3170 }, { "epoch": 1.6992808161899984, "grad_norm": 9.181957372766112, "learning_rate": 4.712039931496855e-07, "logits/chosen": -0.2748224139213562, "logits/rejected": -0.20714323222637177, "logps/chosen": -1.2339074611663818, "logps/rejected": -1.3875453472137451, "loss": 1.6198, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2339074611663818, "rewards/margins": 0.15363790094852448, "rewards/rejected": -1.3875453472137451, "step": 3175 }, { "epoch": 1.7019568489713999, "grad_norm": 5.559172509042021, "learning_rate": 4.6964931766245905e-07, "logits/chosen": -0.1216423287987709, "logits/rejected": -0.0782778263092041, "logps/chosen": -1.2822221517562866, "logps/rejected": -1.4988685846328735, "loss": 1.6129, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2822221517562866, "rewards/margins": 0.21664640307426453, "rewards/rejected": -1.4988685846328735, "step": 3180 }, { "epoch": 1.7046328817528016, "grad_norm": 8.053248508222694, "learning_rate": 4.6809493663768575e-07, "logits/chosen": -0.17452619969844818, "logits/rejected": -0.1666513979434967, "logps/chosen": -1.2001672983169556, "logps/rejected": -1.3536691665649414, "loss": 1.5746, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2001672983169556, "rewards/margins": 0.15350192785263062, "rewards/rejected": -1.3536691665649414, "step": 3185 }, { "epoch": 1.707308914534203, "grad_norm": 8.155853541198367, "learning_rate": 4.6654086515597716e-07, "logits/chosen": -0.21809610724449158, "logits/rejected": -0.09687723219394684, "logps/chosen": -1.2203891277313232, "logps/rejected": -1.4710357189178467, "loss": 1.5438, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2203891277313232, "rewards/margins": 0.2506466507911682, "rewards/rejected": -1.4710357189178467, "step": 3190 }, { "epoch": 1.7099849473156046, "grad_norm": 5.6339085467216, "learning_rate": 4.6498711829494154e-07, "logits/chosen": -0.23436173796653748, "logits/rejected": -0.14631803333759308, "logps/chosen": -1.227378249168396, "logps/rejected": -1.4630035161972046, "loss": 1.5698, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.227378249168396, "rewards/margins": 0.2356252372264862, "rewards/rejected": -1.4630035161972046, "step": 3195 }, { "epoch": 1.7126609800970063, "grad_norm": 8.215408437303552, "learning_rate": 4.6343371112903777e-07, "logits/chosen": -0.1575685441493988, "logits/rejected": -0.03138052672147751, "logps/chosen": -1.2763001918792725, "logps/rejected": -1.5757664442062378, "loss": 1.6012, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2763001918792725, "rewards/margins": 0.2994665205478668, "rewards/rejected": -1.5757664442062378, "step": 3200 }, { "epoch": 1.7126609800970063, "eval_logits/chosen": 0.11330799758434296, "eval_logits/rejected": 0.18483346700668335, "eval_logps/chosen": -1.3009766340255737, "eval_logps/rejected": -1.4718098640441895, "eval_loss": 1.6493672132492065, "eval_rewards/accuracies": 0.5675074458122253, "eval_rewards/chosen": -1.3009766340255737, "eval_rewards/margins": 0.17083337903022766, "eval_rewards/rejected": -1.4718098640441895, "eval_runtime": 40.5995, "eval_samples_per_second": 33.129, "eval_steps_per_second": 8.301, "step": 3200 }, { "epoch": 1.7153370128784078, "grad_norm": 5.519402528838374, "learning_rate": 4.618806587294291e-07, "logits/chosen": -0.2710733413696289, "logits/rejected": -0.18044158816337585, "logps/chosen": -1.2949516773223877, "logps/rejected": -1.4818966388702393, "loss": 1.6292, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2949516773223877, "rewards/margins": 0.1869450807571411, "rewards/rejected": -1.4818966388702393, "step": 3205 }, { "epoch": 1.7180130456598093, "grad_norm": 9.347170465131176, "learning_rate": 4.603279761638365e-07, "logits/chosen": -0.2634008228778839, "logits/rejected": -0.18680989742279053, "logps/chosen": -1.2629354000091553, "logps/rejected": -1.4384006261825562, "loss": 1.6268, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2629354000091553, "rewards/margins": 0.17546531558036804, "rewards/rejected": -1.4384006261825562, "step": 3210 }, { "epoch": 1.720689078441211, "grad_norm": 9.617166522314575, "learning_rate": 4.5877567849639315e-07, "logits/chosen": -0.23709094524383545, "logits/rejected": -0.13341295719146729, "logps/chosen": -1.2301095724105835, "logps/rejected": -1.3809475898742676, "loss": 1.6028, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2301095724105835, "rewards/margins": 0.15083806216716766, "rewards/rejected": -1.3809475898742676, "step": 3215 }, { "epoch": 1.7233651112226125, "grad_norm": 6.722448806937953, "learning_rate": 4.572237807874979e-07, "logits/chosen": -0.23734502494335175, "logits/rejected": -0.08293096721172333, "logps/chosen": -1.3151230812072754, "logps/rejected": -1.4774357080459595, "loss": 1.6529, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.3151230812072754, "rewards/margins": 0.16231267154216766, "rewards/rejected": -1.4774357080459595, "step": 3220 }, { "epoch": 1.726041144004014, "grad_norm": 9.235009754404345, "learning_rate": 4.5567229809366895e-07, "logits/chosen": -0.22738520801067352, "logits/rejected": -0.12039558589458466, "logps/chosen": -1.1815749406814575, "logps/rejected": -1.3902910947799683, "loss": 1.5459, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1815749406814575, "rewards/margins": 0.2087162286043167, "rewards/rejected": -1.3902910947799683, "step": 3225 }, { "epoch": 1.7287171767854157, "grad_norm": 6.67073380977587, "learning_rate": 4.541212454673984e-07, "logits/chosen": -0.23986825346946716, "logits/rejected": -0.1233188733458519, "logps/chosen": -1.22847580909729, "logps/rejected": -1.4644906520843506, "loss": 1.5682, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.22847580909729, "rewards/margins": 0.23601479828357697, "rewards/rejected": -1.4644906520843506, "step": 3230 }, { "epoch": 1.7313932095668172, "grad_norm": 8.575828281719883, "learning_rate": 4.525706379570055e-07, "logits/chosen": -0.1955959051847458, "logits/rejected": -0.15789765119552612, "logps/chosen": -1.2535419464111328, "logps/rejected": -1.437584638595581, "loss": 1.6022, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2535419464111328, "rewards/margins": 0.18404273688793182, "rewards/rejected": -1.437584638595581, "step": 3235 }, { "epoch": 1.7340692423482187, "grad_norm": 6.262181768654312, "learning_rate": 4.510204906064911e-07, "logits/chosen": -0.11629863828420639, "logits/rejected": -0.04444526880979538, "logps/chosen": -1.1999154090881348, "logps/rejected": -1.4560317993164062, "loss": 1.5254, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1999154090881348, "rewards/margins": 0.25611644983291626, "rewards/rejected": -1.4560317993164062, "step": 3240 }, { "epoch": 1.7367452751296204, "grad_norm": 9.168006641027242, "learning_rate": 4.4947081845539177e-07, "logits/chosen": -0.30351483821868896, "logits/rejected": -0.1936475783586502, "logps/chosen": -1.205866813659668, "logps/rejected": -1.3990415334701538, "loss": 1.5854, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.205866813659668, "rewards/margins": 0.193174809217453, "rewards/rejected": -1.3990415334701538, "step": 3245 }, { "epoch": 1.739421307911022, "grad_norm": 7.1099536920776725, "learning_rate": 4.479216365386333e-07, "logits/chosen": -0.10585144907236099, "logits/rejected": -0.015230861492455006, "logps/chosen": -1.2517902851104736, "logps/rejected": -1.476215124130249, "loss": 1.5946, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2517902851104736, "rewards/margins": 0.2244250327348709, "rewards/rejected": -1.476215124130249, "step": 3250 }, { "epoch": 1.7420973406924234, "grad_norm": 7.28474636819934, "learning_rate": 4.4637295988638555e-07, "logits/chosen": -0.12763206660747528, "logits/rejected": -0.06798548251390457, "logps/chosen": -1.3406620025634766, "logps/rejected": -1.4056285619735718, "loss": 1.6821, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3406620025634766, "rewards/margins": 0.0649665966629982, "rewards/rejected": -1.4056285619735718, "step": 3255 }, { "epoch": 1.744773373473825, "grad_norm": 8.174404159145823, "learning_rate": 4.4482480352391623e-07, "logits/chosen": -0.2601187229156494, "logits/rejected": -0.1402393877506256, "logps/chosen": -1.2974830865859985, "logps/rejected": -1.3821454048156738, "loss": 1.642, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2974830865859985, "rewards/margins": 0.0846622958779335, "rewards/rejected": -1.3821454048156738, "step": 3260 }, { "epoch": 1.7474494062552266, "grad_norm": 11.452650486684668, "learning_rate": 4.4327718247144507e-07, "logits/chosen": -0.14017020165920258, "logits/rejected": -0.049223341047763824, "logps/chosen": -1.1975616216659546, "logps/rejected": -1.4038350582122803, "loss": 1.5632, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1975616216659546, "rewards/margins": 0.20627336204051971, "rewards/rejected": -1.4038350582122803, "step": 3265 }, { "epoch": 1.750125439036628, "grad_norm": 8.279683070973517, "learning_rate": 4.417301117439984e-07, "logits/chosen": -0.15070870518684387, "logits/rejected": -0.019910674542188644, "logps/chosen": -1.157780647277832, "logps/rejected": -1.3996648788452148, "loss": 1.5372, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.157780647277832, "rewards/margins": 0.2418842613697052, "rewards/rejected": -1.3996648788452148, "step": 3270 }, { "epoch": 1.7528014718180298, "grad_norm": 9.072661083572072, "learning_rate": 4.401836063512631e-07, "logits/chosen": -0.17229340970516205, "logits/rejected": 0.08592073619365692, "logps/chosen": -1.2706421613693237, "logps/rejected": -1.3676345348358154, "loss": 1.6445, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2706421613693237, "rewards/margins": 0.09699241071939468, "rewards/rejected": -1.3676345348358154, "step": 3275 }, { "epoch": 1.7554775045994313, "grad_norm": 8.30874909165034, "learning_rate": 4.386376812974413e-07, "logits/chosen": -0.16627468168735504, "logits/rejected": -0.08861502259969711, "logps/chosen": -1.2122783660888672, "logps/rejected": -1.3982435464859009, "loss": 1.5866, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2122783660888672, "rewards/margins": 0.18596510589122772, "rewards/rejected": -1.3982435464859009, "step": 3280 }, { "epoch": 1.7581535373808328, "grad_norm": 7.630394959614666, "learning_rate": 4.370923515811048e-07, "logits/chosen": -0.19240662455558777, "logits/rejected": -0.01533288974314928, "logps/chosen": -1.2119776010513306, "logps/rejected": -1.4352428913116455, "loss": 1.6006, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2119776010513306, "rewards/margins": 0.22326524555683136, "rewards/rejected": -1.4352428913116455, "step": 3285 }, { "epoch": 1.7608295701622345, "grad_norm": 8.218490443754595, "learning_rate": 4.35547632195049e-07, "logits/chosen": -0.1405266672372818, "logits/rejected": -0.04814616218209267, "logps/chosen": -1.2359240055084229, "logps/rejected": -1.331750512123108, "loss": 1.6362, "rewards/accuracies": 0.5, "rewards/chosen": -1.2359240055084229, "rewards/margins": 0.0958266407251358, "rewards/rejected": -1.331750512123108, "step": 3290 }, { "epoch": 1.763505602943636, "grad_norm": 9.22563905261601, "learning_rate": 4.340035381261484e-07, "logits/chosen": -0.1587071567773819, "logits/rejected": -0.11247760057449341, "logps/chosen": -1.3131606578826904, "logps/rejected": -1.4168140888214111, "loss": 1.7043, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.3131606578826904, "rewards/margins": 0.10365326702594757, "rewards/rejected": -1.4168140888214111, "step": 3295 }, { "epoch": 1.7661816357250375, "grad_norm": 8.056133465059338, "learning_rate": 4.324600843552104e-07, "logits/chosen": -0.24770304560661316, "logits/rejected": -0.13748589158058167, "logps/chosen": -1.3128284215927124, "logps/rejected": -1.464592695236206, "loss": 1.6644, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3128284215927124, "rewards/margins": 0.15176445245742798, "rewards/rejected": -1.464592695236206, "step": 3300 }, { "epoch": 1.7688576685064392, "grad_norm": 9.249667644920944, "learning_rate": 4.309172858568302e-07, "logits/chosen": -0.23670992255210876, "logits/rejected": -0.12905147671699524, "logps/chosen": -1.2816505432128906, "logps/rejected": -1.3982784748077393, "loss": 1.6566, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2816505432128906, "rewards/margins": 0.11662785708904266, "rewards/rejected": -1.3982784748077393, "step": 3305 }, { "epoch": 1.771533701287841, "grad_norm": 7.673087844994584, "learning_rate": 4.293751575992455e-07, "logits/chosen": -0.09383638203144073, "logits/rejected": -0.05546841025352478, "logps/chosen": -1.2519562244415283, "logps/rejected": -1.4098784923553467, "loss": 1.6148, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2519562244415283, "rewards/margins": 0.15792234241962433, "rewards/rejected": -1.4098784923553467, "step": 3310 }, { "epoch": 1.7742097340692422, "grad_norm": 9.027262073359903, "learning_rate": 4.278337145441916e-07, "logits/chosen": -0.2571185231208801, "logits/rejected": -0.142175555229187, "logps/chosen": -1.216233253479004, "logps/rejected": -1.3984395265579224, "loss": 1.5919, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.216233253479004, "rewards/margins": 0.18220646679401398, "rewards/rejected": -1.3984395265579224, "step": 3315 }, { "epoch": 1.776885766850644, "grad_norm": 7.06508547800236, "learning_rate": 4.262929716467556e-07, "logits/chosen": -0.1786039024591446, "logits/rejected": -0.02842709794640541, "logps/chosen": -1.2456203699111938, "logps/rejected": -1.4679776430130005, "loss": 1.5921, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2456203699111938, "rewards/margins": 0.22235722839832306, "rewards/rejected": -1.4679776430130005, "step": 3320 }, { "epoch": 1.7795617996320456, "grad_norm": 8.012772750013468, "learning_rate": 4.247529438552321e-07, "logits/chosen": -0.26813259720802307, "logits/rejected": -0.11593590676784515, "logps/chosen": -1.2606017589569092, "logps/rejected": -1.45313560962677, "loss": 1.6097, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2606017589569092, "rewards/margins": 0.1925339698791504, "rewards/rejected": -1.45313560962677, "step": 3325 }, { "epoch": 1.782237832413447, "grad_norm": 9.729844540213076, "learning_rate": 4.232136461109773e-07, "logits/chosen": -0.13743022084236145, "logits/rejected": -0.04654218629002571, "logps/chosen": -1.1681630611419678, "logps/rejected": -1.412596583366394, "loss": 1.5242, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1681630611419678, "rewards/margins": 0.2444334775209427, "rewards/rejected": -1.412596583366394, "step": 3330 }, { "epoch": 1.7849138651948486, "grad_norm": 11.710305594086776, "learning_rate": 4.216750933482646e-07, "logits/chosen": -0.1769091784954071, "logits/rejected": -0.03825043514370918, "logps/chosen": -1.2831181287765503, "logps/rejected": -1.4444267749786377, "loss": 1.6162, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2831181287765503, "rewards/margins": 0.16130872070789337, "rewards/rejected": -1.4444267749786377, "step": 3335 }, { "epoch": 1.7875898979762503, "grad_norm": 6.253784938341751, "learning_rate": 4.2013730049413986e-07, "logits/chosen": -0.12877388298511505, "logits/rejected": -0.01266949437558651, "logps/chosen": -1.2255029678344727, "logps/rejected": -1.4497065544128418, "loss": 1.5884, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2255029678344727, "rewards/margins": 0.22420351207256317, "rewards/rejected": -1.4497065544128418, "step": 3340 }, { "epoch": 1.7902659307576518, "grad_norm": 8.849273462673999, "learning_rate": 4.1860028246827594e-07, "logits/chosen": -0.1709320843219757, "logits/rejected": -0.03634721040725708, "logps/chosen": -1.1567004919052124, "logps/rejected": -1.364342451095581, "loss": 1.5424, "rewards/accuracies": 0.5625, "rewards/chosen": -1.1567004919052124, "rewards/margins": 0.20764203369617462, "rewards/rejected": -1.364342451095581, "step": 3345 }, { "epoch": 1.7929419635390533, "grad_norm": 8.5624059736707, "learning_rate": 4.170640541828285e-07, "logits/chosen": -0.27934056520462036, "logits/rejected": -0.15984955430030823, "logps/chosen": -1.321946620941162, "logps/rejected": -1.4497636556625366, "loss": 1.6759, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.321946620941162, "rewards/margins": 0.1278170347213745, "rewards/rejected": -1.4497636556625366, "step": 3350 }, { "epoch": 1.795617996320455, "grad_norm": 10.439526585881291, "learning_rate": 4.1552863054229116e-07, "logits/chosen": -0.035465117543935776, "logits/rejected": -0.004923814907670021, "logps/chosen": -1.3120362758636475, "logps/rejected": -1.3973262310028076, "loss": 1.7083, "rewards/accuracies": 0.46875, "rewards/chosen": -1.3120362758636475, "rewards/margins": 0.08529011905193329, "rewards/rejected": -1.3973262310028076, "step": 3355 }, { "epoch": 1.7982940291018565, "grad_norm": 8.191935181400508, "learning_rate": 4.139940264433508e-07, "logits/chosen": -0.17950890958309174, "logits/rejected": -0.0031684041023254395, "logps/chosen": -1.2063032388687134, "logps/rejected": -1.3835880756378174, "loss": 1.5792, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2063032388687134, "rewards/margins": 0.17728491127490997, "rewards/rejected": -1.3835880756378174, "step": 3360 }, { "epoch": 1.800970061883258, "grad_norm": 5.825131917702379, "learning_rate": 4.1246025677474303e-07, "logits/chosen": -0.19442106783390045, "logits/rejected": -0.06658891588449478, "logps/chosen": -1.2423062324523926, "logps/rejected": -1.4395949840545654, "loss": 1.5854, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2423062324523926, "rewards/margins": 0.1972886472940445, "rewards/rejected": -1.4395949840545654, "step": 3365 }, { "epoch": 1.8036460946646597, "grad_norm": 6.410860625207539, "learning_rate": 4.10927336417108e-07, "logits/chosen": -0.17737777531147003, "logits/rejected": -0.05374649167060852, "logps/chosen": -1.2333533763885498, "logps/rejected": -1.3829739093780518, "loss": 1.6015, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2333533763885498, "rewards/margins": 0.14962056279182434, "rewards/rejected": -1.3829739093780518, "step": 3370 }, { "epoch": 1.8063221274460612, "grad_norm": 7.866950645857972, "learning_rate": 4.093952802428457e-07, "logits/chosen": -0.010890421457588673, "logits/rejected": 0.023261893540620804, "logps/chosen": -1.268868088722229, "logps/rejected": -1.3623992204666138, "loss": 1.6544, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.268868088722229, "rewards/margins": 0.09353115409612656, "rewards/rejected": -1.3623992204666138, "step": 3375 }, { "epoch": 1.8089981602274627, "grad_norm": 5.672252950162126, "learning_rate": 4.0786410311597184e-07, "logits/chosen": -0.24759213626384735, "logits/rejected": -0.1401231288909912, "logps/chosen": -1.2340792417526245, "logps/rejected": -1.4476680755615234, "loss": 1.5787, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2340792417526245, "rewards/margins": 0.21358875930309296, "rewards/rejected": -1.4476680755615234, "step": 3380 }, { "epoch": 1.8116741930088645, "grad_norm": 7.211360904627205, "learning_rate": 4.063338198919737e-07, "logits/chosen": -0.2078475058078766, "logits/rejected": -0.1844664216041565, "logps/chosen": -1.294800043106079, "logps/rejected": -1.4443811178207397, "loss": 1.6502, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.294800043106079, "rewards/margins": 0.14958105981349945, "rewards/rejected": -1.4443811178207397, "step": 3385 }, { "epoch": 1.814350225790266, "grad_norm": 10.003239488954332, "learning_rate": 4.0480444541766575e-07, "logits/chosen": -0.15505781769752502, "logits/rejected": -0.05917396396398544, "logps/chosen": -1.3137118816375732, "logps/rejected": -1.4140517711639404, "loss": 1.6835, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.3137118816375732, "rewards/margins": 0.10033991187810898, "rewards/rejected": -1.4140517711639404, "step": 3390 }, { "epoch": 1.8170262585716674, "grad_norm": 8.940401448632166, "learning_rate": 4.0327599453104606e-07, "logits/chosen": -0.2057075947523117, "logits/rejected": -0.13585631549358368, "logps/chosen": -1.1912782192230225, "logps/rejected": -1.4137266874313354, "loss": 1.5635, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1912782192230225, "rewards/margins": 0.22244830429553986, "rewards/rejected": -1.4137266874313354, "step": 3395 }, { "epoch": 1.8197022913530692, "grad_norm": 8.003560686277208, "learning_rate": 4.017484820611514e-07, "logits/chosen": -0.1746649146080017, "logits/rejected": -0.07764903455972672, "logps/chosen": -1.2590306997299194, "logps/rejected": -1.4080326557159424, "loss": 1.6141, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2590306997299194, "rewards/margins": 0.14900176227092743, "rewards/rejected": -1.4080326557159424, "step": 3400 }, { "epoch": 1.8223783241344707, "grad_norm": 11.346465262398201, "learning_rate": 4.002219228279148e-07, "logits/chosen": -0.18884289264678955, "logits/rejected": -0.061424195766448975, "logps/chosen": -1.2502999305725098, "logps/rejected": -1.4279358386993408, "loss": 1.609, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2502999305725098, "rewards/margins": 0.17763593792915344, "rewards/rejected": -1.4279358386993408, "step": 3405 }, { "epoch": 1.8250543569158721, "grad_norm": 8.144778692952709, "learning_rate": 3.9869633164202045e-07, "logits/chosen": -0.18082715570926666, "logits/rejected": -0.0020881250966340303, "logps/chosen": -1.3831933736801147, "logps/rejected": -1.4894465208053589, "loss": 1.7307, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3831933736801147, "rewards/margins": 0.10625322163105011, "rewards/rejected": -1.4894465208053589, "step": 3410 }, { "epoch": 1.8277303896972739, "grad_norm": 7.661754333515176, "learning_rate": 3.9717172330476077e-07, "logits/chosen": -0.16274294257164001, "logits/rejected": -0.07826801389455795, "logps/chosen": -1.2358801364898682, "logps/rejected": -1.4467062950134277, "loss": 1.5922, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2358801364898682, "rewards/margins": 0.2108263075351715, "rewards/rejected": -1.4467062950134277, "step": 3415 }, { "epoch": 1.8304064224786754, "grad_norm": 9.58276576176987, "learning_rate": 3.956481126078927e-07, "logits/chosen": -0.11591336876153946, "logits/rejected": -0.014336993917822838, "logps/chosen": -1.2792669534683228, "logps/rejected": -1.5205464363098145, "loss": 1.6288, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2792669534683228, "rewards/margins": 0.24127936363220215, "rewards/rejected": -1.5205464363098145, "step": 3420 }, { "epoch": 1.8330824552600768, "grad_norm": 6.151300442369192, "learning_rate": 3.941255143334937e-07, "logits/chosen": -0.22207434475421906, "logits/rejected": -0.18767404556274414, "logps/chosen": -1.242661714553833, "logps/rejected": -1.417602777481079, "loss": 1.6019, "rewards/accuracies": 0.5625, "rewards/chosen": -1.242661714553833, "rewards/margins": 0.17494121193885803, "rewards/rejected": -1.417602777481079, "step": 3425 }, { "epoch": 1.8357584880414786, "grad_norm": 10.485462818829816, "learning_rate": 3.9260394325381895e-07, "logits/chosen": -0.17935669422149658, "logits/rejected": -0.07608787715435028, "logps/chosen": -1.3051313161849976, "logps/rejected": -1.4965819120407104, "loss": 1.637, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3051313161849976, "rewards/margins": 0.19145064055919647, "rewards/rejected": -1.4965819120407104, "step": 3430 }, { "epoch": 1.83843452082288, "grad_norm": 10.217071873269104, "learning_rate": 3.9108341413115784e-07, "logits/chosen": -0.19460782408714294, "logits/rejected": -0.12509635090827942, "logps/chosen": -1.2595460414886475, "logps/rejected": -1.4745066165924072, "loss": 1.5893, "rewards/accuracies": 0.625, "rewards/chosen": -1.2595460414886475, "rewards/margins": 0.21496033668518066, "rewards/rejected": -1.4745066165924072, "step": 3435 }, { "epoch": 1.8411105536042816, "grad_norm": 10.063470604313583, "learning_rate": 3.895639417176905e-07, "logits/chosen": -0.2412361204624176, "logits/rejected": -0.18948474526405334, "logps/chosen": -1.1810823678970337, "logps/rejected": -1.4140219688415527, "loss": 1.5661, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1810823678970337, "rewards/margins": 0.2329397201538086, "rewards/rejected": -1.4140219688415527, "step": 3440 }, { "epoch": 1.8437865863856833, "grad_norm": 8.01386847459897, "learning_rate": 3.8804554075534497e-07, "logits/chosen": -0.23835010826587677, "logits/rejected": -0.043347861617803574, "logps/chosen": -1.2297178506851196, "logps/rejected": -1.407577395439148, "loss": 1.5735, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2297178506851196, "rewards/margins": 0.17785969376564026, "rewards/rejected": -1.407577395439148, "step": 3445 }, { "epoch": 1.8464626191670848, "grad_norm": 10.227781776336018, "learning_rate": 3.8652822597565403e-07, "logits/chosen": -0.30860528349876404, "logits/rejected": -0.15392982959747314, "logps/chosen": -1.250962257385254, "logps/rejected": -1.5010929107666016, "loss": 1.585, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.250962257385254, "rewards/margins": 0.25013068318367004, "rewards/rejected": -1.5010929107666016, "step": 3450 }, { "epoch": 1.8491386519484863, "grad_norm": 6.974518358323705, "learning_rate": 3.850120120996123e-07, "logits/chosen": -0.18968996405601501, "logits/rejected": -0.06573452800512314, "logps/chosen": -1.4123960733413696, "logps/rejected": -1.6055738925933838, "loss": 1.7374, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.4123960733413696, "rewards/margins": 0.19317783415317535, "rewards/rejected": -1.6055738925933838, "step": 3455 }, { "epoch": 1.851814684729888, "grad_norm": 10.776685643361438, "learning_rate": 3.8349691383753356e-07, "logits/chosen": -0.06698723137378693, "logits/rejected": 0.0407029464840889, "logps/chosen": -1.2353590726852417, "logps/rejected": -1.4258449077606201, "loss": 1.6391, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2353590726852417, "rewards/margins": 0.19048604369163513, "rewards/rejected": -1.4258449077606201, "step": 3460 }, { "epoch": 1.8544907175112895, "grad_norm": 6.886761549669779, "learning_rate": 3.819829458889078e-07, "logits/chosen": -0.2288968861103058, "logits/rejected": -0.11740916967391968, "logps/chosen": -1.2035908699035645, "logps/rejected": -1.3677371740341187, "loss": 1.5767, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2035908699035645, "rewards/margins": 0.16414625942707062, "rewards/rejected": -1.3677371740341187, "step": 3465 }, { "epoch": 1.857166750292691, "grad_norm": 7.653110287741205, "learning_rate": 3.804701229422585e-07, "logits/chosen": -0.24052855372428894, "logits/rejected": -0.15510748326778412, "logps/chosen": -1.3313173055648804, "logps/rejected": -1.4727556705474854, "loss": 1.6644, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3313173055648804, "rewards/margins": 0.1414383500814438, "rewards/rejected": -1.4727556705474854, "step": 3470 }, { "epoch": 1.8598427830740927, "grad_norm": 6.938993849604643, "learning_rate": 3.789584596750007e-07, "logits/chosen": -0.2517121434211731, "logits/rejected": -0.1962416172027588, "logps/chosen": -1.2569118738174438, "logps/rejected": -1.4523240327835083, "loss": 1.6172, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2569118738174438, "rewards/margins": 0.19541211426258087, "rewards/rejected": -1.4523240327835083, "step": 3475 }, { "epoch": 1.8625188158554944, "grad_norm": 7.922434750323968, "learning_rate": 3.77447970753298e-07, "logits/chosen": -0.1458994597196579, "logits/rejected": -0.12055318057537079, "logps/chosen": -1.2811120748519897, "logps/rejected": -1.4992064237594604, "loss": 1.6195, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2811120748519897, "rewards/margins": 0.2180943489074707, "rewards/rejected": -1.4992064237594604, "step": 3480 }, { "epoch": 1.8651948486368957, "grad_norm": 10.217164584913801, "learning_rate": 3.7593867083192057e-07, "logits/chosen": -0.16993644833564758, "logits/rejected": -0.08232876658439636, "logps/chosen": -1.2428160905838013, "logps/rejected": -1.4211010932922363, "loss": 1.6092, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2428160905838013, "rewards/margins": 0.17828497290611267, "rewards/rejected": -1.4211010932922363, "step": 3485 }, { "epoch": 1.8678708814182974, "grad_norm": 6.790309217738528, "learning_rate": 3.7443057455410276e-07, "logits/chosen": -0.15006773173809052, "logits/rejected": -0.043843645602464676, "logps/chosen": -1.300969123840332, "logps/rejected": -1.3989123106002808, "loss": 1.6606, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.300969123840332, "rewards/margins": 0.09794305264949799, "rewards/rejected": -1.3989123106002808, "step": 3490 }, { "epoch": 1.870546914199699, "grad_norm": 8.197605176941899, "learning_rate": 3.7292369655140145e-07, "logits/chosen": -0.2248447686433792, "logits/rejected": -0.07948695868253708, "logps/chosen": -1.252953290939331, "logps/rejected": -1.4324767589569092, "loss": 1.5976, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.252953290939331, "rewards/margins": 0.1795234978199005, "rewards/rejected": -1.4324767589569092, "step": 3495 }, { "epoch": 1.8732229469811004, "grad_norm": 8.118952561685207, "learning_rate": 3.714180514435534e-07, "logits/chosen": -0.15942063927650452, "logits/rejected": -0.023639973253011703, "logps/chosen": -1.298370599746704, "logps/rejected": -1.5295395851135254, "loss": 1.6139, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.298370599746704, "rewards/margins": 0.2311691790819168, "rewards/rejected": -1.5295395851135254, "step": 3500 }, { "epoch": 1.875898979762502, "grad_norm": 9.018634962880915, "learning_rate": 3.6991365383833426e-07, "logits/chosen": -0.15019235014915466, "logits/rejected": -0.05135111138224602, "logps/chosen": -1.2949392795562744, "logps/rejected": -1.4940239191055298, "loss": 1.6456, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2949392795562744, "rewards/margins": 0.19908465445041656, "rewards/rejected": -1.4940239191055298, "step": 3505 }, { "epoch": 1.8785750125439038, "grad_norm": 9.397012119740829, "learning_rate": 3.684105183314162e-07, "logits/chosen": -0.19316525757312775, "logits/rejected": -0.1316845864057541, "logps/chosen": -1.2419188022613525, "logps/rejected": -1.3934087753295898, "loss": 1.5919, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2419188022613525, "rewards/margins": 0.15148988366127014, "rewards/rejected": -1.3934087753295898, "step": 3510 }, { "epoch": 1.881251045325305, "grad_norm": 9.365950918947213, "learning_rate": 3.669086595062263e-07, "logits/chosen": -0.16677063703536987, "logits/rejected": 0.004392133094370365, "logps/chosen": -1.2826802730560303, "logps/rejected": -1.4858967065811157, "loss": 1.6239, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2826802730560303, "rewards/margins": 0.20321647822856903, "rewards/rejected": -1.4858967065811157, "step": 3515 }, { "epoch": 1.8839270781067068, "grad_norm": 8.981100395468093, "learning_rate": 3.654080919338056e-07, "logits/chosen": -0.2394132912158966, "logits/rejected": -0.13284596800804138, "logps/chosen": -1.2914131879806519, "logps/rejected": -1.4775323867797852, "loss": 1.6358, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2914131879806519, "rewards/margins": 0.18611934781074524, "rewards/rejected": -1.4775323867797852, "step": 3520 }, { "epoch": 1.8866031108881085, "grad_norm": 7.288596464022811, "learning_rate": 3.639088301726673e-07, "logits/chosen": -0.1558699905872345, "logits/rejected": 0.013392647728323936, "logps/chosen": -1.255486249923706, "logps/rejected": -1.4498775005340576, "loss": 1.6129, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.255486249923706, "rewards/margins": 0.19439134001731873, "rewards/rejected": -1.4498775005340576, "step": 3525 }, { "epoch": 1.88927914366951, "grad_norm": 11.646149377669992, "learning_rate": 3.624108887686556e-07, "logits/chosen": -0.14714086055755615, "logits/rejected": -0.0844908356666565, "logps/chosen": -1.2383390665054321, "logps/rejected": -1.4483082294464111, "loss": 1.5954, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2383390665054321, "rewards/margins": 0.2099691927433014, "rewards/rejected": -1.4483082294464111, "step": 3530 }, { "epoch": 1.8919551764509115, "grad_norm": 6.508255128650294, "learning_rate": 3.6091428225480433e-07, "logits/chosen": -0.23644837737083435, "logits/rejected": -0.12065120041370392, "logps/chosen": -1.2052524089813232, "logps/rejected": -1.347082257270813, "loss": 1.6029, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2052524089813232, "rewards/margins": 0.14182987809181213, "rewards/rejected": -1.347082257270813, "step": 3535 }, { "epoch": 1.8946312092323132, "grad_norm": 10.987640646933064, "learning_rate": 3.5941902515119674e-07, "logits/chosen": -0.17603799700737, "logits/rejected": 0.02899671159684658, "logps/chosen": -1.2338905334472656, "logps/rejected": -1.4330108165740967, "loss": 1.5923, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2338905334472656, "rewards/margins": 0.19912008941173553, "rewards/rejected": -1.4330108165740967, "step": 3540 }, { "epoch": 1.8973072420137147, "grad_norm": 9.210600086405625, "learning_rate": 3.5792513196482373e-07, "logits/chosen": -0.3058917820453644, "logits/rejected": -0.07458896934986115, "logps/chosen": -1.236205816268921, "logps/rejected": -1.3533177375793457, "loss": 1.6074, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.236205816268921, "rewards/margins": 0.11711195856332779, "rewards/rejected": -1.3533177375793457, "step": 3545 }, { "epoch": 1.8999832747951162, "grad_norm": 7.006254201358214, "learning_rate": 3.5643261718944346e-07, "logits/chosen": -0.15428760647773743, "logits/rejected": -0.07689885795116425, "logps/chosen": -1.230529546737671, "logps/rejected": -1.3739771842956543, "loss": 1.5886, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.230529546737671, "rewards/margins": 0.14344754815101624, "rewards/rejected": -1.3739771842956543, "step": 3550 }, { "epoch": 1.902659307576518, "grad_norm": 6.0580380854265705, "learning_rate": 3.5494149530544087e-07, "logits/chosen": -0.27520883083343506, "logits/rejected": -0.15692627429962158, "logps/chosen": -1.1899023056030273, "logps/rejected": -1.3904666900634766, "loss": 1.5944, "rewards/accuracies": 0.59375, "rewards/chosen": -1.1899023056030273, "rewards/margins": 0.2005644589662552, "rewards/rejected": -1.3904666900634766, "step": 3555 }, { "epoch": 1.9053353403579194, "grad_norm": 11.666799658627408, "learning_rate": 3.534517807796871e-07, "logits/chosen": -0.16600708663463593, "logits/rejected": -0.10134400427341461, "logps/chosen": -1.282378911972046, "logps/rejected": -1.4656873941421509, "loss": 1.6334, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -1.282378911972046, "rewards/margins": 0.18330873548984528, "rewards/rejected": -1.4656873941421509, "step": 3560 }, { "epoch": 1.908011373139321, "grad_norm": 5.999143019086315, "learning_rate": 3.519634880653988e-07, "logits/chosen": -0.18405933678150177, "logits/rejected": -0.14047543704509735, "logps/chosen": -1.2129871845245361, "logps/rejected": -1.4327234029769897, "loss": 1.5922, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2129871845245361, "rewards/margins": 0.21973614394664764, "rewards/rejected": -1.4327234029769897, "step": 3565 }, { "epoch": 1.9106874059207226, "grad_norm": 7.594276353635611, "learning_rate": 3.504766316019987e-07, "logits/chosen": -0.2190113514661789, "logits/rejected": -0.10907818377017975, "logps/chosen": -1.241443157196045, "logps/rejected": -1.4419244527816772, "loss": 1.605, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.241443157196045, "rewards/margins": 0.20048122107982635, "rewards/rejected": -1.4419244527816772, "step": 3570 }, { "epoch": 1.913363438702124, "grad_norm": 7.467008251101331, "learning_rate": 3.489912258149745e-07, "logits/chosen": -0.11534838378429413, "logits/rejected": -0.011506294831633568, "logps/chosen": -1.2071187496185303, "logps/rejected": -1.4129838943481445, "loss": 1.5861, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2071187496185303, "rewards/margins": 0.20586495101451874, "rewards/rejected": -1.4129838943481445, "step": 3575 }, { "epoch": 1.9160394714835256, "grad_norm": 8.656615666522297, "learning_rate": 3.475072851157397e-07, "logits/chosen": -0.17539773881435394, "logits/rejected": -0.13611510396003723, "logps/chosen": -1.224860668182373, "logps/rejected": -1.4651345014572144, "loss": 1.578, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.224860668182373, "rewards/margins": 0.24027390778064728, "rewards/rejected": -1.4651345014572144, "step": 3580 }, { "epoch": 1.9187155042649273, "grad_norm": 8.01258193139656, "learning_rate": 3.460248239014936e-07, "logits/chosen": -0.10515443235635757, "logits/rejected": -0.06597565114498138, "logps/chosen": -1.3152985572814941, "logps/rejected": -1.4554119110107422, "loss": 1.665, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3152985572814941, "rewards/margins": 0.14011338353157043, "rewards/rejected": -1.4554119110107422, "step": 3585 }, { "epoch": 1.9213915370463288, "grad_norm": 7.927567069296681, "learning_rate": 3.4454385655508134e-07, "logits/chosen": -0.14241893589496613, "logits/rejected": -0.07710736989974976, "logps/chosen": -1.2774500846862793, "logps/rejected": -1.4123210906982422, "loss": 1.6315, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2774500846862793, "rewards/margins": 0.13487093150615692, "rewards/rejected": -1.4123210906982422, "step": 3590 }, { "epoch": 1.9240675698277303, "grad_norm": 6.7613467796414035, "learning_rate": 3.4306439744485447e-07, "logits/chosen": -0.26878923177719116, "logits/rejected": -0.08853518962860107, "logps/chosen": -1.2294676303863525, "logps/rejected": -1.4297891855239868, "loss": 1.5832, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2294676303863525, "rewards/margins": 0.2003215253353119, "rewards/rejected": -1.4297891855239868, "step": 3595 }, { "epoch": 1.926743602609132, "grad_norm": 10.040042117098851, "learning_rate": 3.415864609245322e-07, "logits/chosen": -0.12516078352928162, "logits/rejected": 0.03827281668782234, "logps/chosen": -1.2093441486358643, "logps/rejected": -1.4573616981506348, "loss": 1.5646, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2093441486358643, "rewards/margins": 0.24801769852638245, "rewards/rejected": -1.4573616981506348, "step": 3600 }, { "epoch": 1.926743602609132, "eval_logits/chosen": 0.0770111232995987, "eval_logits/rejected": 0.1466367095708847, "eval_logps/chosen": -1.2986595630645752, "eval_logps/rejected": -1.477584719657898, "eval_loss": 1.6478548049926758, "eval_rewards/accuracies": 0.5682492852210999, "eval_rewards/chosen": -1.2986595630645752, "eval_rewards/margins": 0.17892497777938843, "eval_rewards/rejected": -1.477584719657898, "eval_runtime": 40.6731, "eval_samples_per_second": 33.069, "eval_steps_per_second": 8.286, "step": 3600 }, { "epoch": 1.9294196353905335, "grad_norm": 7.840330360879674, "learning_rate": 3.401100613330605e-07, "logits/chosen": -0.22566242516040802, "logits/rejected": -0.19398054480552673, "logps/chosen": -1.2587788105010986, "logps/rejected": -1.3885822296142578, "loss": 1.6406, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2587788105010986, "rewards/margins": 0.1298035830259323, "rewards/rejected": -1.3885822296142578, "step": 3605 }, { "epoch": 1.932095668171935, "grad_norm": 6.400210435776477, "learning_rate": 3.3863521299447514e-07, "logits/chosen": -0.15038132667541504, "logits/rejected": -0.04821936413645744, "logps/chosen": -1.2305161952972412, "logps/rejected": -1.4341785907745361, "loss": 1.5762, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2305161952972412, "rewards/margins": 0.20366230607032776, "rewards/rejected": -1.4341785907745361, "step": 3610 }, { "epoch": 1.9347717009533367, "grad_norm": 6.307705093571466, "learning_rate": 3.371619302177609e-07, "logits/chosen": -0.0919400304555893, "logits/rejected": 0.0020879185758531094, "logps/chosen": -1.3276575803756714, "logps/rejected": -1.436106562614441, "loss": 1.6671, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3276575803756714, "rewards/margins": 0.1084490641951561, "rewards/rejected": -1.436106562614441, "step": 3615 }, { "epoch": 1.9374477337347382, "grad_norm": 10.19933447906074, "learning_rate": 3.3569022729671393e-07, "logits/chosen": -0.14851398766040802, "logits/rejected": -0.09383167326450348, "logps/chosen": -1.2694073915481567, "logps/rejected": -1.383233904838562, "loss": 1.6595, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2694073915481567, "rewards/margins": 0.11382659524679184, "rewards/rejected": -1.383233904838562, "step": 3620 }, { "epoch": 1.9401237665161397, "grad_norm": 8.719753915697652, "learning_rate": 3.342201185098024e-07, "logits/chosen": -0.10923445224761963, "logits/rejected": -0.09326710551977158, "logps/chosen": -1.2533637285232544, "logps/rejected": -1.4721083641052246, "loss": 1.601, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2533637285232544, "rewards/margins": 0.21874471008777618, "rewards/rejected": -1.4721083641052246, "step": 3625 }, { "epoch": 1.9427997992975414, "grad_norm": 8.66556904039224, "learning_rate": 3.3275161812002807e-07, "logits/chosen": -0.17842338979244232, "logits/rejected": -0.13167932629585266, "logps/chosen": -1.2734184265136719, "logps/rejected": -1.4388821125030518, "loss": 1.6404, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2734184265136719, "rewards/margins": 0.16546377539634705, "rewards/rejected": -1.4388821125030518, "step": 3630 }, { "epoch": 1.945475832078943, "grad_norm": 10.758803567850421, "learning_rate": 3.312847403747883e-07, "logits/chosen": -0.20899248123168945, "logits/rejected": -0.13900670409202576, "logps/chosen": -1.2907582521438599, "logps/rejected": -1.4487711191177368, "loss": 1.6511, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2907582521438599, "rewards/margins": 0.1580127775669098, "rewards/rejected": -1.4487711191177368, "step": 3635 }, { "epoch": 1.9481518648603444, "grad_norm": 7.904694182249486, "learning_rate": 3.2981949950573733e-07, "logits/chosen": -0.16778028011322021, "logits/rejected": -0.06348295509815216, "logps/chosen": -1.3668639659881592, "logps/rejected": -1.4366734027862549, "loss": 1.7136, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3668639659881592, "rewards/margins": 0.06980935484170914, "rewards/rejected": -1.4366734027862549, "step": 3640 }, { "epoch": 1.9508278976417461, "grad_norm": 7.477827845292579, "learning_rate": 3.283559097286486e-07, "logits/chosen": -0.1922161877155304, "logits/rejected": -0.08164320886135101, "logps/chosen": -1.3601343631744385, "logps/rejected": -1.4841011762619019, "loss": 1.6988, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3601343631744385, "rewards/margins": 0.12396688759326935, "rewards/rejected": -1.4841011762619019, "step": 3645 }, { "epoch": 1.9535039304231478, "grad_norm": 7.4573408788650175, "learning_rate": 3.268939852432765e-07, "logits/chosen": -0.2014189213514328, "logits/rejected": -0.11250327527523041, "logps/chosen": -1.248978853225708, "logps/rejected": -1.386046290397644, "loss": 1.613, "rewards/accuracies": 0.5625, "rewards/chosen": -1.248978853225708, "rewards/margins": 0.13706740736961365, "rewards/rejected": -1.386046290397644, "step": 3650 }, { "epoch": 1.9561799632045491, "grad_norm": 11.245097511626474, "learning_rate": 3.254337402332187e-07, "logits/chosen": -0.1812329888343811, "logits/rejected": -0.060154665261507034, "logps/chosen": -1.3011744022369385, "logps/rejected": -1.4550087451934814, "loss": 1.6674, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.3011744022369385, "rewards/margins": 0.15383443236351013, "rewards/rejected": -1.4550087451934814, "step": 3655 }, { "epoch": 1.9588559959859508, "grad_norm": 11.108946193918412, "learning_rate": 3.239751888657788e-07, "logits/chosen": -0.21206536889076233, "logits/rejected": -0.10278163105249405, "logps/chosen": -1.2035177946090698, "logps/rejected": -1.3656055927276611, "loss": 1.5971, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2035177946090698, "rewards/margins": 0.16208767890930176, "rewards/rejected": -1.3656055927276611, "step": 3660 }, { "epoch": 1.9615320287673526, "grad_norm": 8.840385185457135, "learning_rate": 3.2251834529182856e-07, "logits/chosen": -0.14388488233089447, "logits/rejected": -0.03836951404809952, "logps/chosen": -1.2217673063278198, "logps/rejected": -1.3933287858963013, "loss": 1.6081, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2217673063278198, "rewards/margins": 0.17156127095222473, "rewards/rejected": -1.3933287858963013, "step": 3665 }, { "epoch": 1.9642080615487538, "grad_norm": 6.492069858070431, "learning_rate": 3.2106322364567075e-07, "logits/chosen": -0.21870478987693787, "logits/rejected": -0.09473247826099396, "logps/chosen": -1.2585182189941406, "logps/rejected": -1.4683822393417358, "loss": 1.6007, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2585182189941406, "rewards/margins": 0.20986399054527283, "rewards/rejected": -1.4683822393417358, "step": 3670 }, { "epoch": 1.9668840943301555, "grad_norm": 7.383800873219924, "learning_rate": 3.1960983804490183e-07, "logits/chosen": -0.16788841784000397, "logits/rejected": -0.04624394327402115, "logps/chosen": -1.2878868579864502, "logps/rejected": -1.57819402217865, "loss": 1.606, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2878868579864502, "rewards/margins": 0.29030725359916687, "rewards/rejected": -1.57819402217865, "step": 3675 }, { "epoch": 1.9695601271115573, "grad_norm": 8.130005522098017, "learning_rate": 3.1815820259027537e-07, "logits/chosen": -0.18945178389549255, "logits/rejected": -0.09167564660310745, "logps/chosen": -1.138102412223816, "logps/rejected": -1.3502238988876343, "loss": 1.5201, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.138102412223816, "rewards/margins": 0.21212148666381836, "rewards/rejected": -1.3502238988876343, "step": 3680 }, { "epoch": 1.9722361598929585, "grad_norm": 8.7096371713661, "learning_rate": 3.16708331365565e-07, "logits/chosen": -0.2038750946521759, "logits/rejected": -0.1391317993402481, "logps/chosen": -1.2368247509002686, "logps/rejected": -1.4675143957138062, "loss": 1.5714, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2368247509002686, "rewards/margins": 0.23068973422050476, "rewards/rejected": -1.4675143957138062, "step": 3685 }, { "epoch": 1.9749121926743602, "grad_norm": 7.496384500443867, "learning_rate": 3.152602384374275e-07, "logits/chosen": -0.17525169253349304, "logits/rejected": -0.04767988994717598, "logps/chosen": -1.2744166851043701, "logps/rejected": -1.4640617370605469, "loss": 1.6311, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2744166851043701, "rewards/margins": 0.18964512646198273, "rewards/rejected": -1.4640617370605469, "step": 3690 }, { "epoch": 1.977588225455762, "grad_norm": 7.29091326836066, "learning_rate": 3.1381393785526697e-07, "logits/chosen": -0.12062673270702362, "logits/rejected": -0.09473898261785507, "logps/chosen": -1.3177236318588257, "logps/rejected": -1.549801230430603, "loss": 1.6315, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.3177236318588257, "rewards/margins": 0.2320774793624878, "rewards/rejected": -1.549801230430603, "step": 3695 }, { "epoch": 1.9802642582371635, "grad_norm": 8.520817560244373, "learning_rate": 3.123694436510979e-07, "logits/chosen": -0.12591895461082458, "logits/rejected": -0.035199157893657684, "logps/chosen": -1.2405253648757935, "logps/rejected": -1.4357420206069946, "loss": 1.5927, "rewards/accuracies": 0.625, "rewards/chosen": -1.2405253648757935, "rewards/margins": 0.1952165812253952, "rewards/rejected": -1.4357420206069946, "step": 3700 }, { "epoch": 1.982940291018565, "grad_norm": 8.825227178659631, "learning_rate": 3.1092676983940946e-07, "logits/chosen": -0.16672348976135254, "logits/rejected": -0.10381430387496948, "logps/chosen": -1.290474772453308, "logps/rejected": -1.491180181503296, "loss": 1.6401, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.290474772453308, "rewards/margins": 0.20070545375347137, "rewards/rejected": -1.491180181503296, "step": 3705 }, { "epoch": 1.9856163237999667, "grad_norm": 7.8779454890869784, "learning_rate": 3.094859304170293e-07, "logits/chosen": -0.04672155901789665, "logits/rejected": 0.0022973031736910343, "logps/chosen": -1.2978581190109253, "logps/rejected": -1.4588404893875122, "loss": 1.6502, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2978581190109253, "rewards/margins": 0.16098220646381378, "rewards/rejected": -1.4588404893875122, "step": 3710 }, { "epoch": 1.9882923565813682, "grad_norm": 5.76963985896493, "learning_rate": 3.0804693936298795e-07, "logits/chosen": -0.12356910854578018, "logits/rejected": -0.05897799879312515, "logps/chosen": -1.2731387615203857, "logps/rejected": -1.4750876426696777, "loss": 1.6181, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2731387615203857, "rewards/margins": 0.201948881149292, "rewards/rejected": -1.4750876426696777, "step": 3715 }, { "epoch": 1.9909683893627697, "grad_norm": 6.953061439873837, "learning_rate": 3.066098106383826e-07, "logits/chosen": -0.14972397685050964, "logits/rejected": -0.08427880704402924, "logps/chosen": -1.2518134117126465, "logps/rejected": -1.4181315898895264, "loss": 1.6279, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2518134117126465, "rewards/margins": 0.16631817817687988, "rewards/rejected": -1.4181315898895264, "step": 3720 }, { "epoch": 1.9936444221441714, "grad_norm": 6.830614461497077, "learning_rate": 3.0517455818624263e-07, "logits/chosen": -0.22917135059833527, "logits/rejected": -0.14272315800189972, "logps/chosen": -1.255526065826416, "logps/rejected": -1.423462152481079, "loss": 1.6155, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.255526065826416, "rewards/margins": 0.16793592274188995, "rewards/rejected": -1.423462152481079, "step": 3725 }, { "epoch": 1.9963204549255729, "grad_norm": 7.49871108289989, "learning_rate": 3.037411959313936e-07, "logits/chosen": -0.1212211400270462, "logits/rejected": -0.016563747078180313, "logps/chosen": -1.218724012374878, "logps/rejected": -1.4318233728408813, "loss": 1.5747, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.218724012374878, "rewards/margins": 0.21309927105903625, "rewards/rejected": -1.4318233728408813, "step": 3730 }, { "epoch": 1.9989964877069744, "grad_norm": 7.409246277510923, "learning_rate": 3.023097377803224e-07, "logits/chosen": -0.07228492200374603, "logits/rejected": -0.003942572977393866, "logps/chosen": -1.340967059135437, "logps/rejected": -1.43936026096344, "loss": 1.6969, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.340967059135437, "rewards/margins": 0.09839321672916412, "rewards/rejected": -1.43936026096344, "step": 3735 }, { "epoch": 2.001672520488376, "grad_norm": 6.988169196963502, "learning_rate": 3.008801976210423e-07, "logits/chosen": -0.0938621237874031, "logits/rejected": -0.05825387313961983, "logps/chosen": -1.3282005786895752, "logps/rejected": -1.4351487159729004, "loss": 1.6719, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.3282005786895752, "rewards/margins": 0.10694797337055206, "rewards/rejected": -1.4351487159729004, "step": 3740 }, { "epoch": 2.0043485532697773, "grad_norm": 6.621951457816368, "learning_rate": 2.994525893229581e-07, "logits/chosen": -0.17387568950653076, "logits/rejected": -0.09450852870941162, "logps/chosen": -1.275235891342163, "logps/rejected": -1.4434269666671753, "loss": 1.6201, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.275235891342163, "rewards/margins": 0.16819104552268982, "rewards/rejected": -1.4434269666671753, "step": 3745 }, { "epoch": 2.007024586051179, "grad_norm": 6.744276302001362, "learning_rate": 2.98026926736732e-07, "logits/chosen": -0.21224412322044373, "logits/rejected": -0.14395299553871155, "logps/chosen": -1.1820372343063354, "logps/rejected": -1.4528343677520752, "loss": 1.5254, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1820372343063354, "rewards/margins": 0.27079716324806213, "rewards/rejected": -1.4528343677520752, "step": 3750 }, { "epoch": 2.0097006188325808, "grad_norm": 7.532463308561457, "learning_rate": 2.9660322369414846e-07, "logits/chosen": -0.17556189000606537, "logits/rejected": -0.09353695809841156, "logps/chosen": -1.2044193744659424, "logps/rejected": -1.5262267589569092, "loss": 1.5395, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2044193744659424, "rewards/margins": 0.321807324886322, "rewards/rejected": -1.5262267589569092, "step": 3755 }, { "epoch": 2.0123766516139825, "grad_norm": 7.32971584019061, "learning_rate": 2.9518149400798063e-07, "logits/chosen": -0.24406211078166962, "logits/rejected": -0.23254680633544922, "logps/chosen": -1.243602991104126, "logps/rejected": -1.4762613773345947, "loss": 1.5733, "rewards/accuracies": 0.625, "rewards/chosen": -1.243602991104126, "rewards/margins": 0.23265841603279114, "rewards/rejected": -1.4762613773345947, "step": 3760 }, { "epoch": 2.0150526843953838, "grad_norm": 8.947032815941157, "learning_rate": 2.9376175147185633e-07, "logits/chosen": -0.10410241782665253, "logits/rejected": 0.04200471192598343, "logps/chosen": -1.2236241102218628, "logps/rejected": -1.4989502429962158, "loss": 1.5573, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2236241102218628, "rewards/margins": 0.2753261923789978, "rewards/rejected": -1.4989502429962158, "step": 3765 }, { "epoch": 2.0177287171767855, "grad_norm": 8.738730226260707, "learning_rate": 2.9234400986012376e-07, "logits/chosen": -0.2576754689216614, "logits/rejected": -0.12882332503795624, "logps/chosen": -1.1754076480865479, "logps/rejected": -1.583466649055481, "loss": 1.5065, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.1754076480865479, "rewards/margins": 0.4080590605735779, "rewards/rejected": -1.583466649055481, "step": 3770 }, { "epoch": 2.020404749958187, "grad_norm": 7.85080913279154, "learning_rate": 2.9092828292771817e-07, "logits/chosen": -0.20311729609966278, "logits/rejected": -0.15802323818206787, "logps/chosen": -1.229494571685791, "logps/rejected": -1.4390629529953003, "loss": 1.5964, "rewards/accuracies": 0.5, "rewards/chosen": -1.229494571685791, "rewards/margins": 0.20956823229789734, "rewards/rejected": -1.4390629529953003, "step": 3775 }, { "epoch": 2.0230807827395885, "grad_norm": 7.165995730205779, "learning_rate": 2.8951458441002875e-07, "logits/chosen": -0.13184405863285065, "logits/rejected": -0.1076413244009018, "logps/chosen": -1.2524865865707397, "logps/rejected": -1.4842063188552856, "loss": 1.6219, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2524865865707397, "rewards/margins": 0.2317197024822235, "rewards/rejected": -1.4842063188552856, "step": 3780 }, { "epoch": 2.02575681552099, "grad_norm": 6.470693358098353, "learning_rate": 2.881029280227643e-07, "logits/chosen": -0.18296106159687042, "logits/rejected": -0.0861835852265358, "logps/chosen": -1.2711610794067383, "logps/rejected": -1.5663596391677856, "loss": 1.5752, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2711610794067383, "rewards/margins": 0.295198529958725, "rewards/rejected": -1.5663596391677856, "step": 3785 }, { "epoch": 2.028432848302392, "grad_norm": 6.194312588328929, "learning_rate": 2.8669332746182177e-07, "logits/chosen": -0.2417146861553192, "logits/rejected": -0.0933353453874588, "logps/chosen": -1.2275288105010986, "logps/rejected": -1.4598970413208008, "loss": 1.556, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2275288105010986, "rewards/margins": 0.23236839473247528, "rewards/rejected": -1.4598970413208008, "step": 3790 }, { "epoch": 2.031108881083793, "grad_norm": 6.4618508390800615, "learning_rate": 2.8528579640315156e-07, "logits/chosen": -0.1630270779132843, "logits/rejected": -0.14098826050758362, "logps/chosen": -1.2046821117401123, "logps/rejected": -1.3948853015899658, "loss": 1.5704, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2046821117401123, "rewards/margins": 0.19020314514636993, "rewards/rejected": -1.3948853015899658, "step": 3795 }, { "epoch": 2.033784913865195, "grad_norm": 8.434133849723032, "learning_rate": 2.8388034850262646e-07, "logits/chosen": -0.15991130471229553, "logits/rejected": -0.059218693524599075, "logps/chosen": -1.2866146564483643, "logps/rejected": -1.5523066520690918, "loss": 1.6305, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2866146564483643, "rewards/margins": 0.2656919062137604, "rewards/rejected": -1.5523066520690918, "step": 3800 }, { "epoch": 2.0364609466465966, "grad_norm": 11.017392314905319, "learning_rate": 2.824769973959079e-07, "logits/chosen": -0.12081266939640045, "logits/rejected": -0.023985737934708595, "logps/chosen": -1.1901729106903076, "logps/rejected": -1.4317840337753296, "loss": 1.5337, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1901729106903076, "rewards/margins": 0.24161119759082794, "rewards/rejected": -1.4317840337753296, "step": 3805 }, { "epoch": 2.039136979427998, "grad_norm": 7.846576772195456, "learning_rate": 2.81075756698315e-07, "logits/chosen": -0.047540389001369476, "logits/rejected": 0.03174520656466484, "logps/chosen": -1.2199788093566895, "logps/rejected": -1.4722950458526611, "loss": 1.5653, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2199788093566895, "rewards/margins": 0.25231608748435974, "rewards/rejected": -1.4722950458526611, "step": 3810 }, { "epoch": 2.0418130122093996, "grad_norm": 7.380796979122265, "learning_rate": 2.7967664000469035e-07, "logits/chosen": -0.25073981285095215, "logits/rejected": -0.14145082235336304, "logps/chosen": -1.259158968925476, "logps/rejected": -1.4133001565933228, "loss": 1.6182, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.259158968925476, "rewards/margins": 0.15414123237133026, "rewards/rejected": -1.4133001565933228, "step": 3815 }, { "epoch": 2.0444890449908013, "grad_norm": 9.109162210572233, "learning_rate": 2.7827966088927095e-07, "logits/chosen": -0.23392069339752197, "logits/rejected": -0.06347467750310898, "logps/chosen": -1.3198875188827515, "logps/rejected": -1.4128139019012451, "loss": 1.6813, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -1.3198875188827515, "rewards/margins": 0.09292645007371902, "rewards/rejected": -1.4128139019012451, "step": 3820 }, { "epoch": 2.0471650777722026, "grad_norm": 8.398550301698064, "learning_rate": 2.768848329055538e-07, "logits/chosen": -0.20006170868873596, "logits/rejected": -0.10423286259174347, "logps/chosen": -1.228200078010559, "logps/rejected": -1.374854326248169, "loss": 1.5936, "rewards/accuracies": 0.59375, "rewards/chosen": -1.228200078010559, "rewards/margins": 0.14665424823760986, "rewards/rejected": -1.374854326248169, "step": 3825 }, { "epoch": 2.0498411105536043, "grad_norm": 8.515263228939604, "learning_rate": 2.7549216958616657e-07, "logits/chosen": -0.24098920822143555, "logits/rejected": -0.11884365230798721, "logps/chosen": -1.2912288904190063, "logps/rejected": -1.5368410348892212, "loss": 1.6143, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2912288904190063, "rewards/margins": 0.24561218917369843, "rewards/rejected": -1.5368410348892212, "step": 3830 }, { "epoch": 2.052517143335006, "grad_norm": 7.409912246633359, "learning_rate": 2.741016844427344e-07, "logits/chosen": -0.15882329642772675, "logits/rejected": -0.04877109453082085, "logps/chosen": -1.2781291007995605, "logps/rejected": -1.4769856929779053, "loss": 1.6229, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2781291007995605, "rewards/margins": 0.1988566666841507, "rewards/rejected": -1.4769856929779053, "step": 3835 }, { "epoch": 2.0551931761164073, "grad_norm": 8.036877785952855, "learning_rate": 2.7271339096575073e-07, "logits/chosen": -0.1380457580089569, "logits/rejected": -0.06133418530225754, "logps/chosen": -1.1802645921707153, "logps/rejected": -1.5075302124023438, "loss": 1.5241, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1802645921707153, "rewards/margins": 0.3272656500339508, "rewards/rejected": -1.5075302124023438, "step": 3840 }, { "epoch": 2.057869208897809, "grad_norm": 7.1543336062609955, "learning_rate": 2.713273026244446e-07, "logits/chosen": -0.27739790081977844, "logits/rejected": -0.07038364559412003, "logps/chosen": -1.3000078201293945, "logps/rejected": -1.5286756753921509, "loss": 1.6336, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3000078201293945, "rewards/margins": 0.22866801917552948, "rewards/rejected": -1.5286756753921509, "step": 3845 }, { "epoch": 2.0605452416792107, "grad_norm": 7.797180213118202, "learning_rate": 2.6994343286665156e-07, "logits/chosen": -0.18627727031707764, "logits/rejected": -0.04998369142413139, "logps/chosen": -1.2930742502212524, "logps/rejected": -1.4864602088928223, "loss": 1.6311, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2930742502212524, "rewards/margins": 0.1933860182762146, "rewards/rejected": -1.4864602088928223, "step": 3850 }, { "epoch": 2.063221274460612, "grad_norm": 7.29308685777475, "learning_rate": 2.6856179511868156e-07, "logits/chosen": -0.14326825737953186, "logits/rejected": 0.013318806886672974, "logps/chosen": -1.2368152141571045, "logps/rejected": -1.5324199199676514, "loss": 1.5788, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2368152141571045, "rewards/margins": 0.2956047058105469, "rewards/rejected": -1.5324199199676514, "step": 3855 }, { "epoch": 2.0658973072420137, "grad_norm": 6.185287692033218, "learning_rate": 2.6718240278519056e-07, "logits/chosen": -0.14798951148986816, "logits/rejected": -0.008536259643733501, "logps/chosen": -1.2535607814788818, "logps/rejected": -1.4576365947723389, "loss": 1.6073, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2535607814788818, "rewards/margins": 0.20407557487487793, "rewards/rejected": -1.4576365947723389, "step": 3860 }, { "epoch": 2.0685733400234154, "grad_norm": 10.875993443912234, "learning_rate": 2.6580526924904866e-07, "logits/chosen": -0.29120516777038574, "logits/rejected": -0.135166734457016, "logps/chosen": -1.2904469966888428, "logps/rejected": -1.4185327291488647, "loss": 1.6384, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2904469966888428, "rewards/margins": 0.1280859410762787, "rewards/rejected": -1.4185327291488647, "step": 3865 }, { "epoch": 2.0712493728048167, "grad_norm": 9.152730161588952, "learning_rate": 2.6443040787121186e-07, "logits/chosen": -0.19847515225410461, "logits/rejected": -0.1773093342781067, "logps/chosen": -1.1396772861480713, "logps/rejected": -1.3321276903152466, "loss": 1.5195, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1396772861480713, "rewards/margins": 0.19245044887065887, "rewards/rejected": -1.3321276903152466, "step": 3870 }, { "epoch": 2.0739254055862184, "grad_norm": 6.883716697989702, "learning_rate": 2.6305783199059084e-07, "logits/chosen": -0.19664858281612396, "logits/rejected": -0.09270669519901276, "logps/chosen": -1.2300426959991455, "logps/rejected": -1.4747213125228882, "loss": 1.568, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2300426959991455, "rewards/margins": 0.2446787804365158, "rewards/rejected": -1.4747213125228882, "step": 3875 }, { "epoch": 2.07660143836762, "grad_norm": 6.804884424226387, "learning_rate": 2.6168755492392324e-07, "logits/chosen": -0.21317140758037567, "logits/rejected": -0.08506114780902863, "logps/chosen": -1.1500242948532104, "logps/rejected": -1.3966689109802246, "loss": 1.4991, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1500242948532104, "rewards/margins": 0.24664464592933655, "rewards/rejected": -1.3966689109802246, "step": 3880 }, { "epoch": 2.0792774711490214, "grad_norm": 7.374519752982465, "learning_rate": 2.6031958996564274e-07, "logits/chosen": -0.22471275925636292, "logits/rejected": -0.09944645315408707, "logps/chosen": -1.1963858604431152, "logps/rejected": -1.4682543277740479, "loss": 1.533, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.1963858604431152, "rewards/margins": 0.27186843752861023, "rewards/rejected": -1.4682543277740479, "step": 3885 }, { "epoch": 2.081953503930423, "grad_norm": 10.671471040265736, "learning_rate": 2.589539503877518e-07, "logits/chosen": -0.14481914043426514, "logits/rejected": -0.06979800015687943, "logps/chosen": -1.2372180223464966, "logps/rejected": -1.419391393661499, "loss": 1.6031, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2372180223464966, "rewards/margins": 0.18217317759990692, "rewards/rejected": -1.419391393661499, "step": 3890 }, { "epoch": 2.084629536711825, "grad_norm": 9.820407721904644, "learning_rate": 2.5759064943969125e-07, "logits/chosen": -0.18728923797607422, "logits/rejected": 0.020716819912195206, "logps/chosen": -1.2106778621673584, "logps/rejected": -1.4359009265899658, "loss": 1.5687, "rewards/accuracies": 0.625, "rewards/chosen": -1.2106778621673584, "rewards/margins": 0.2252231389284134, "rewards/rejected": -1.4359009265899658, "step": 3895 }, { "epoch": 2.087305569493226, "grad_norm": 8.506512351210667, "learning_rate": 2.562297003482131e-07, "logits/chosen": -0.08664155006408691, "logits/rejected": -0.08900171518325806, "logps/chosen": -1.216568946838379, "logps/rejected": -1.4304020404815674, "loss": 1.5729, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.216568946838379, "rewards/margins": 0.2138332575559616, "rewards/rejected": -1.4304020404815674, "step": 3900 }, { "epoch": 2.089981602274628, "grad_norm": 7.499504238653955, "learning_rate": 2.548711163172512e-07, "logits/chosen": -0.1457064300775528, "logits/rejected": -0.05645012855529785, "logps/chosen": -1.2313451766967773, "logps/rejected": -1.4008392095565796, "loss": 1.5811, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2313451766967773, "rewards/margins": 0.16949406266212463, "rewards/rejected": -1.4008392095565796, "step": 3905 }, { "epoch": 2.0926576350560295, "grad_norm": 6.809512264296725, "learning_rate": 2.53514910527794e-07, "logits/chosen": -0.10277318954467773, "logits/rejected": -0.003941692411899567, "logps/chosen": -1.1524231433868408, "logps/rejected": -1.392700433731079, "loss": 1.5247, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1524231433868408, "rewards/margins": 0.24027732014656067, "rewards/rejected": -1.392700433731079, "step": 3910 }, { "epoch": 2.095333667837431, "grad_norm": 5.704136859915198, "learning_rate": 2.5216109613775573e-07, "logits/chosen": -0.189134418964386, "logits/rejected": -0.05074596405029297, "logps/chosen": -1.2564955949783325, "logps/rejected": -1.5363318920135498, "loss": 1.5776, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2564955949783325, "rewards/margins": 0.2798364758491516, "rewards/rejected": -1.5363318920135498, "step": 3915 }, { "epoch": 2.0980097006188325, "grad_norm": 6.740453561840508, "learning_rate": 2.5080968628184993e-07, "logits/chosen": -0.19768080115318298, "logits/rejected": -0.05966009944677353, "logps/chosen": -1.265911340713501, "logps/rejected": -1.5537766218185425, "loss": 1.5768, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.265911340713501, "rewards/margins": 0.28786543011665344, "rewards/rejected": -1.5537766218185425, "step": 3920 }, { "epoch": 2.1006857334002342, "grad_norm": 7.8854205875236625, "learning_rate": 2.494606940714605e-07, "logits/chosen": -0.1744232028722763, "logits/rejected": -0.07103218138217926, "logps/chosen": -1.1927406787872314, "logps/rejected": -1.4526997804641724, "loss": 1.5685, "rewards/accuracies": 0.5625, "rewards/chosen": -1.1927406787872314, "rewards/margins": 0.25995901226997375, "rewards/rejected": -1.4526997804641724, "step": 3925 }, { "epoch": 2.103361766181636, "grad_norm": 6.305959398254729, "learning_rate": 2.4811413259451625e-07, "logits/chosen": -0.25274330377578735, "logits/rejected": -0.13995765149593353, "logps/chosen": -1.2792290449142456, "logps/rejected": -1.4351447820663452, "loss": 1.6262, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2792290449142456, "rewards/margins": 0.15591581165790558, "rewards/rejected": -1.4351447820663452, "step": 3930 }, { "epoch": 2.106037798963037, "grad_norm": 7.203853391762811, "learning_rate": 2.46770014915362e-07, "logits/chosen": -0.09655294567346573, "logits/rejected": -0.04600976034998894, "logps/chosen": -1.2321531772613525, "logps/rejected": -1.5000993013381958, "loss": 1.5579, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2321531772613525, "rewards/margins": 0.26794594526290894, "rewards/rejected": -1.5000993013381958, "step": 3935 }, { "epoch": 2.108713831744439, "grad_norm": 9.213519824001787, "learning_rate": 2.45428354074634e-07, "logits/chosen": -0.13530853390693665, "logits/rejected": -0.10061784088611603, "logps/chosen": -1.1457960605621338, "logps/rejected": -1.5235750675201416, "loss": 1.4669, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1457960605621338, "rewards/margins": 0.37777888774871826, "rewards/rejected": -1.5235750675201416, "step": 3940 }, { "epoch": 2.1113898645258407, "grad_norm": 7.951658906871399, "learning_rate": 2.4408916308913105e-07, "logits/chosen": -0.15835228562355042, "logits/rejected": -0.006830123718827963, "logps/chosen": -1.276841402053833, "logps/rejected": -1.4330775737762451, "loss": 1.631, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.276841402053833, "rewards/margins": 0.15623629093170166, "rewards/rejected": -1.4330775737762451, "step": 3945 }, { "epoch": 2.114065897307242, "grad_norm": 9.809389717974016, "learning_rate": 2.4275245495169025e-07, "logits/chosen": -0.08743122965097427, "logits/rejected": 0.04935204237699509, "logps/chosen": -1.2254064083099365, "logps/rejected": -1.3818477392196655, "loss": 1.6021, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.2254064083099365, "rewards/margins": 0.15644130110740662, "rewards/rejected": -1.3818477392196655, "step": 3950 }, { "epoch": 2.1167419300886436, "grad_norm": 7.760483638978795, "learning_rate": 2.414182426310597e-07, "logits/chosen": -0.240851491689682, "logits/rejected": -0.16412867605686188, "logps/chosen": -1.1903098821640015, "logps/rejected": -1.4226571321487427, "loss": 1.5391, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.1903098821640015, "rewards/margins": 0.23234716057777405, "rewards/rejected": -1.4226571321487427, "step": 3955 }, { "epoch": 2.1194179628700454, "grad_norm": 11.898354640392771, "learning_rate": 2.400865390717734e-07, "logits/chosen": -0.1388021856546402, "logits/rejected": -0.03632751852273941, "logps/chosen": -1.2601227760314941, "logps/rejected": -1.5602470636367798, "loss": 1.5906, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2601227760314941, "rewards/margins": 0.3001241087913513, "rewards/rejected": -1.5602470636367798, "step": 3960 }, { "epoch": 2.1220939956514466, "grad_norm": 7.148422792573522, "learning_rate": 2.3875735719402475e-07, "logits/chosen": -0.08727221935987473, "logits/rejected": 0.004704681225121021, "logps/chosen": -1.1741969585418701, "logps/rejected": -1.4731131792068481, "loss": 1.5135, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1741969585418701, "rewards/margins": 0.29891616106033325, "rewards/rejected": -1.4731131792068481, "step": 3965 }, { "epoch": 2.1247700284328483, "grad_norm": 7.22524858929848, "learning_rate": 2.3743070989354258e-07, "logits/chosen": -0.16493722796440125, "logits/rejected": -0.09080357104539871, "logps/chosen": -1.2232723236083984, "logps/rejected": -1.5072360038757324, "loss": 1.5719, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2232723236083984, "rewards/margins": 0.28396373987197876, "rewards/rejected": -1.5072360038757324, "step": 3970 }, { "epoch": 2.12744606121425, "grad_norm": 6.966234959523901, "learning_rate": 2.3610661004146454e-07, "logits/chosen": -0.09802804887294769, "logits/rejected": -0.02745756134390831, "logps/chosen": -1.1443368196487427, "logps/rejected": -1.40862238407135, "loss": 1.5115, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1443368196487427, "rewards/margins": 0.26428547501564026, "rewards/rejected": -1.40862238407135, "step": 3975 }, { "epoch": 2.1301220939956513, "grad_norm": 8.690475995299611, "learning_rate": 2.3478507048421314e-07, "logits/chosen": -0.21995559334754944, "logits/rejected": -0.16529780626296997, "logps/chosen": -1.142746925354004, "logps/rejected": -1.4897350072860718, "loss": 1.4869, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.142746925354004, "rewards/margins": 0.34698814153671265, "rewards/rejected": -1.4897350072860718, "step": 3980 }, { "epoch": 2.132798126777053, "grad_norm": 11.365330215938014, "learning_rate": 2.334661040433713e-07, "logits/chosen": -0.24035818874835968, "logits/rejected": -0.15376703441143036, "logps/chosen": -1.217773199081421, "logps/rejected": -1.4895384311676025, "loss": 1.5593, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.217773199081421, "rewards/margins": 0.27176520228385925, "rewards/rejected": -1.4895384311676025, "step": 3985 }, { "epoch": 2.1354741595584548, "grad_norm": 7.236942359737125, "learning_rate": 2.321497235155568e-07, "logits/chosen": -0.2544265687465668, "logits/rejected": -0.1432047337293625, "logps/chosen": -1.1897337436676025, "logps/rejected": -1.4237349033355713, "loss": 1.5348, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1897337436676025, "rewards/margins": 0.23400099575519562, "rewards/rejected": -1.4237349033355713, "step": 3990 }, { "epoch": 2.138150192339856, "grad_norm": 7.941620296898851, "learning_rate": 2.3083594167229965e-07, "logits/chosen": -0.302916556596756, "logits/rejected": -0.0914468914270401, "logps/chosen": -1.2181164026260376, "logps/rejected": -1.5334707498550415, "loss": 1.5571, "rewards/accuracies": 0.625, "rewards/chosen": -1.2181164026260376, "rewards/margins": 0.3153544068336487, "rewards/rejected": -1.5334707498550415, "step": 3995 }, { "epoch": 2.1408262251212578, "grad_norm": 9.043760190413138, "learning_rate": 2.295247712599167e-07, "logits/chosen": -0.19980423152446747, "logits/rejected": -0.10652852058410645, "logps/chosen": -1.19603431224823, "logps/rejected": -1.461136817932129, "loss": 1.5351, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.19603431224823, "rewards/margins": 0.2651025950908661, "rewards/rejected": -1.461136817932129, "step": 4000 }, { "epoch": 2.1408262251212578, "eval_logits/chosen": 0.07140856981277466, "eval_logits/rejected": 0.1418316513299942, "eval_logps/chosen": -1.3020397424697876, "eval_logps/rejected": -1.496042013168335, "eval_loss": 1.6470303535461426, "eval_rewards/accuracies": 0.5697329640388489, "eval_rewards/chosen": -1.3020397424697876, "eval_rewards/margins": 0.19400210678577423, "eval_rewards/rejected": -1.496042013168335, "eval_runtime": 40.7989, "eval_samples_per_second": 32.967, "eval_steps_per_second": 8.26, "step": 4000 }, { "epoch": 2.1435022579026595, "grad_norm": 6.69689106524811, "learning_rate": 2.2821622499938948e-07, "logits/chosen": -0.19220581650733948, "logits/rejected": -0.00428193062543869, "logps/chosen": -1.3547049760818481, "logps/rejected": -1.513340950012207, "loss": 1.6739, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3547049760818481, "rewards/margins": 0.15863582491874695, "rewards/rejected": -1.513340950012207, "step": 4005 }, { "epoch": 2.1461782906840607, "grad_norm": 6.439031174437564, "learning_rate": 2.269103155862391e-07, "logits/chosen": -0.2250560224056244, "logits/rejected": -0.14797084033489227, "logps/chosen": -1.2645237445831299, "logps/rejected": -1.4367012977600098, "loss": 1.6099, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2645237445831299, "rewards/margins": 0.1721775382757187, "rewards/rejected": -1.4367012977600098, "step": 4010 }, { "epoch": 2.1488543234654625, "grad_norm": 7.528147440207392, "learning_rate": 2.2560705569040483e-07, "logits/chosen": -0.2006227672100067, "logits/rejected": 0.018616771325469017, "logps/chosen": -1.2492566108703613, "logps/rejected": -1.4439423084259033, "loss": 1.6014, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2492566108703613, "rewards/margins": 0.19468554854393005, "rewards/rejected": -1.4439423084259033, "step": 4015 }, { "epoch": 2.151530356246864, "grad_norm": 6.76322281964501, "learning_rate": 2.2430645795611963e-07, "logits/chosen": -0.27711382508277893, "logits/rejected": -0.16055673360824585, "logps/chosen": -1.2736780643463135, "logps/rejected": -1.437705636024475, "loss": 1.6194, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2736780643463135, "rewards/margins": 0.16402749717235565, "rewards/rejected": -1.437705636024475, "step": 4020 }, { "epoch": 2.1542063890282654, "grad_norm": 8.161004313533256, "learning_rate": 2.230085350017884e-07, "logits/chosen": -0.20443399250507355, "logits/rejected": -0.12384285032749176, "logps/chosen": -1.1757631301879883, "logps/rejected": -1.4113733768463135, "loss": 1.5517, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1757631301879883, "rewards/margins": 0.23561033606529236, "rewards/rejected": -1.4113733768463135, "step": 4025 }, { "epoch": 2.156882421809667, "grad_norm": 12.639231430085822, "learning_rate": 2.2171329941986554e-07, "logits/chosen": -0.23632891476154327, "logits/rejected": -0.17125725746154785, "logps/chosen": -1.194329857826233, "logps/rejected": -1.4643354415893555, "loss": 1.5404, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.194329857826233, "rewards/margins": 0.27000561356544495, "rewards/rejected": -1.4643354415893555, "step": 4030 }, { "epoch": 2.159558454591069, "grad_norm": 13.035082920965463, "learning_rate": 2.2042076377673202e-07, "logits/chosen": -0.22286994755268097, "logits/rejected": -0.19176195561885834, "logps/chosen": -1.2081761360168457, "logps/rejected": -1.312530755996704, "loss": 1.6095, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2081761360168457, "rewards/margins": 0.10435459762811661, "rewards/rejected": -1.312530755996704, "step": 4035 }, { "epoch": 2.16223448737247, "grad_norm": 6.83123896732779, "learning_rate": 2.1913094061257476e-07, "logits/chosen": -0.22944974899291992, "logits/rejected": -0.19934967160224915, "logps/chosen": -1.1782562732696533, "logps/rejected": -1.3568545579910278, "loss": 1.55, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.1782562732696533, "rewards/margins": 0.1785982847213745, "rewards/rejected": -1.3568545579910278, "step": 4040 }, { "epoch": 2.164910520153872, "grad_norm": 9.180685223568698, "learning_rate": 2.178438424412633e-07, "logits/chosen": -0.16974474489688873, "logits/rejected": -0.06564159691333771, "logps/chosen": -1.2702267169952393, "logps/rejected": -1.4524881839752197, "loss": 1.6181, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2702267169952393, "rewards/margins": 0.18226134777069092, "rewards/rejected": -1.4524881839752197, "step": 4045 }, { "epoch": 2.1675865529352736, "grad_norm": 4.982141323452237, "learning_rate": 2.165594817502302e-07, "logits/chosen": -0.24664337933063507, "logits/rejected": -0.16238948702812195, "logps/chosen": -1.255029559135437, "logps/rejected": -1.4013950824737549, "loss": 1.6247, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.255029559135437, "rewards/margins": 0.14636535942554474, "rewards/rejected": -1.4013950824737549, "step": 4050 }, { "epoch": 2.170262585716675, "grad_norm": 6.9593790107088855, "learning_rate": 2.1527787100034806e-07, "logits/chosen": -0.1397363841533661, "logits/rejected": -0.09621234238147736, "logps/chosen": -1.23110032081604, "logps/rejected": -1.4188356399536133, "loss": 1.5766, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.23110032081604, "rewards/margins": 0.1877353936433792, "rewards/rejected": -1.4188356399536133, "step": 4055 }, { "epoch": 2.1729386184980766, "grad_norm": 8.771723975407568, "learning_rate": 2.1399902262581037e-07, "logits/chosen": -0.047494079917669296, "logits/rejected": 0.030322402715682983, "logps/chosen": -1.1686232089996338, "logps/rejected": -1.3430014848709106, "loss": 1.538, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1686232089996338, "rewards/margins": 0.17437830567359924, "rewards/rejected": -1.3430014848709106, "step": 4060 }, { "epoch": 2.1756146512794783, "grad_norm": 8.165202393914798, "learning_rate": 2.127229490340094e-07, "logits/chosen": -0.2657929062843323, "logits/rejected": -0.19261549413204193, "logps/chosen": -1.2344993352890015, "logps/rejected": -1.4854099750518799, "loss": 1.557, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2344993352890015, "rewards/margins": 0.2509106397628784, "rewards/rejected": -1.4854099750518799, "step": 4065 }, { "epoch": 2.1782906840608796, "grad_norm": 11.282347383089665, "learning_rate": 2.1144966260541698e-07, "logits/chosen": -0.147164449095726, "logits/rejected": 0.02844533883035183, "logps/chosen": -1.193411111831665, "logps/rejected": -1.4447523355484009, "loss": 1.542, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.193411111831665, "rewards/margins": 0.25134122371673584, "rewards/rejected": -1.4447523355484009, "step": 4070 }, { "epoch": 2.1809667168422813, "grad_norm": 7.959387334694807, "learning_rate": 2.1017917569346332e-07, "logits/chosen": -0.19362321496009827, "logits/rejected": -0.038269199430942535, "logps/chosen": -1.2862924337387085, "logps/rejected": -1.4655495882034302, "loss": 1.6203, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2862924337387085, "rewards/margins": 0.17925693094730377, "rewards/rejected": -1.4655495882034302, "step": 4075 }, { "epoch": 2.183642749623683, "grad_norm": 5.997204567632148, "learning_rate": 2.0891150062441837e-07, "logits/chosen": -0.19443969428539276, "logits/rejected": -0.08346713334321976, "logps/chosen": -1.2668898105621338, "logps/rejected": -1.4902750253677368, "loss": 1.6037, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2668898105621338, "rewards/margins": 0.22338524460792542, "rewards/rejected": -1.4902750253677368, "step": 4080 }, { "epoch": 2.1863187824050843, "grad_norm": 7.002594484782536, "learning_rate": 2.0764664969727086e-07, "logits/chosen": -0.1699420064687729, "logits/rejected": -0.0828448086977005, "logps/chosen": -1.2525122165679932, "logps/rejected": -1.4004987478256226, "loss": 1.6426, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2525122165679932, "rewards/margins": 0.147986501455307, "rewards/rejected": -1.4004987478256226, "step": 4085 }, { "epoch": 2.188994815186486, "grad_norm": 6.749307057443181, "learning_rate": 2.0638463518361033e-07, "logits/chosen": -0.23159702122211456, "logits/rejected": -0.07271578907966614, "logps/chosen": -1.2239384651184082, "logps/rejected": -1.4089152812957764, "loss": 1.5835, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2239384651184082, "rewards/margins": 0.18497692048549652, "rewards/rejected": -1.4089152812957764, "step": 4090 }, { "epoch": 2.1916708479678877, "grad_norm": 8.019662305992837, "learning_rate": 2.0512546932750702e-07, "logits/chosen": -0.2117147445678711, "logits/rejected": -0.14385417103767395, "logps/chosen": -1.3089931011199951, "logps/rejected": -1.4399337768554688, "loss": 1.65, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3089931011199951, "rewards/margins": 0.13094060122966766, "rewards/rejected": -1.4399337768554688, "step": 4095 }, { "epoch": 2.194346880749289, "grad_norm": 9.726428326876832, "learning_rate": 2.0386916434539343e-07, "logits/chosen": -0.16599038243293762, "logits/rejected": -0.06162125989794731, "logps/chosen": -1.1586806774139404, "logps/rejected": -1.4184343814849854, "loss": 1.5374, "rewards/accuracies": 0.59375, "rewards/chosen": -1.1586806774139404, "rewards/margins": 0.2597537636756897, "rewards/rejected": -1.4184343814849854, "step": 4100 }, { "epoch": 2.1970229135306907, "grad_norm": 9.762474966817123, "learning_rate": 2.0261573242594627e-07, "logits/chosen": -0.18855324387550354, "logits/rejected": -0.014698073267936707, "logps/chosen": -1.280236840248108, "logps/rejected": -1.449611783027649, "loss": 1.6209, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.280236840248108, "rewards/margins": 0.16937494277954102, "rewards/rejected": -1.449611783027649, "step": 4105 }, { "epoch": 2.1996989463120924, "grad_norm": 9.559046606243188, "learning_rate": 2.0136518572996724e-07, "logits/chosen": -0.14109668135643005, "logits/rejected": -0.002549926284700632, "logps/chosen": -1.2072608470916748, "logps/rejected": -1.392521619796753, "loss": 1.5823, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2072608470916748, "rewards/margins": 0.18526069819927216, "rewards/rejected": -1.392521619796753, "step": 4110 }, { "epoch": 2.202374979093494, "grad_norm": 7.579963806366611, "learning_rate": 2.0011753639026617e-07, "logits/chosen": -0.1384604573249817, "logits/rejected": -0.1108267530798912, "logps/chosen": -1.216487169265747, "logps/rejected": -1.496044397354126, "loss": 1.554, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.216487169265747, "rewards/margins": 0.27955740690231323, "rewards/rejected": -1.496044397354126, "step": 4115 }, { "epoch": 2.2050510118748954, "grad_norm": 5.4827811338445, "learning_rate": 1.988727965115421e-07, "logits/chosen": -0.17637844383716583, "logits/rejected": -0.12951934337615967, "logps/chosen": -1.1752429008483887, "logps/rejected": -1.428956151008606, "loss": 1.5207, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1752429008483887, "rewards/margins": 0.2537131905555725, "rewards/rejected": -1.428956151008606, "step": 4120 }, { "epoch": 2.207727044656297, "grad_norm": 6.844973266135418, "learning_rate": 1.9763097817026713e-07, "logits/chosen": -0.212055966258049, "logits/rejected": -0.04291188344359398, "logps/chosen": -1.2000548839569092, "logps/rejected": -1.415569543838501, "loss": 1.5606, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2000548839569092, "rewards/margins": 0.21551451086997986, "rewards/rejected": -1.415569543838501, "step": 4125 }, { "epoch": 2.210403077437699, "grad_norm": 7.6682117055039125, "learning_rate": 1.9639209341456796e-07, "logits/chosen": -0.15989407896995544, "logits/rejected": -0.09153582155704498, "logps/chosen": -1.2050529718399048, "logps/rejected": -1.4642305374145508, "loss": 1.5452, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2050529718399048, "rewards/margins": 0.25917768478393555, "rewards/rejected": -1.4642305374145508, "step": 4130 }, { "epoch": 2.2130791102191, "grad_norm": 14.738132300531646, "learning_rate": 1.951561542641102e-07, "logits/chosen": -0.16685988008975983, "logits/rejected": -0.1748218834400177, "logps/chosen": -1.2561938762664795, "logps/rejected": -1.4929753541946411, "loss": 1.6073, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2561938762664795, "rewards/margins": 0.23678138852119446, "rewards/rejected": -1.4929753541946411, "step": 4135 }, { "epoch": 2.215755143000502, "grad_norm": 8.486783587459678, "learning_rate": 1.939231727099806e-07, "logits/chosen": -0.28792333602905273, "logits/rejected": -0.21578791737556458, "logps/chosen": -1.1780260801315308, "logps/rejected": -1.4389264583587646, "loss": 1.5305, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1780260801315308, "rewards/margins": 0.26090025901794434, "rewards/rejected": -1.4389264583587646, "step": 4140 }, { "epoch": 2.2184311757819035, "grad_norm": 8.16180910963854, "learning_rate": 1.926931607145719e-07, "logits/chosen": -0.09905657172203064, "logits/rejected": 0.023060614243149757, "logps/chosen": -1.2769787311553955, "logps/rejected": -1.5458099842071533, "loss": 1.6054, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2769787311553955, "rewards/margins": 0.26883116364479065, "rewards/rejected": -1.5458099842071533, "step": 4145 }, { "epoch": 2.221107208563305, "grad_norm": 8.070106294047847, "learning_rate": 1.9146613021146564e-07, "logits/chosen": -0.14280208945274353, "logits/rejected": -0.0710291638970375, "logps/chosen": -1.1692044734954834, "logps/rejected": -1.434899926185608, "loss": 1.5355, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1692044734954834, "rewards/margins": 0.26569539308547974, "rewards/rejected": -1.434899926185608, "step": 4150 }, { "epoch": 2.2237832413447065, "grad_norm": 8.036551483049287, "learning_rate": 1.9024209310531736e-07, "logits/chosen": -0.09412866830825806, "logits/rejected": -0.12163051217794418, "logps/chosen": -1.2059614658355713, "logps/rejected": -1.4259283542633057, "loss": 1.5517, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2059614658355713, "rewards/margins": 0.2199670821428299, "rewards/rejected": -1.4259283542633057, "step": 4155 }, { "epoch": 2.2264592741261082, "grad_norm": 9.278979880678373, "learning_rate": 1.890210612717401e-07, "logits/chosen": -0.17106539011001587, "logits/rejected": -0.04072389751672745, "logps/chosen": -1.2591288089752197, "logps/rejected": -1.4694256782531738, "loss": 1.6059, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2591288089752197, "rewards/margins": 0.21029677987098694, "rewards/rejected": -1.4694256782531738, "step": 4160 }, { "epoch": 2.2291353069075095, "grad_norm": 7.100599192462308, "learning_rate": 1.8780304655719054e-07, "logits/chosen": -0.1789466142654419, "logits/rejected": -0.09607045352458954, "logps/chosen": -1.26167631149292, "logps/rejected": -1.4886329174041748, "loss": 1.6086, "rewards/accuracies": 0.518750011920929, "rewards/chosen": -1.26167631149292, "rewards/margins": 0.22695684432983398, "rewards/rejected": -1.4886329174041748, "step": 4165 }, { "epoch": 2.231811339688911, "grad_norm": 8.472961986461971, "learning_rate": 1.865880607788523e-07, "logits/chosen": -0.0813198909163475, "logits/rejected": -0.036982350051403046, "logps/chosen": -1.2395800352096558, "logps/rejected": -1.4827731847763062, "loss": 1.575, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2395800352096558, "rewards/margins": 0.2431931495666504, "rewards/rejected": -1.4827731847763062, "step": 4170 }, { "epoch": 2.234487372470313, "grad_norm": 7.059742378344645, "learning_rate": 1.8537611572452316e-07, "logits/chosen": -0.1719505935907364, "logits/rejected": -0.08126050978899002, "logps/chosen": -1.2195240259170532, "logps/rejected": -1.388882040977478, "loss": 1.5901, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2195240259170532, "rewards/margins": 0.1693582832813263, "rewards/rejected": -1.388882040977478, "step": 4175 }, { "epoch": 2.237163405251714, "grad_norm": 10.458292911193098, "learning_rate": 1.84167223152499e-07, "logits/chosen": -0.20625898241996765, "logits/rejected": -0.017097296193242073, "logps/chosen": -1.1701924800872803, "logps/rejected": -1.4478870630264282, "loss": 1.5156, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.1701924800872803, "rewards/margins": 0.2776945233345032, "rewards/rejected": -1.4478870630264282, "step": 4180 }, { "epoch": 2.239839438033116, "grad_norm": 8.809485044183429, "learning_rate": 1.8296139479146112e-07, "logits/chosen": -0.21345221996307373, "logits/rejected": -0.21348659694194794, "logps/chosen": -1.1450181007385254, "logps/rejected": -1.4158300161361694, "loss": 1.5114, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1450181007385254, "rewards/margins": 0.270812064409256, "rewards/rejected": -1.4158300161361694, "step": 4185 }, { "epoch": 2.2425154708145176, "grad_norm": 8.494491689490282, "learning_rate": 1.8175864234036132e-07, "logits/chosen": -0.10274024307727814, "logits/rejected": -0.007196763064712286, "logps/chosen": -1.2229564189910889, "logps/rejected": -1.435478925704956, "loss": 1.5785, "rewards/accuracies": 0.625, "rewards/chosen": -1.2229564189910889, "rewards/margins": 0.21252259612083435, "rewards/rejected": -1.435478925704956, "step": 4190 }, { "epoch": 2.245191503595919, "grad_norm": 5.3920501659104145, "learning_rate": 1.805589774683094e-07, "logits/chosen": -0.2902788817882538, "logits/rejected": -0.15456892549991608, "logps/chosen": -1.2663679122924805, "logps/rejected": -1.4145959615707397, "loss": 1.6248, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2663679122924805, "rewards/margins": 0.14822818338871002, "rewards/rejected": -1.4145959615707397, "step": 4195 }, { "epoch": 2.2478675363773206, "grad_norm": 8.537807124499022, "learning_rate": 1.79362411814459e-07, "logits/chosen": -0.0195014588534832, "logits/rejected": -0.04255833104252815, "logps/chosen": -1.2723864316940308, "logps/rejected": -1.5024255514144897, "loss": 1.601, "rewards/accuracies": 0.625, "rewards/chosen": -1.2723864316940308, "rewards/margins": 0.2300390899181366, "rewards/rejected": -1.5024255514144897, "step": 4200 }, { "epoch": 2.2505435691587223, "grad_norm": 8.356553445152443, "learning_rate": 1.7816895698789552e-07, "logits/chosen": -0.21217474341392517, "logits/rejected": -0.14514055848121643, "logps/chosen": -1.2000718116760254, "logps/rejected": -1.4663290977478027, "loss": 1.5215, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.2000718116760254, "rewards/margins": 0.266257107257843, "rewards/rejected": -1.4663290977478027, "step": 4205 }, { "epoch": 2.2532196019401236, "grad_norm": 8.480499781545147, "learning_rate": 1.7697862456752271e-07, "logits/chosen": -0.19726264476776123, "logits/rejected": -0.11173512786626816, "logps/chosen": -1.213542103767395, "logps/rejected": -1.5523083209991455, "loss": 1.5522, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.213542103767395, "rewards/margins": 0.3387662172317505, "rewards/rejected": -1.5523083209991455, "step": 4210 }, { "epoch": 2.2558956347215253, "grad_norm": 6.8471923153577725, "learning_rate": 1.7579142610195124e-07, "logits/chosen": -0.17487090826034546, "logits/rejected": -0.06448483467102051, "logps/chosen": -1.2431567907333374, "logps/rejected": -1.452376365661621, "loss": 1.5932, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2431567907333374, "rewards/margins": 0.2092195749282837, "rewards/rejected": -1.452376365661621, "step": 4215 }, { "epoch": 2.258571667502927, "grad_norm": 7.816383861597878, "learning_rate": 1.7460737310938568e-07, "logits/chosen": -0.22700047492980957, "logits/rejected": -0.05646955966949463, "logps/chosen": -1.2269337177276611, "logps/rejected": -1.4205477237701416, "loss": 1.5907, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2269337177276611, "rewards/margins": 0.19361400604248047, "rewards/rejected": -1.4205477237701416, "step": 4220 }, { "epoch": 2.2612477002843283, "grad_norm": 10.822570695434614, "learning_rate": 1.734264770775133e-07, "logits/chosen": -0.22074365615844727, "logits/rejected": -0.07267844676971436, "logps/chosen": -1.2606627941131592, "logps/rejected": -1.5198907852172852, "loss": 1.5868, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2606627941131592, "rewards/margins": 0.25922802090644836, "rewards/rejected": -1.5198907852172852, "step": 4225 }, { "epoch": 2.26392373306573, "grad_norm": 7.833504416013592, "learning_rate": 1.7224874946339241e-07, "logits/chosen": -0.2175557166337967, "logits/rejected": -0.15451601147651672, "logps/chosen": -1.2821791172027588, "logps/rejected": -1.581095576286316, "loss": 1.6205, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2821791172027588, "rewards/margins": 0.2989165782928467, "rewards/rejected": -1.581095576286316, "step": 4230 }, { "epoch": 2.2665997658471317, "grad_norm": 7.657476050617335, "learning_rate": 1.7107420169334186e-07, "logits/chosen": -0.14881347119808197, "logits/rejected": -0.04365311190485954, "logps/chosen": -1.2934807538986206, "logps/rejected": -1.379242181777954, "loss": 1.6734, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -1.2934807538986206, "rewards/margins": 0.08576139062643051, "rewards/rejected": -1.379242181777954, "step": 4235 }, { "epoch": 2.269275798628533, "grad_norm": 8.447673546250616, "learning_rate": 1.6990284516282893e-07, "logits/chosen": -0.16056771576404572, "logits/rejected": -0.06181789189577103, "logps/chosen": -1.2400410175323486, "logps/rejected": -1.3618853092193604, "loss": 1.6195, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2400410175323486, "rewards/margins": 0.12184418737888336, "rewards/rejected": -1.3618853092193604, "step": 4240 }, { "epoch": 2.2719518314099347, "grad_norm": 7.006714687663983, "learning_rate": 1.687346912363602e-07, "logits/chosen": -0.23508353531360626, "logits/rejected": -0.13327956199645996, "logps/chosen": -1.2458765506744385, "logps/rejected": -1.371870517730713, "loss": 1.6025, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2458765506744385, "rewards/margins": 0.125994011759758, "rewards/rejected": -1.371870517730713, "step": 4245 }, { "epoch": 2.2746278641913364, "grad_norm": 5.102888582064418, "learning_rate": 1.675697512473697e-07, "logits/chosen": -0.1529920995235443, "logits/rejected": -0.008715217001736164, "logps/chosen": -1.2363736629486084, "logps/rejected": -1.4021012783050537, "loss": 1.6104, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2363736629486084, "rewards/margins": 0.16572771966457367, "rewards/rejected": -1.4021012783050537, "step": 4250 }, { "epoch": 2.2773038969727377, "grad_norm": 10.903502119428019, "learning_rate": 1.6640803649811087e-07, "logits/chosen": -0.21960356831550598, "logits/rejected": -0.020394813269376755, "logps/chosen": -1.2543021440505981, "logps/rejected": -1.4993212223052979, "loss": 1.5882, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2543021440505981, "rewards/margins": 0.24501904845237732, "rewards/rejected": -1.4993212223052979, "step": 4255 }, { "epoch": 2.2799799297541394, "grad_norm": 8.237759215848795, "learning_rate": 1.6524955825954472e-07, "logits/chosen": -0.18711432814598083, "logits/rejected": -0.08437317609786987, "logps/chosen": -1.1727478504180908, "logps/rejected": -1.4643090963363647, "loss": 1.5211, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1727478504180908, "rewards/margins": 0.29156142473220825, "rewards/rejected": -1.4643090963363647, "step": 4260 }, { "epoch": 2.282655962535541, "grad_norm": 6.595624346666196, "learning_rate": 1.6409432777123277e-07, "logits/chosen": -0.2630254626274109, "logits/rejected": -0.12001305818557739, "logps/chosen": -1.2227981090545654, "logps/rejected": -1.4482836723327637, "loss": 1.5619, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2227981090545654, "rewards/margins": 0.22548551857471466, "rewards/rejected": -1.4482836723327637, "step": 4265 }, { "epoch": 2.285331995316943, "grad_norm": 5.90018653535347, "learning_rate": 1.6294235624122577e-07, "logits/chosen": -0.10724867880344391, "logits/rejected": 0.10244635492563248, "logps/chosen": -1.2431776523590088, "logps/rejected": -1.5928304195404053, "loss": 1.5783, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2431776523590088, "rewards/margins": 0.34965258836746216, "rewards/rejected": -1.5928304195404053, "step": 4270 }, { "epoch": 2.288008028098344, "grad_norm": 8.869573524282167, "learning_rate": 1.6179365484595697e-07, "logits/chosen": -0.1616896092891693, "logits/rejected": -0.09324827045202255, "logps/chosen": -1.2973737716674805, "logps/rejected": -1.4981673955917358, "loss": 1.6484, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2973737716674805, "rewards/margins": 0.2007937729358673, "rewards/rejected": -1.4981673955917358, "step": 4275 }, { "epoch": 2.290684060879746, "grad_norm": 6.99952481826747, "learning_rate": 1.60648234730132e-07, "logits/chosen": -0.18767237663269043, "logits/rejected": -0.1236768588423729, "logps/chosen": -1.1969908475875854, "logps/rejected": -1.3880600929260254, "loss": 1.5465, "rewards/accuracies": 0.625, "rewards/chosen": -1.1969908475875854, "rewards/margins": 0.19106918573379517, "rewards/rejected": -1.3880600929260254, "step": 4280 }, { "epoch": 2.293360093661147, "grad_norm": 10.175508917314751, "learning_rate": 1.595061070066222e-07, "logits/chosen": -0.1178465262055397, "logits/rejected": -0.1252368986606598, "logps/chosen": -1.1660603284835815, "logps/rejected": -1.4133269786834717, "loss": 1.5046, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.1660603284835815, "rewards/margins": 0.24726668000221252, "rewards/rejected": -1.4133269786834717, "step": 4285 }, { "epoch": 2.296036126442549, "grad_norm": 10.362950866375071, "learning_rate": 1.5836728275635542e-07, "logits/chosen": -0.23263172805309296, "logits/rejected": -0.07924433052539825, "logps/chosen": -1.269500970840454, "logps/rejected": -1.436907172203064, "loss": 1.611, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.269500970840454, "rewards/margins": 0.16740620136260986, "rewards/rejected": -1.436907172203064, "step": 4290 }, { "epoch": 2.2987121592239506, "grad_norm": 8.875971527758225, "learning_rate": 1.5723177302820984e-07, "logits/chosen": -0.18567875027656555, "logits/rejected": -0.1478704959154129, "logps/chosen": -1.2316734790802002, "logps/rejected": -1.4582233428955078, "loss": 1.5816, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2316734790802002, "rewards/margins": 0.2265499383211136, "rewards/rejected": -1.4582233428955078, "step": 4295 }, { "epoch": 2.3013881920053523, "grad_norm": 11.997345478520547, "learning_rate": 1.5609958883890544e-07, "logits/chosen": -0.19638779759407043, "logits/rejected": -0.1040961742401123, "logps/chosen": -1.2302956581115723, "logps/rejected": -1.3700530529022217, "loss": 1.5969, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2302956581115723, "rewards/margins": 0.13975724577903748, "rewards/rejected": -1.3700530529022217, "step": 4300 }, { "epoch": 2.3040642247867535, "grad_norm": 9.273363482162067, "learning_rate": 1.5497074117289865e-07, "logits/chosen": -0.2623113989830017, "logits/rejected": -0.16000813245773315, "logps/chosen": -1.2087552547454834, "logps/rejected": -1.4842298030853271, "loss": 1.5421, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2087552547454834, "rewards/margins": 0.27547457814216614, "rewards/rejected": -1.4842298030853271, "step": 4305 }, { "epoch": 2.3067402575681553, "grad_norm": 8.284717897141489, "learning_rate": 1.5384524098227402e-07, "logits/chosen": -0.20485445857048035, "logits/rejected": -0.035859428346157074, "logps/chosen": -1.22329843044281, "logps/rejected": -1.4752705097198486, "loss": 1.5734, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.22329843044281, "rewards/margins": 0.2519720196723938, "rewards/rejected": -1.4752705097198486, "step": 4310 }, { "epoch": 2.3094162903495565, "grad_norm": 7.149524801004495, "learning_rate": 1.5272309918663974e-07, "logits/chosen": -0.18672919273376465, "logits/rejected": -0.06092787906527519, "logps/chosen": -1.2216147184371948, "logps/rejected": -1.401799201965332, "loss": 1.5917, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2216147184371948, "rewards/margins": 0.1801842749118805, "rewards/rejected": -1.401799201965332, "step": 4315 }, { "epoch": 2.3120923231309582, "grad_norm": 8.360417135910502, "learning_rate": 1.516043266730201e-07, "logits/chosen": -0.20249688625335693, "logits/rejected": -0.09940443933010101, "logps/chosen": -1.2791379690170288, "logps/rejected": -1.4019443988800049, "loss": 1.6402, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2791379690170288, "rewards/margins": 0.1228063553571701, "rewards/rejected": -1.4019443988800049, "step": 4320 }, { "epoch": 2.31476835591236, "grad_norm": 11.234185197249264, "learning_rate": 1.504889342957512e-07, "logits/chosen": -0.21147003769874573, "logits/rejected": -0.07601438462734222, "logps/chosen": -1.203464150428772, "logps/rejected": -1.4852374792099, "loss": 1.5519, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.203464150428772, "rewards/margins": 0.28177323937416077, "rewards/rejected": -1.4852374792099, "step": 4325 }, { "epoch": 2.3174443886937617, "grad_norm": 6.501783311715216, "learning_rate": 1.4937693287637453e-07, "logits/chosen": -0.1648644059896469, "logits/rejected": -0.07460461556911469, "logps/chosen": -1.304404377937317, "logps/rejected": -1.4362707138061523, "loss": 1.6622, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.304404377937317, "rewards/margins": 0.131866455078125, "rewards/rejected": -1.4362707138061523, "step": 4330 }, { "epoch": 2.320120421475163, "grad_norm": 10.205396624845779, "learning_rate": 1.4826833320353305e-07, "logits/chosen": -0.14972421526908875, "logits/rejected": -0.09903161227703094, "logps/chosen": -1.281598448753357, "logps/rejected": -1.4960078001022339, "loss": 1.6442, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.281598448753357, "rewards/margins": 0.2144092321395874, "rewards/rejected": -1.4960078001022339, "step": 4335 }, { "epoch": 2.3227964542565647, "grad_norm": 8.916306505575866, "learning_rate": 1.4716314603286528e-07, "logits/chosen": -0.18709716200828552, "logits/rejected": -0.04617304354906082, "logps/chosen": -1.1499556303024292, "logps/rejected": -1.5380523204803467, "loss": 1.4906, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.1499556303024292, "rewards/margins": 0.3880966603755951, "rewards/rejected": -1.5380523204803467, "step": 4340 }, { "epoch": 2.3254724870379664, "grad_norm": 9.022344805920719, "learning_rate": 1.4606138208690233e-07, "logits/chosen": -0.2786336839199066, "logits/rejected": -0.20892539620399475, "logps/chosen": -1.3439069986343384, "logps/rejected": -1.3802927732467651, "loss": 1.7034, "rewards/accuracies": 0.53125, "rewards/chosen": -1.3439069986343384, "rewards/margins": 0.03638586029410362, "rewards/rejected": -1.3802927732467651, "step": 4345 }, { "epoch": 2.3281485198193677, "grad_norm": 7.711159128746238, "learning_rate": 1.4496305205496251e-07, "logits/chosen": -0.1719704121351242, "logits/rejected": -0.11904128640890121, "logps/chosen": -1.239412784576416, "logps/rejected": -1.5017035007476807, "loss": 1.5726, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.239412784576416, "rewards/margins": 0.2622907757759094, "rewards/rejected": -1.5017035007476807, "step": 4350 }, { "epoch": 2.3308245526007694, "grad_norm": 6.595624727606789, "learning_rate": 1.4386816659304895e-07, "logits/chosen": -0.2478518784046173, "logits/rejected": -0.17337559163570404, "logps/chosen": -1.2752437591552734, "logps/rejected": -1.4763338565826416, "loss": 1.6006, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2752437591552734, "rewards/margins": 0.2010900229215622, "rewards/rejected": -1.4763338565826416, "step": 4355 }, { "epoch": 2.333500585382171, "grad_norm": 7.420736127491981, "learning_rate": 1.4277673632374492e-07, "logits/chosen": -0.2834548354148865, "logits/rejected": -0.10949674993753433, "logps/chosen": -1.3200583457946777, "logps/rejected": -1.4551689624786377, "loss": 1.6533, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3200583457946777, "rewards/margins": 0.13511063158512115, "rewards/rejected": -1.4551689624786377, "step": 4360 }, { "epoch": 2.3361766181635724, "grad_norm": 7.875809964585136, "learning_rate": 1.416887718361119e-07, "logits/chosen": -0.13666775822639465, "logits/rejected": -0.14465972781181335, "logps/chosen": -1.1882762908935547, "logps/rejected": -1.4436252117156982, "loss": 1.5433, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1882762908935547, "rewards/margins": 0.2553488314151764, "rewards/rejected": -1.4436252117156982, "step": 4365 }, { "epoch": 2.338852650944974, "grad_norm": 9.979204525063246, "learning_rate": 1.406042836855859e-07, "logits/chosen": -0.2052156627178192, "logits/rejected": -0.10023872554302216, "logps/chosen": -1.155705451965332, "logps/rejected": -1.534826636314392, "loss": 1.4966, "rewards/accuracies": 0.625, "rewards/chosen": -1.155705451965332, "rewards/margins": 0.37912115454673767, "rewards/rejected": -1.534826636314392, "step": 4370 }, { "epoch": 2.341528683726376, "grad_norm": 10.66112545100067, "learning_rate": 1.3952328239387595e-07, "logits/chosen": -0.2963188588619232, "logits/rejected": -0.14353522658348083, "logps/chosen": -1.2621066570281982, "logps/rejected": -1.4838224649429321, "loss": 1.6012, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2621066570281982, "rewards/margins": 0.22171583771705627, "rewards/rejected": -1.4838224649429321, "step": 4375 }, { "epoch": 2.344204716507777, "grad_norm": 7.217437454739501, "learning_rate": 1.3844577844886109e-07, "logits/chosen": -0.20205311477184296, "logits/rejected": -0.0537579171359539, "logps/chosen": -1.2881264686584473, "logps/rejected": -1.448127269744873, "loss": 1.6281, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2881264686584473, "rewards/margins": 0.16000090539455414, "rewards/rejected": -1.448127269744873, "step": 4380 }, { "epoch": 2.346880749289179, "grad_norm": 9.649330496694866, "learning_rate": 1.3737178230448955e-07, "logits/chosen": -0.25397199392318726, "logits/rejected": -0.12670288980007172, "logps/chosen": -1.2086747884750366, "logps/rejected": -1.5265108346939087, "loss": 1.5239, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2086747884750366, "rewards/margins": 0.31783586740493774, "rewards/rejected": -1.5265108346939087, "step": 4385 }, { "epoch": 2.3495567820705805, "grad_norm": 6.571409844411033, "learning_rate": 1.363013043806764e-07, "logits/chosen": -0.19165340065956116, "logits/rejected": -0.07862716913223267, "logps/chosen": -1.2117689847946167, "logps/rejected": -1.3977199792861938, "loss": 1.5744, "rewards/accuracies": 0.625, "rewards/chosen": -1.2117689847946167, "rewards/margins": 0.18595094978809357, "rewards/rejected": -1.3977199792861938, "step": 4390 }, { "epoch": 2.3522328148519818, "grad_norm": 6.7949582778565665, "learning_rate": 1.352343550632034e-07, "logits/chosen": -0.1610317975282669, "logits/rejected": -0.03247717395424843, "logps/chosen": -1.237166166305542, "logps/rejected": -1.5563886165618896, "loss": 1.5657, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.237166166305542, "rewards/margins": 0.3192223906517029, "rewards/rejected": -1.5563886165618896, "step": 4395 }, { "epoch": 2.3549088476333835, "grad_norm": 6.170379624826009, "learning_rate": 1.3417094470361722e-07, "logits/chosen": -0.2740820050239563, "logits/rejected": -0.14671733975410461, "logps/chosen": -1.198980450630188, "logps/rejected": -1.4838148355484009, "loss": 1.5309, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.198980450630188, "rewards/margins": 0.28483447432518005, "rewards/rejected": -1.4838148355484009, "step": 4400 }, { "epoch": 2.3549088476333835, "eval_logits/chosen": 0.04385645315051079, "eval_logits/rejected": 0.1132112517952919, "eval_logps/chosen": -1.3050687313079834, "eval_logps/rejected": -1.5041919946670532, "eval_loss": 1.6466861963272095, "eval_rewards/accuracies": 0.5727003216743469, "eval_rewards/chosen": -1.3050687313079834, "eval_rewards/margins": 0.1991230994462967, "eval_rewards/rejected": -1.5041919946670532, "eval_runtime": 40.8223, "eval_samples_per_second": 32.948, "eval_steps_per_second": 8.255, "step": 4400 }, { "epoch": 2.357584880414785, "grad_norm": 7.5349630898521225, "learning_rate": 1.3311108361913015e-07, "logits/chosen": -0.2757181227207184, "logits/rejected": -0.23484423756599426, "logps/chosen": -1.213034987449646, "logps/rejected": -1.3631339073181152, "loss": 1.5892, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.213034987449646, "rewards/margins": 0.1500987857580185, "rewards/rejected": -1.3631339073181152, "step": 4405 }, { "epoch": 2.3602609131961865, "grad_norm": 6.592700251523777, "learning_rate": 1.3205478209251874e-07, "logits/chosen": -0.19986172020435333, "logits/rejected": -0.17671015858650208, "logps/chosen": -1.3033101558685303, "logps/rejected": -1.5694794654846191, "loss": 1.6318, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.3033101558685303, "rewards/margins": 0.2661692500114441, "rewards/rejected": -1.5694794654846191, "step": 4410 }, { "epoch": 2.362936945977588, "grad_norm": 7.564965929459961, "learning_rate": 1.310020503720254e-07, "logits/chosen": -0.16970126330852509, "logits/rejected": -0.03718472644686699, "logps/chosen": -1.2483030557632446, "logps/rejected": -1.4484217166900635, "loss": 1.5892, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2483030557632446, "rewards/margins": 0.20011882483959198, "rewards/rejected": -1.4484217166900635, "step": 4415 }, { "epoch": 2.36561297875899, "grad_norm": 8.4764314085826, "learning_rate": 1.2995289867125752e-07, "logits/chosen": -0.21752643585205078, "logits/rejected": -0.15322408080101013, "logps/chosen": -1.279433012008667, "logps/rejected": -1.4012091159820557, "loss": 1.6496, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.279433012008667, "rewards/margins": 0.12177594006061554, "rewards/rejected": -1.4012091159820557, "step": 4420 }, { "epoch": 2.368289011540391, "grad_norm": 8.181139885606376, "learning_rate": 1.2890733716908986e-07, "logits/chosen": -0.20343129336833954, "logits/rejected": -0.10383858531713486, "logps/chosen": -1.1922613382339478, "logps/rejected": -1.4458949565887451, "loss": 1.5287, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1922613382339478, "rewards/margins": 0.2536335587501526, "rewards/rejected": -1.4458949565887451, "step": 4425 }, { "epoch": 2.370965044321793, "grad_norm": 5.960231929117382, "learning_rate": 1.2786537600956454e-07, "logits/chosen": -0.19824449717998505, "logits/rejected": -0.05848120525479317, "logps/chosen": -1.2846362590789795, "logps/rejected": -1.5001429319381714, "loss": 1.6152, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2846362590789795, "rewards/margins": 0.21550679206848145, "rewards/rejected": -1.5001429319381714, "step": 4430 }, { "epoch": 2.3736410771031946, "grad_norm": 6.663144772697994, "learning_rate": 1.268270253017933e-07, "logits/chosen": -0.28592783212661743, "logits/rejected": -0.096594899892807, "logps/chosen": -1.1786837577819824, "logps/rejected": -1.4318368434906006, "loss": 1.5258, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1786837577819824, "rewards/margins": 0.253152996301651, "rewards/rejected": -1.4318368434906006, "step": 4435 }, { "epoch": 2.376317109884596, "grad_norm": 9.6216050200051, "learning_rate": 1.257922951198591e-07, "logits/chosen": -0.2888815701007843, "logits/rejected": -0.1167341023683548, "logps/chosen": -1.2764893770217896, "logps/rejected": -1.4089630842208862, "loss": 1.6323, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2764893770217896, "rewards/margins": 0.1324736624956131, "rewards/rejected": -1.4089630842208862, "step": 4440 }, { "epoch": 2.3789931426659976, "grad_norm": 8.165315173346368, "learning_rate": 1.24761195502719e-07, "logits/chosen": -0.2527289092540741, "logits/rejected": -0.10613594949245453, "logps/chosen": -1.2289607524871826, "logps/rejected": -1.4288841485977173, "loss": 1.6052, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2289607524871826, "rewards/margins": 0.19992344081401825, "rewards/rejected": -1.4288841485977173, "step": 4445 }, { "epoch": 2.3816691754473993, "grad_norm": 9.785709031069278, "learning_rate": 1.2373373645410573e-07, "logits/chosen": -0.18451814353466034, "logits/rejected": -0.07983697950839996, "logps/chosen": -1.2855627536773682, "logps/rejected": -1.500300407409668, "loss": 1.6164, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2855627536773682, "rewards/margins": 0.2147376537322998, "rewards/rejected": -1.500300407409668, "step": 4450 }, { "epoch": 2.384345208228801, "grad_norm": 8.058393238708456, "learning_rate": 1.2270992794243175e-07, "logits/chosen": -0.26786744594573975, "logits/rejected": -0.19314830005168915, "logps/chosen": -1.2687526941299438, "logps/rejected": -1.460627794265747, "loss": 1.6077, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2687526941299438, "rewards/margins": 0.19187510013580322, "rewards/rejected": -1.460627794265747, "step": 4455 }, { "epoch": 2.3870212410102023, "grad_norm": 7.4465935675716235, "learning_rate": 1.2168977990069147e-07, "logits/chosen": -0.27446383237838745, "logits/rejected": -0.0762770026922226, "logps/chosen": -1.187028169631958, "logps/rejected": -1.5143496990203857, "loss": 1.5082, "rewards/accuracies": 0.65625, "rewards/chosen": -1.187028169631958, "rewards/margins": 0.3273215591907501, "rewards/rejected": -1.5143496990203857, "step": 4460 }, { "epoch": 2.389697273791604, "grad_norm": 6.9486768670050285, "learning_rate": 1.206733022263659e-07, "logits/chosen": -0.27223438024520874, "logits/rejected": -0.11611995846033096, "logps/chosen": -1.287155270576477, "logps/rejected": -1.4335745573043823, "loss": 1.645, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.287155270576477, "rewards/margins": 0.146419495344162, "rewards/rejected": -1.4335745573043823, "step": 4465 }, { "epoch": 2.3923733065730053, "grad_norm": 8.014042276809453, "learning_rate": 1.1966050478132572e-07, "logits/chosen": -0.1054086685180664, "logits/rejected": -0.039462633430957794, "logps/chosen": -1.1409763097763062, "logps/rejected": -1.3532037734985352, "loss": 1.5266, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1409763097763062, "rewards/margins": 0.21222741901874542, "rewards/rejected": -1.3532037734985352, "step": 4470 }, { "epoch": 2.395049339354407, "grad_norm": 8.033512264834458, "learning_rate": 1.1865139739173635e-07, "logits/chosen": -0.2219352424144745, "logits/rejected": -0.0237773135304451, "logps/chosen": -1.2053747177124023, "logps/rejected": -1.429945707321167, "loss": 1.5515, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.2053747177124023, "rewards/margins": 0.22457103431224823, "rewards/rejected": -1.429945707321167, "step": 4475 }, { "epoch": 2.3977253721358087, "grad_norm": 8.917144958810724, "learning_rate": 1.1764598984796187e-07, "logits/chosen": -0.17329058051109314, "logits/rejected": -0.10808472335338593, "logps/chosen": -1.143885612487793, "logps/rejected": -1.339099407196045, "loss": 1.5354, "rewards/accuracies": 0.59375, "rewards/chosen": -1.143885612487793, "rewards/margins": 0.19521372020244598, "rewards/rejected": -1.339099407196045, "step": 4480 }, { "epoch": 2.4004014049172104, "grad_norm": 11.834309102099507, "learning_rate": 1.1664429190447095e-07, "logits/chosen": -0.2013109028339386, "logits/rejected": -0.1036062240600586, "logps/chosen": -1.2618813514709473, "logps/rejected": -1.518686056137085, "loss": 1.6, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2618813514709473, "rewards/margins": 0.2568047046661377, "rewards/rejected": -1.518686056137085, "step": 4485 }, { "epoch": 2.4030774376986117, "grad_norm": 7.175929463146946, "learning_rate": 1.1564631327974122e-07, "logits/chosen": -0.2383660525083542, "logits/rejected": -0.04933794587850571, "logps/chosen": -1.2206124067306519, "logps/rejected": -1.5133615732192993, "loss": 1.5399, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.2206124067306519, "rewards/margins": 0.2927490770816803, "rewards/rejected": -1.5133615732192993, "step": 4490 }, { "epoch": 2.4057534704800134, "grad_norm": 9.16204325508004, "learning_rate": 1.1465206365616587e-07, "logits/chosen": -0.2820783853530884, "logits/rejected": -0.09989559650421143, "logps/chosen": -1.2064526081085205, "logps/rejected": -1.4397554397583008, "loss": 1.5541, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2064526081085205, "rewards/margins": 0.23330290615558624, "rewards/rejected": -1.4397554397583008, "step": 4495 }, { "epoch": 2.408429503261415, "grad_norm": 6.78544051173504, "learning_rate": 1.1366155267995887e-07, "logits/chosen": -0.11139881610870361, "logits/rejected": -0.11971314251422882, "logps/chosen": -1.2364609241485596, "logps/rejected": -1.4335353374481201, "loss": 1.5937, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2364609241485596, "rewards/margins": 0.19707438349723816, "rewards/rejected": -1.4335353374481201, "step": 4500 }, { "epoch": 2.4111055360428164, "grad_norm": 10.310534197220848, "learning_rate": 1.1267478996106228e-07, "logits/chosen": -0.2633117437362671, "logits/rejected": -0.14126761257648468, "logps/chosen": -1.2213430404663086, "logps/rejected": -1.5242010354995728, "loss": 1.56, "rewards/accuracies": 0.625, "rewards/chosen": -1.2213430404663086, "rewards/margins": 0.3028578460216522, "rewards/rejected": -1.5242010354995728, "step": 4505 }, { "epoch": 2.413781568824218, "grad_norm": 11.1093160012141, "learning_rate": 1.116917850730521e-07, "logits/chosen": -0.24887220561504364, "logits/rejected": -0.14276175200939178, "logps/chosen": -1.1748237609863281, "logps/rejected": -1.3930141925811768, "loss": 1.5492, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1748237609863281, "rewards/margins": 0.21819040179252625, "rewards/rejected": -1.3930141925811768, "step": 4510 }, { "epoch": 2.41645760160562, "grad_norm": 6.210320759773411, "learning_rate": 1.1071254755304637e-07, "logits/chosen": -0.22284992039203644, "logits/rejected": -0.16558507084846497, "logps/chosen": -1.2325079441070557, "logps/rejected": -1.4816372394561768, "loss": 1.5629, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2325079441070557, "rewards/margins": 0.24912920594215393, "rewards/rejected": -1.4816372394561768, "step": 4515 }, { "epoch": 2.419133634387021, "grad_norm": 7.705672138397239, "learning_rate": 1.0973708690161143e-07, "logits/chosen": -0.20282363891601562, "logits/rejected": -0.13200493156909943, "logps/chosen": -1.209581732749939, "logps/rejected": -1.4997475147247314, "loss": 1.5331, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.209581732749939, "rewards/margins": 0.2901657223701477, "rewards/rejected": -1.4997475147247314, "step": 4520 }, { "epoch": 2.421809667168423, "grad_norm": 10.387820564546052, "learning_rate": 1.0876541258267119e-07, "logits/chosen": -0.2831111550331116, "logits/rejected": -0.13041360676288605, "logps/chosen": -1.3232536315917969, "logps/rejected": -1.5723609924316406, "loss": 1.6341, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3232536315917969, "rewards/margins": 0.24910736083984375, "rewards/rejected": -1.5723609924316406, "step": 4525 }, { "epoch": 2.4244856999498245, "grad_norm": 6.50545879373684, "learning_rate": 1.0779753402341379e-07, "logits/chosen": -0.242244154214859, "logits/rejected": -0.16366708278656006, "logps/chosen": -1.2527737617492676, "logps/rejected": -1.4390571117401123, "loss": 1.6024, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2527737617492676, "rewards/margins": 0.18628332018852234, "rewards/rejected": -1.4390571117401123, "step": 4530 }, { "epoch": 2.427161732731226, "grad_norm": 7.166247614143461, "learning_rate": 1.0683346061420157e-07, "logits/chosen": -0.0786890909075737, "logits/rejected": 0.01348264329135418, "logps/chosen": -1.2025784254074097, "logps/rejected": -1.4283499717712402, "loss": 1.5804, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2025784254074097, "rewards/margins": 0.22577135264873505, "rewards/rejected": -1.4283499717712402, "step": 4535 }, { "epoch": 2.4298377655126275, "grad_norm": 7.152108536625329, "learning_rate": 1.0587320170847874e-07, "logits/chosen": -0.1070355772972107, "logits/rejected": -0.047557950019836426, "logps/chosen": -1.1764482259750366, "logps/rejected": -1.3941954374313354, "loss": 1.5467, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.1764482259750366, "rewards/margins": 0.2177472561597824, "rewards/rejected": -1.3941954374313354, "step": 4540 }, { "epoch": 2.4325137982940293, "grad_norm": 7.238332196173785, "learning_rate": 1.0491676662268156e-07, "logits/chosen": -0.1882878541946411, "logits/rejected": -0.07006659358739853, "logps/chosen": -1.1832000017166138, "logps/rejected": -1.424682378768921, "loss": 1.5582, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1832000017166138, "rewards/margins": 0.24148237705230713, "rewards/rejected": -1.424682378768921, "step": 4545 }, { "epoch": 2.4351898310754305, "grad_norm": 10.485834168254481, "learning_rate": 1.0396416463614732e-07, "logits/chosen": -0.24194899201393127, "logits/rejected": -0.14754289388656616, "logps/chosen": -1.1658766269683838, "logps/rejected": -1.4418859481811523, "loss": 1.5452, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1658766269683838, "rewards/margins": 0.27600932121276855, "rewards/rejected": -1.4418859481811523, "step": 4550 }, { "epoch": 2.4378658638568322, "grad_norm": 7.914123241626193, "learning_rate": 1.0301540499102479e-07, "logits/chosen": -0.2195291817188263, "logits/rejected": -0.1631467640399933, "logps/chosen": -1.2823065519332886, "logps/rejected": -1.430878758430481, "loss": 1.6272, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2823065519332886, "rewards/margins": 0.14857208728790283, "rewards/rejected": -1.430878758430481, "step": 4555 }, { "epoch": 2.440541896638234, "grad_norm": 9.804187017833879, "learning_rate": 1.0207049689218405e-07, "logits/chosen": -0.2617951035499573, "logits/rejected": -0.1206180602312088, "logps/chosen": -1.1838595867156982, "logps/rejected": -1.3840501308441162, "loss": 1.5599, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1838595867156982, "rewards/margins": 0.20019057393074036, "rewards/rejected": -1.3840501308441162, "step": 4560 }, { "epoch": 2.4432179294196352, "grad_norm": 9.221087836615421, "learning_rate": 1.0112944950712782e-07, "logits/chosen": -0.1988077610731125, "logits/rejected": -0.140293687582016, "logps/chosen": -1.284519076347351, "logps/rejected": -1.474821925163269, "loss": 1.6212, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.284519076347351, "rewards/margins": 0.19030292332172394, "rewards/rejected": -1.474821925163269, "step": 4565 }, { "epoch": 2.445893962201037, "grad_norm": 6.782121730236154, "learning_rate": 1.0019227196590174e-07, "logits/chosen": -0.13662488758563995, "logits/rejected": -0.0032837330363690853, "logps/chosen": -1.2362960577011108, "logps/rejected": -1.4920551776885986, "loss": 1.5736, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2362960577011108, "rewards/margins": 0.255759060382843, "rewards/rejected": -1.4920551776885986, "step": 4570 }, { "epoch": 2.4485699949824387, "grad_norm": 7.190362233091777, "learning_rate": 9.925897336100664e-08, "logits/chosen": -0.13475404679775238, "logits/rejected": -0.09383302927017212, "logps/chosen": -1.2316467761993408, "logps/rejected": -1.4339395761489868, "loss": 1.5886, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2316467761993408, "rewards/margins": 0.20229271054267883, "rewards/rejected": -1.4339395761489868, "step": 4575 }, { "epoch": 2.45124602776384, "grad_norm": 8.377076003160667, "learning_rate": 9.832956274730946e-08, "logits/chosen": -0.2030809372663498, "logits/rejected": -0.1647244095802307, "logps/chosen": -1.179179072380066, "logps/rejected": -1.4651811122894287, "loss": 1.5232, "rewards/accuracies": 0.65625, "rewards/chosen": -1.179179072380066, "rewards/margins": 0.2860020101070404, "rewards/rejected": -1.4651811122894287, "step": 4580 }, { "epoch": 2.4539220605452416, "grad_norm": 6.6990802347524685, "learning_rate": 9.740404914195633e-08, "logits/chosen": -0.16281422972679138, "logits/rejected": -0.04782953858375549, "logps/chosen": -1.291107416152954, "logps/rejected": -1.4581815004348755, "loss": 1.6578, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.291107416152954, "rewards/margins": 0.16707420349121094, "rewards/rejected": -1.4581815004348755, "step": 4585 }, { "epoch": 2.4565980933266434, "grad_norm": 6.900910408002965, "learning_rate": 9.648244152428392e-08, "logits/chosen": -0.27961355447769165, "logits/rejected": -0.11388464272022247, "logps/chosen": -1.1816189289093018, "logps/rejected": -1.426766037940979, "loss": 1.5509, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1816189289093018, "rewards/margins": 0.24514737725257874, "rewards/rejected": -1.426766037940979, "step": 4590 }, { "epoch": 2.4592741261080446, "grad_norm": 7.585177745899499, "learning_rate": 9.556474883573379e-08, "logits/chosen": -0.23860666155815125, "logits/rejected": -0.13447125256061554, "logps/chosen": -1.2374709844589233, "logps/rejected": -1.509106993675232, "loss": 1.5831, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2374709844589233, "rewards/margins": 0.271636039018631, "rewards/rejected": -1.509106993675232, "step": 4595 }, { "epoch": 2.4619501588894463, "grad_norm": 8.26242453249571, "learning_rate": 9.465097997976412e-08, "logits/chosen": -0.21761243045330048, "logits/rejected": 0.014684316702187061, "logps/chosen": -1.2814009189605713, "logps/rejected": -1.5540639162063599, "loss": 1.6085, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2814009189605713, "rewards/margins": 0.2726629972457886, "rewards/rejected": -1.5540639162063599, "step": 4600 }, { "epoch": 2.464626191670848, "grad_norm": 8.565484930594367, "learning_rate": 9.374114382176457e-08, "logits/chosen": -0.15770654380321503, "logits/rejected": -0.048599086701869965, "logps/chosen": -1.258568525314331, "logps/rejected": -1.5116450786590576, "loss": 1.6018, "rewards/accuracies": 0.625, "rewards/chosen": -1.258568525314331, "rewards/margins": 0.2530765235424042, "rewards/rejected": -1.5116450786590576, "step": 4605 }, { "epoch": 2.46730222445225, "grad_norm": 8.024702890810318, "learning_rate": 9.283524918896945e-08, "logits/chosen": -0.20571458339691162, "logits/rejected": -0.06943124532699585, "logps/chosen": -1.2410836219787598, "logps/rejected": -1.5107513666152954, "loss": 1.5583, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.2410836219787598, "rewards/margins": 0.26966771483421326, "rewards/rejected": -1.5107513666152954, "step": 4610 }, { "epoch": 2.469978257233651, "grad_norm": 7.078771121377501, "learning_rate": 9.193330487037232e-08, "logits/chosen": -0.149086594581604, "logits/rejected": -0.0335305891931057, "logps/chosen": -1.2589304447174072, "logps/rejected": -1.5451947450637817, "loss": 1.5774, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2589304447174072, "rewards/margins": 0.28626421093940735, "rewards/rejected": -1.5451947450637817, "step": 4615 }, { "epoch": 2.4726542900150528, "grad_norm": 7.410281710233767, "learning_rate": 9.103531961664118e-08, "logits/chosen": -0.1585276871919632, "logits/rejected": 0.013647395186126232, "logps/chosen": -1.1524250507354736, "logps/rejected": -1.4461370706558228, "loss": 1.4992, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.1524250507354736, "rewards/margins": 0.29371196031570435, "rewards/rejected": -1.4461370706558228, "step": 4620 }, { "epoch": 2.475330322796454, "grad_norm": 6.4861215775128365, "learning_rate": 9.014130214003269e-08, "logits/chosen": -0.27106332778930664, "logits/rejected": -0.22946906089782715, "logps/chosen": -1.2726995944976807, "logps/rejected": -1.4708665609359741, "loss": 1.5953, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2726995944976807, "rewards/margins": 0.19816681742668152, "rewards/rejected": -1.4708665609359741, "step": 4625 }, { "epoch": 2.4780063555778558, "grad_norm": 11.593695563467303, "learning_rate": 8.925126111430848e-08, "logits/chosen": -0.18593397736549377, "logits/rejected": -0.11719465255737305, "logps/chosen": -1.1971778869628906, "logps/rejected": -1.4603909254074097, "loss": 1.5598, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1971778869628906, "rewards/margins": 0.26321300864219666, "rewards/rejected": -1.4603909254074097, "step": 4630 }, { "epoch": 2.4806823883592575, "grad_norm": 9.842887811694869, "learning_rate": 8.83652051746504e-08, "logits/chosen": -0.09734443575143814, "logits/rejected": 0.02725895121693611, "logps/chosen": -1.2012536525726318, "logps/rejected": -1.4989259243011475, "loss": 1.5646, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2012536525726318, "rewards/margins": 0.2976721227169037, "rewards/rejected": -1.4989259243011475, "step": 4635 }, { "epoch": 2.483358421140659, "grad_norm": 7.375338188343925, "learning_rate": 8.748314291757696e-08, "logits/chosen": -0.14858081936836243, "logits/rejected": -0.05358070880174637, "logps/chosen": -1.2337394952774048, "logps/rejected": -1.4716202020645142, "loss": 1.5705, "rewards/accuracies": 0.625, "rewards/chosen": -1.2337394952774048, "rewards/margins": 0.2378806620836258, "rewards/rejected": -1.4716202020645142, "step": 4640 }, { "epoch": 2.4860344539220605, "grad_norm": 9.083724144875998, "learning_rate": 8.660508290086032e-08, "logits/chosen": -0.15422284603118896, "logits/rejected": -0.032006729394197464, "logps/chosen": -1.211015224456787, "logps/rejected": -1.4706156253814697, "loss": 1.5634, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.211015224456787, "rewards/margins": 0.2596002519130707, "rewards/rejected": -1.4706156253814697, "step": 4645 }, { "epoch": 2.488710486703462, "grad_norm": 8.495728501232144, "learning_rate": 8.573103364344231e-08, "logits/chosen": -0.19348394870758057, "logits/rejected": 0.028809908777475357, "logps/chosen": -1.2423301935195923, "logps/rejected": -1.5411301851272583, "loss": 1.5551, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2423301935195923, "rewards/margins": 0.298799991607666, "rewards/rejected": -1.5411301851272583, "step": 4650 }, { "epoch": 2.4913865194848634, "grad_norm": 7.953930669088138, "learning_rate": 8.486100362535292e-08, "logits/chosen": -0.21532246470451355, "logits/rejected": -0.08617198467254639, "logps/chosen": -1.2144615650177002, "logps/rejected": -1.3963156938552856, "loss": 1.5693, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2144615650177002, "rewards/margins": 0.18185389041900635, "rewards/rejected": -1.3963156938552856, "step": 4655 }, { "epoch": 2.494062552266265, "grad_norm": 6.625260423362665, "learning_rate": 8.399500128762693e-08, "logits/chosen": -0.20581689476966858, "logits/rejected": -0.09944789111614227, "logps/chosen": -1.2891731262207031, "logps/rejected": -1.4467524290084839, "loss": 1.6299, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2891731262207031, "rewards/margins": 0.15757936239242554, "rewards/rejected": -1.4467524290084839, "step": 4660 }, { "epoch": 2.496738585047667, "grad_norm": 7.384548483424394, "learning_rate": 8.313303503222313e-08, "logits/chosen": -0.1977105438709259, "logits/rejected": -0.14704328775405884, "logps/chosen": -1.3137004375457764, "logps/rejected": -1.5258251428604126, "loss": 1.645, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3137004375457764, "rewards/margins": 0.21212443709373474, "rewards/rejected": -1.5258251428604126, "step": 4665 }, { "epoch": 2.4994146178290686, "grad_norm": 8.198769766484682, "learning_rate": 8.227511322194164e-08, "logits/chosen": -0.18010950088500977, "logits/rejected": -0.06886539608240128, "logps/chosen": -1.280510663986206, "logps/rejected": -1.4507843255996704, "loss": 1.6213, "rewards/accuracies": 0.59375, "rewards/chosen": -1.280510663986206, "rewards/margins": 0.17027351260185242, "rewards/rejected": -1.4507843255996704, "step": 4670 }, { "epoch": 2.50209065061047, "grad_norm": 6.855115745421152, "learning_rate": 8.142124418034385e-08, "logits/chosen": -0.1507149189710617, "logits/rejected": -0.026994511485099792, "logps/chosen": -1.1871156692504883, "logps/rejected": -1.4419450759887695, "loss": 1.5485, "rewards/accuracies": 0.5625, "rewards/chosen": -1.1871156692504883, "rewards/margins": 0.25482943654060364, "rewards/rejected": -1.4419450759887695, "step": 4675 }, { "epoch": 2.5047666833918716, "grad_norm": 8.982655143567454, "learning_rate": 8.057143619167073e-08, "logits/chosen": -0.14716149866580963, "logits/rejected": -0.050116412341594696, "logps/chosen": -1.233823537826538, "logps/rejected": -1.4536529779434204, "loss": 1.587, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.233823537826538, "rewards/margins": 0.2198294699192047, "rewards/rejected": -1.4536529779434204, "step": 4680 }, { "epoch": 2.507442716173273, "grad_norm": 6.2140511328578905, "learning_rate": 7.97256975007633e-08, "logits/chosen": -0.2550865113735199, "logits/rejected": -0.09868437051773071, "logps/chosen": -1.2446149587631226, "logps/rejected": -1.4486770629882812, "loss": 1.5898, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2446149587631226, "rewards/margins": 0.20406225323677063, "rewards/rejected": -1.4486770629882812, "step": 4685 }, { "epoch": 2.5101187489546746, "grad_norm": 9.244400120324464, "learning_rate": 7.888403631298186e-08, "logits/chosen": -0.14784127473831177, "logits/rejected": -0.11010942608118057, "logps/chosen": -1.2629327774047852, "logps/rejected": -1.443296194076538, "loss": 1.6106, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2629327774047852, "rewards/margins": 0.18036355078220367, "rewards/rejected": -1.443296194076538, "step": 4690 }, { "epoch": 2.5127947817360763, "grad_norm": 8.161542062031891, "learning_rate": 7.804646079412719e-08, "logits/chosen": -0.1550903618335724, "logits/rejected": 0.014736851677298546, "logps/chosen": -1.2535508871078491, "logps/rejected": -1.4470226764678955, "loss": 1.5888, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2535508871078491, "rewards/margins": 0.19347189366817474, "rewards/rejected": -1.4470226764678955, "step": 4695 }, { "epoch": 2.515470814517478, "grad_norm": 7.057383355262316, "learning_rate": 7.72129790703604e-08, "logits/chosen": -0.2678453326225281, "logits/rejected": -0.17103447020053864, "logps/chosen": -1.2096585035324097, "logps/rejected": -1.4120450019836426, "loss": 1.5652, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2096585035324097, "rewards/margins": 0.20238666236400604, "rewards/rejected": -1.4120450019836426, "step": 4700 }, { "epoch": 2.5181468472988793, "grad_norm": 8.83446443379937, "learning_rate": 7.638359922812504e-08, "logits/chosen": -0.11662398278713226, "logits/rejected": -0.09670352190732956, "logps/chosen": -1.2910524606704712, "logps/rejected": -1.4886689186096191, "loss": 1.6368, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2910524606704712, "rewards/margins": 0.19761626422405243, "rewards/rejected": -1.4886689186096191, "step": 4705 }, { "epoch": 2.520822880080281, "grad_norm": 10.996392501015148, "learning_rate": 7.555832931406774e-08, "logits/chosen": -0.19205592572689056, "logits/rejected": -0.06259692460298538, "logps/chosen": -1.2577614784240723, "logps/rejected": -1.4748903512954712, "loss": 1.5913, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2577614784240723, "rewards/margins": 0.21712878346443176, "rewards/rejected": -1.4748903512954712, "step": 4710 }, { "epoch": 2.5234989128616827, "grad_norm": 7.080472540676264, "learning_rate": 7.47371773349611e-08, "logits/chosen": -0.2080480307340622, "logits/rejected": -0.1923224925994873, "logps/chosen": -1.3185064792633057, "logps/rejected": -1.525776982307434, "loss": 1.6493, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3185064792633057, "rewards/margins": 0.2072705775499344, "rewards/rejected": -1.525776982307434, "step": 4715 }, { "epoch": 2.526174945643084, "grad_norm": 9.847853778010714, "learning_rate": 7.392015125762496e-08, "logits/chosen": -0.14937856793403625, "logits/rejected": -0.07092615962028503, "logps/chosen": -1.1753498315811157, "logps/rejected": -1.4301412105560303, "loss": 1.5205, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1753498315811157, "rewards/margins": 0.2547914385795593, "rewards/rejected": -1.4301412105560303, "step": 4720 }, { "epoch": 2.5288509784244857, "grad_norm": 8.166106113851754, "learning_rate": 7.310725900885018e-08, "logits/chosen": -0.2096148282289505, "logits/rejected": -0.1626475304365158, "logps/chosen": -1.258299469947815, "logps/rejected": -1.4184331893920898, "loss": 1.6108, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.258299469947815, "rewards/margins": 0.16013376414775848, "rewards/rejected": -1.4184331893920898, "step": 4725 }, { "epoch": 2.5315270112058874, "grad_norm": 8.605523645573196, "learning_rate": 7.229850847532076e-08, "logits/chosen": -0.17315617203712463, "logits/rejected": -0.07473836839199066, "logps/chosen": -1.1580817699432373, "logps/rejected": -1.3972079753875732, "loss": 1.5282, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1580817699432373, "rewards/margins": 0.23912616074085236, "rewards/rejected": -1.3972079753875732, "step": 4730 }, { "epoch": 2.5342030439872887, "grad_norm": 7.252170164669517, "learning_rate": 7.149390750353779e-08, "logits/chosen": -0.1420806497335434, "logits/rejected": -0.2054506242275238, "logps/chosen": -1.2775603532791138, "logps/rejected": -1.4530556201934814, "loss": 1.6186, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2775603532791138, "rewards/margins": 0.17549541592597961, "rewards/rejected": -1.4530556201934814, "step": 4735 }, { "epoch": 2.5368790767686904, "grad_norm": 9.076334442762303, "learning_rate": 7.069346389974374e-08, "logits/chosen": -0.24810704588890076, "logits/rejected": -0.12928973138332367, "logps/chosen": -1.2737659215927124, "logps/rejected": -1.5296587944030762, "loss": 1.6067, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2737659215927124, "rewards/margins": 0.2558927536010742, "rewards/rejected": -1.5296587944030762, "step": 4740 }, { "epoch": 2.539555109550092, "grad_norm": 7.678519243284076, "learning_rate": 6.989718542984563e-08, "logits/chosen": -0.19102391600608826, "logits/rejected": -0.15975889563560486, "logps/chosen": -1.2788324356079102, "logps/rejected": -1.504441499710083, "loss": 1.6004, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2788324356079102, "rewards/margins": 0.22560922801494598, "rewards/rejected": -1.504441499710083, "step": 4745 }, { "epoch": 2.5422311423314934, "grad_norm": 6.267154014115284, "learning_rate": 6.9105079819341e-08, "logits/chosen": -0.12780414521694183, "logits/rejected": 0.04071787744760513, "logps/chosen": -1.2187215089797974, "logps/rejected": -1.6234019994735718, "loss": 1.5236, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.2187215089797974, "rewards/margins": 0.40468043088912964, "rewards/rejected": -1.6234019994735718, "step": 4750 }, { "epoch": 2.544907175112895, "grad_norm": 6.474135443347511, "learning_rate": 6.831715475324163e-08, "logits/chosen": -0.18726108968257904, "logits/rejected": -0.04195536673069, "logps/chosen": -1.1510288715362549, "logps/rejected": -1.485931158065796, "loss": 1.4969, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1510288715362549, "rewards/margins": 0.3349021375179291, "rewards/rejected": -1.485931158065796, "step": 4755 }, { "epoch": 2.547583207894297, "grad_norm": 7.059864002376292, "learning_rate": 6.753341787600026e-08, "logits/chosen": -0.250051885843277, "logits/rejected": -0.12963862717151642, "logps/chosen": -1.1672437191009521, "logps/rejected": -1.3937098979949951, "loss": 1.5341, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1672437191009521, "rewards/margins": 0.22646614909172058, "rewards/rejected": -1.3937098979949951, "step": 4760 }, { "epoch": 2.5502592406756985, "grad_norm": 8.424779242777895, "learning_rate": 6.67538767914353e-08, "logits/chosen": -0.272745817899704, "logits/rejected": -0.11777077615261078, "logps/chosen": -1.2558521032333374, "logps/rejected": -1.4366075992584229, "loss": 1.6214, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2558521032333374, "rewards/margins": 0.18075552582740784, "rewards/rejected": -1.4366075992584229, "step": 4765 }, { "epoch": 2.5529352734571, "grad_norm": 8.11145710217, "learning_rate": 6.597853906265793e-08, "logits/chosen": -0.22406180202960968, "logits/rejected": -0.14268115162849426, "logps/chosen": -1.2580957412719727, "logps/rejected": -1.5128915309906006, "loss": 1.5874, "rewards/accuracies": 0.625, "rewards/chosen": -1.2580957412719727, "rewards/margins": 0.25479573011398315, "rewards/rejected": -1.5128915309906006, "step": 4770 }, { "epoch": 2.5556113062385015, "grad_norm": 5.545354795700497, "learning_rate": 6.5207412211998e-08, "logits/chosen": -0.06541035324335098, "logits/rejected": 0.02902628853917122, "logps/chosen": -1.1863327026367188, "logps/rejected": -1.4349019527435303, "loss": 1.5328, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.1863327026367188, "rewards/margins": 0.2485692948102951, "rewards/rejected": -1.4349019527435303, "step": 4775 }, { "epoch": 2.558287339019903, "grad_norm": 5.089746429603335, "learning_rate": 6.444050372093186e-08, "logits/chosen": -0.16423115134239197, "logits/rejected": -0.09638829529285431, "logps/chosen": -1.2353355884552002, "logps/rejected": -1.4341075420379639, "loss": 1.5856, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2353355884552002, "rewards/margins": 0.19877199828624725, "rewards/rejected": -1.4341075420379639, "step": 4780 }, { "epoch": 2.5609633718013045, "grad_norm": 10.491530025375786, "learning_rate": 6.367782103000873e-08, "logits/chosen": -0.15491080284118652, "logits/rejected": -0.10681736469268799, "logps/chosen": -1.2595205307006836, "logps/rejected": -1.4247640371322632, "loss": 1.6131, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2595205307006836, "rewards/margins": 0.1652435064315796, "rewards/rejected": -1.4247640371322632, "step": 4785 }, { "epoch": 2.5636394045827062, "grad_norm": 6.394223585792894, "learning_rate": 6.29193715387798e-08, "logits/chosen": -0.2331111878156662, "logits/rejected": -0.10800018161535263, "logps/chosen": -1.2580963373184204, "logps/rejected": -1.5037890672683716, "loss": 1.5824, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2580963373184204, "rewards/margins": 0.24569272994995117, "rewards/rejected": -1.5037890672683716, "step": 4790 }, { "epoch": 2.566315437364108, "grad_norm": 11.05058571995248, "learning_rate": 6.216516260572502e-08, "logits/chosen": -0.14371183514595032, "logits/rejected": -0.09054752439260483, "logps/chosen": -1.2728910446166992, "logps/rejected": -1.4731087684631348, "loss": 1.6238, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2728910446166992, "rewards/margins": 0.20021755993366241, "rewards/rejected": -1.4731087684631348, "step": 4795 }, { "epoch": 2.568991470145509, "grad_norm": 8.411984021616512, "learning_rate": 6.141520154818297e-08, "logits/chosen": -0.19774667918682098, "logits/rejected": -0.10813900083303452, "logps/chosen": -1.1911413669586182, "logps/rejected": -1.4454268217086792, "loss": 1.5444, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.1911413669586182, "rewards/margins": 0.25428539514541626, "rewards/rejected": -1.4454268217086792, "step": 4800 }, { "epoch": 2.568991470145509, "eval_logits/chosen": 0.069010429084301, "eval_logits/rejected": 0.1403309553861618, "eval_logps/chosen": -1.303430199623108, "eval_logps/rejected": -1.5013798475265503, "eval_loss": 1.6473349332809448, "eval_rewards/accuracies": 0.5719584822654724, "eval_rewards/chosen": -1.303430199623108, "eval_rewards/margins": 0.19794969260692596, "eval_rewards/rejected": -1.5013798475265503, "eval_runtime": 40.9803, "eval_samples_per_second": 32.821, "eval_steps_per_second": 8.223, "step": 4800 }, { "epoch": 2.571667502926911, "grad_norm": 9.907133386402505, "learning_rate": 6.066949564227897e-08, "logits/chosen": -0.2680151164531708, "logits/rejected": -0.1689799576997757, "logps/chosen": -1.2977831363677979, "logps/rejected": -1.565671682357788, "loss": 1.6163, "rewards/accuracies": 0.625, "rewards/chosen": -1.2977831363677979, "rewards/margins": 0.2678885757923126, "rewards/rejected": -1.565671682357788, "step": 4805 }, { "epoch": 2.574343535708312, "grad_norm": 6.806730281339653, "learning_rate": 5.992805212285523e-08, "logits/chosen": -0.14087334275245667, "logits/rejected": -0.02274361439049244, "logps/chosen": -1.3015844821929932, "logps/rejected": -1.5247794389724731, "loss": 1.6321, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.3015844821929932, "rewards/margins": 0.223194882273674, "rewards/rejected": -1.5247794389724731, "step": 4810 }, { "epoch": 2.577019568489714, "grad_norm": 8.51556092180203, "learning_rate": 5.9190878183399684e-08, "logits/chosen": -0.1476089209318161, "logits/rejected": -0.04703466221690178, "logps/chosen": -1.146867036819458, "logps/rejected": -1.5179466009140015, "loss": 1.4784, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.146867036819458, "rewards/margins": 0.37107959389686584, "rewards/rejected": -1.5179466009140015, "step": 4815 }, { "epoch": 2.5796956012711156, "grad_norm": 8.72271523604443, "learning_rate": 5.845798097597748e-08, "logits/chosen": -0.14900721609592438, "logits/rejected": -0.0796498954296112, "logps/chosen": -1.3060250282287598, "logps/rejected": -1.4967185258865356, "loss": 1.6296, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3060250282287598, "rewards/margins": 0.1906934529542923, "rewards/rejected": -1.4967185258865356, "step": 4820 }, { "epoch": 2.5823716340525174, "grad_norm": 10.002189649313841, "learning_rate": 5.772936761116026e-08, "logits/chosen": -0.14325377345085144, "logits/rejected": -0.05125095695257187, "logps/chosen": -1.20247220993042, "logps/rejected": -1.3964070081710815, "loss": 1.5706, "rewards/accuracies": 0.5625, "rewards/chosen": -1.20247220993042, "rewards/margins": 0.19393488764762878, "rewards/rejected": -1.3964070081710815, "step": 4825 }, { "epoch": 2.5850476668339186, "grad_norm": 5.577297169590059, "learning_rate": 5.700504515795829e-08, "logits/chosen": -0.18866679072380066, "logits/rejected": -0.06108071282505989, "logps/chosen": -1.287221908569336, "logps/rejected": -1.4399975538253784, "loss": 1.6316, "rewards/accuracies": 0.5625, "rewards/chosen": -1.287221908569336, "rewards/margins": 0.1527756154537201, "rewards/rejected": -1.4399975538253784, "step": 4830 }, { "epoch": 2.5877236996153203, "grad_norm": 10.808848370754, "learning_rate": 5.628502064375101e-08, "logits/chosen": -0.3119337558746338, "logits/rejected": -0.15793856978416443, "logps/chosen": -1.2294279336929321, "logps/rejected": -1.4302443265914917, "loss": 1.5851, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2294279336929321, "rewards/margins": 0.2008165419101715, "rewards/rejected": -1.4302443265914917, "step": 4835 }, { "epoch": 2.5903997323967216, "grad_norm": 17.377615026752256, "learning_rate": 5.55693010542197e-08, "logits/chosen": -0.26505112648010254, "logits/rejected": -0.11424796283245087, "logps/chosen": -1.2466034889221191, "logps/rejected": -1.443650484085083, "loss": 1.5895, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2466034889221191, "rewards/margins": 0.19704698026180267, "rewards/rejected": -1.443650484085083, "step": 4840 }, { "epoch": 2.5930757651781233, "grad_norm": 6.862180399559727, "learning_rate": 5.485789333327856e-08, "logits/chosen": -0.23199248313903809, "logits/rejected": -0.1249762773513794, "logps/chosen": -1.20241379737854, "logps/rejected": -1.4607778787612915, "loss": 1.531, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.20241379737854, "rewards/margins": 0.25836381316185, "rewards/rejected": -1.4607778787612915, "step": 4845 }, { "epoch": 2.595751797959525, "grad_norm": 11.417551410862856, "learning_rate": 5.4150804383008675e-08, "logits/chosen": -0.3245030343532562, "logits/rejected": -0.18769951164722443, "logps/chosen": -1.1993231773376465, "logps/rejected": -1.4456372261047363, "loss": 1.5505, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1993231773376465, "rewards/margins": 0.24631407856941223, "rewards/rejected": -1.4456372261047363, "step": 4850 }, { "epoch": 2.5984278307409268, "grad_norm": 10.991552356367377, "learning_rate": 5.344804106359002e-08, "logits/chosen": -0.1707092821598053, "logits/rejected": -0.040122199803590775, "logps/chosen": -1.1905627250671387, "logps/rejected": -1.3950374126434326, "loss": 1.5777, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.1905627250671387, "rewards/margins": 0.204474538564682, "rewards/rejected": -1.3950374126434326, "step": 4855 }, { "epoch": 2.601103863522328, "grad_norm": 10.67200349867524, "learning_rate": 5.274961019323559e-08, "logits/chosen": -0.20811323821544647, "logits/rejected": -0.15802529454231262, "logps/chosen": -1.1508222818374634, "logps/rejected": -1.5101139545440674, "loss": 1.4898, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.1508222818374634, "rewards/margins": 0.3592917025089264, "rewards/rejected": -1.5101139545440674, "step": 4860 }, { "epoch": 2.6037798963037297, "grad_norm": 7.816821358452244, "learning_rate": 5.205551854812451e-08, "logits/chosen": -0.2871929407119751, "logits/rejected": -0.21601462364196777, "logps/chosen": -1.2162559032440186, "logps/rejected": -1.519219994544983, "loss": 1.5526, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2162559032440186, "rewards/margins": 0.3029640316963196, "rewards/rejected": -1.519219994544983, "step": 4865 }, { "epoch": 2.606455929085131, "grad_norm": 10.124198148417449, "learning_rate": 5.1365772862337177e-08, "logits/chosen": -0.16407471895217896, "logits/rejected": -0.0544242337346077, "logps/chosen": -1.2421973943710327, "logps/rejected": -1.4441970586776733, "loss": 1.5994, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2421973943710327, "rewards/margins": 0.20199963450431824, "rewards/rejected": -1.4441970586776733, "step": 4870 }, { "epoch": 2.6091319618665327, "grad_norm": 8.284000148873433, "learning_rate": 5.068037982778905e-08, "logits/chosen": -0.03732685372233391, "logits/rejected": 0.022808223962783813, "logps/chosen": -1.2028716802597046, "logps/rejected": -1.5486289262771606, "loss": 1.5438, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.2028716802597046, "rewards/margins": 0.34575721621513367, "rewards/rejected": -1.5486289262771606, "step": 4875 }, { "epoch": 2.6118079946479344, "grad_norm": 9.236354556919212, "learning_rate": 4.999934609416656e-08, "logits/chosen": -0.11083243042230606, "logits/rejected": -0.01954919472336769, "logps/chosen": -1.1937742233276367, "logps/rejected": -1.4890735149383545, "loss": 1.5265, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1937742233276367, "rewards/margins": 0.29529935121536255, "rewards/rejected": -1.4890735149383545, "step": 4880 }, { "epoch": 2.614484027429336, "grad_norm": 8.007453309916604, "learning_rate": 4.932267826886183e-08, "logits/chosen": -0.12022782862186432, "logits/rejected": -0.05459488555788994, "logps/chosen": -1.3056491613388062, "logps/rejected": -1.5686249732971191, "loss": 1.6349, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3056491613388062, "rewards/margins": 0.26297587156295776, "rewards/rejected": -1.5686249732971191, "step": 4885 }, { "epoch": 2.6171600602107374, "grad_norm": 9.567757022684548, "learning_rate": 4.8650382916909206e-08, "logits/chosen": -0.2669495940208435, "logits/rejected": -0.12905225157737732, "logps/chosen": -1.235668420791626, "logps/rejected": -1.5042340755462646, "loss": 1.5718, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.235668420791626, "rewards/margins": 0.268565833568573, "rewards/rejected": -1.5042340755462646, "step": 4890 }, { "epoch": 2.619836092992139, "grad_norm": 6.9529581708975625, "learning_rate": 4.7982466560920976e-08, "logits/chosen": -0.19365879893302917, "logits/rejected": -0.11342108249664307, "logps/chosen": -1.2696317434310913, "logps/rejected": -1.4244247674942017, "loss": 1.6236, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2696317434310913, "rewards/margins": 0.1547931283712387, "rewards/rejected": -1.4244247674942017, "step": 4895 }, { "epoch": 2.622512125773541, "grad_norm": 6.010126890420108, "learning_rate": 4.7318935681024685e-08, "logits/chosen": -0.14299534261226654, "logits/rejected": -0.024873051792383194, "logps/chosen": -1.2177270650863647, "logps/rejected": -1.4972065687179565, "loss": 1.5518, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2177270650863647, "rewards/margins": 0.2794795632362366, "rewards/rejected": -1.4972065687179565, "step": 4900 }, { "epoch": 2.625188158554942, "grad_norm": 6.432348217455601, "learning_rate": 4.6659796714799745e-08, "logits/chosen": -0.19225557148456573, "logits/rejected": -0.06428401172161102, "logps/chosen": -1.2267589569091797, "logps/rejected": -1.492279291152954, "loss": 1.5688, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2267589569091797, "rewards/margins": 0.2655201852321625, "rewards/rejected": -1.492279291152954, "step": 4905 }, { "epoch": 2.627864191336344, "grad_norm": 7.250219709550545, "learning_rate": 4.60050560572155e-08, "logits/chosen": -0.2531941831111908, "logits/rejected": -0.2812545895576477, "logps/chosen": -1.2359212636947632, "logps/rejected": -1.5699865818023682, "loss": 1.5642, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2359212636947632, "rewards/margins": 0.33406534790992737, "rewards/rejected": -1.5699865818023682, "step": 4910 }, { "epoch": 2.6305402241177456, "grad_norm": 8.31195879409968, "learning_rate": 4.535472006056834e-08, "logits/chosen": -0.16209043562412262, "logits/rejected": -0.08424313366413116, "logps/chosen": -1.151822566986084, "logps/rejected": -1.447588324546814, "loss": 1.4963, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.151822566986084, "rewards/margins": 0.2957659363746643, "rewards/rejected": -1.447588324546814, "step": 4915 }, { "epoch": 2.6332162568991473, "grad_norm": 9.425398992684931, "learning_rate": 4.470879503442132e-08, "logits/chosen": -0.1478452831506729, "logits/rejected": -0.09613370895385742, "logps/chosen": -1.2299785614013672, "logps/rejected": -1.4136711359024048, "loss": 1.5997, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2299785614013672, "rewards/margins": 0.18369272351264954, "rewards/rejected": -1.4136711359024048, "step": 4920 }, { "epoch": 2.6358922896805486, "grad_norm": 8.625771505341113, "learning_rate": 4.406728724554154e-08, "logits/chosen": -0.3559058904647827, "logits/rejected": -0.14755351841449738, "logps/chosen": -1.1982418298721313, "logps/rejected": -1.4626787900924683, "loss": 1.5445, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1982418298721313, "rewards/margins": 0.26443710923194885, "rewards/rejected": -1.4626787900924683, "step": 4925 }, { "epoch": 2.6385683224619503, "grad_norm": 6.369269401234845, "learning_rate": 4.3430202917840664e-08, "logits/chosen": -0.12286825478076935, "logits/rejected": -0.0016535192262381315, "logps/chosen": -1.274578332901001, "logps/rejected": -1.5974867343902588, "loss": 1.5924, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.274578332901001, "rewards/margins": 0.32290831208229065, "rewards/rejected": -1.5974867343902588, "step": 4930 }, { "epoch": 2.6412443552433515, "grad_norm": 10.84824946742521, "learning_rate": 4.279754823231346e-08, "logits/chosen": -0.25615420937538147, "logits/rejected": -0.1267656832933426, "logps/chosen": -1.2327163219451904, "logps/rejected": -1.4325826168060303, "loss": 1.5876, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2327163219451904, "rewards/margins": 0.19986644387245178, "rewards/rejected": -1.4325826168060303, "step": 4935 }, { "epoch": 2.6439203880247533, "grad_norm": 6.695587205505035, "learning_rate": 4.216932932697859e-08, "logits/chosen": -0.19970975816249847, "logits/rejected": -0.14457407593727112, "logps/chosen": -1.2329965829849243, "logps/rejected": -1.37167489528656, "loss": 1.5926, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2329965829849243, "rewards/margins": 0.13867846131324768, "rewards/rejected": -1.37167489528656, "step": 4940 }, { "epoch": 2.646596420806155, "grad_norm": 7.353167999405972, "learning_rate": 4.154555229681844e-08, "logits/chosen": -0.18768291175365448, "logits/rejected": -0.010452374815940857, "logps/chosen": -1.2278202772140503, "logps/rejected": -1.536572813987732, "loss": 1.5483, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2278202772140503, "rewards/margins": 0.3087523877620697, "rewards/rejected": -1.536572813987732, "step": 4945 }, { "epoch": 2.6492724535875567, "grad_norm": 12.227792809999936, "learning_rate": 4.092622319372069e-08, "logits/chosen": -0.22052931785583496, "logits/rejected": -0.12232838571071625, "logps/chosen": -1.1905124187469482, "logps/rejected": -1.4453401565551758, "loss": 1.5503, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1905124187469482, "rewards/margins": 0.2548278272151947, "rewards/rejected": -1.4453401565551758, "step": 4950 }, { "epoch": 2.651948486368958, "grad_norm": 8.578760957784235, "learning_rate": 4.031134802641889e-08, "logits/chosen": -0.21375855803489685, "logits/rejected": -0.22308556735515594, "logps/chosen": -1.2271889448165894, "logps/rejected": -1.5128446817398071, "loss": 1.5502, "rewards/accuracies": 0.625, "rewards/chosen": -1.2271889448165894, "rewards/margins": 0.2856557071208954, "rewards/rejected": -1.5128446817398071, "step": 4955 }, { "epoch": 2.6546245191503597, "grad_norm": 7.265117101221275, "learning_rate": 3.970093276043468e-08, "logits/chosen": -0.11913114786148071, "logits/rejected": -0.041926462203264236, "logps/chosen": -1.2388899326324463, "logps/rejected": -1.4442875385284424, "loss": 1.5851, "rewards/accuracies": 0.625, "rewards/chosen": -1.2388899326324463, "rewards/margins": 0.20539765059947968, "rewards/rejected": -1.4442875385284424, "step": 4960 }, { "epoch": 2.657300551931761, "grad_norm": 8.696253922209147, "learning_rate": 3.9094983318019584e-08, "logits/chosen": -0.23500557243824005, "logits/rejected": -0.12808677554130554, "logps/chosen": -1.140141487121582, "logps/rejected": -1.3926841020584106, "loss": 1.4881, "rewards/accuracies": 0.65625, "rewards/chosen": -1.140141487121582, "rewards/margins": 0.2525426745414734, "rewards/rejected": -1.3926841020584106, "step": 4965 }, { "epoch": 2.6599765847131627, "grad_norm": 11.446705748719662, "learning_rate": 3.849350557809789e-08, "logits/chosen": -0.13588054478168488, "logits/rejected": -0.08402448147535324, "logps/chosen": -1.1959439516067505, "logps/rejected": -1.4947633743286133, "loss": 1.5293, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1959439516067505, "rewards/margins": 0.29881924390792847, "rewards/rejected": -1.4947633743286133, "step": 4970 }, { "epoch": 2.6626526174945644, "grad_norm": 8.997039686082442, "learning_rate": 3.789650537620903e-08, "logits/chosen": -0.21160972118377686, "logits/rejected": -0.17673712968826294, "logps/chosen": -1.2614823579788208, "logps/rejected": -1.4655091762542725, "loss": 1.6031, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2614823579788208, "rewards/margins": 0.2040266990661621, "rewards/rejected": -1.4655091762542725, "step": 4975 }, { "epoch": 2.665328650275966, "grad_norm": 10.466268389357104, "learning_rate": 3.730398850445182e-08, "logits/chosen": -0.09437865763902664, "logits/rejected": -0.035064686089754105, "logps/chosen": -1.3431638479232788, "logps/rejected": -1.5557583570480347, "loss": 1.6559, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.3431638479232788, "rewards/margins": 0.21259450912475586, "rewards/rejected": -1.5557583570480347, "step": 4980 }, { "epoch": 2.6680046830573674, "grad_norm": 11.519904973680102, "learning_rate": 3.671596071142735e-08, "logits/chosen": -0.14721126854419708, "logits/rejected": -0.006001046393066645, "logps/chosen": -1.1870566606521606, "logps/rejected": -1.4658606052398682, "loss": 1.5466, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1870566606521606, "rewards/margins": 0.2788039743900299, "rewards/rejected": -1.4658606052398682, "step": 4985 }, { "epoch": 2.670680715838769, "grad_norm": 10.178088788076465, "learning_rate": 3.6132427702183996e-08, "logits/chosen": -0.26032769680023193, "logits/rejected": -0.0708533450961113, "logps/chosen": -1.1861783266067505, "logps/rejected": -1.4879579544067383, "loss": 1.5326, "rewards/accuracies": 0.625, "rewards/chosen": -1.1861783266067505, "rewards/margins": 0.301779568195343, "rewards/rejected": -1.4879579544067383, "step": 4990 }, { "epoch": 2.6733567486201704, "grad_norm": 9.76883783497797, "learning_rate": 3.555339513816147e-08, "logits/chosen": -0.23129093647003174, "logits/rejected": -0.223618745803833, "logps/chosen": -1.2377936840057373, "logps/rejected": -1.4420830011367798, "loss": 1.5989, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2377936840057373, "rewards/margins": 0.20428943634033203, "rewards/rejected": -1.4420830011367798, "step": 4995 }, { "epoch": 2.676032781401572, "grad_norm": 8.338276783853692, "learning_rate": 3.497886863713639e-08, "logits/chosen": -0.24142029881477356, "logits/rejected": -0.23088951408863068, "logps/chosen": -1.2321035861968994, "logps/rejected": -1.5279680490493774, "loss": 1.5747, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2321035861968994, "rewards/margins": 0.295864462852478, "rewards/rejected": -1.5279680490493774, "step": 5000 }, { "epoch": 2.678708814182974, "grad_norm": 9.870028644774022, "learning_rate": 3.440885377316721e-08, "logits/chosen": -0.1602504551410675, "logits/rejected": -0.12937873601913452, "logps/chosen": -1.220651388168335, "logps/rejected": -1.4318891763687134, "loss": 1.562, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.220651388168335, "rewards/margins": 0.21123778820037842, "rewards/rejected": -1.4318891763687134, "step": 5005 }, { "epoch": 2.6813848469643755, "grad_norm": 9.349431556525046, "learning_rate": 3.384335607654082e-08, "logits/chosen": -0.13725784420967102, "logits/rejected": -0.06370799988508224, "logps/chosen": -1.3512427806854248, "logps/rejected": -1.5464354753494263, "loss": 1.6933, "rewards/accuracies": 0.5625, "rewards/chosen": -1.3512427806854248, "rewards/margins": 0.1951926350593567, "rewards/rejected": -1.5464354753494263, "step": 5010 }, { "epoch": 2.684060879745777, "grad_norm": 10.099696408988086, "learning_rate": 3.328238103371811e-08, "logits/chosen": -0.25667843222618103, "logits/rejected": -0.20781448483467102, "logps/chosen": -1.2320544719696045, "logps/rejected": -1.5089162588119507, "loss": 1.573, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2320544719696045, "rewards/margins": 0.2768619656562805, "rewards/rejected": -1.5089162588119507, "step": 5015 }, { "epoch": 2.6867369125271785, "grad_norm": 8.636269338621892, "learning_rate": 3.272593408728169e-08, "logits/chosen": -0.25457826256752014, "logits/rejected": -0.09930596500635147, "logps/chosen": -1.2071006298065186, "logps/rejected": -1.385921835899353, "loss": 1.5817, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2071006298065186, "rewards/margins": 0.1788211315870285, "rewards/rejected": -1.385921835899353, "step": 5020 }, { "epoch": 2.6894129453085798, "grad_norm": 7.812458444564327, "learning_rate": 3.217402063588204e-08, "logits/chosen": -0.25639787316322327, "logits/rejected": -0.14365923404693604, "logps/chosen": -1.263940691947937, "logps/rejected": -1.4570047855377197, "loss": 1.6064, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.263940691947937, "rewards/margins": 0.19306401908397675, "rewards/rejected": -1.4570047855377197, "step": 5025 }, { "epoch": 2.6920889780899815, "grad_norm": 8.15837912990784, "learning_rate": 3.162664603418608e-08, "logits/chosen": -0.18479934334754944, "logits/rejected": -0.13748101890087128, "logps/chosen": -1.2296315431594849, "logps/rejected": -1.449176549911499, "loss": 1.5842, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2296315431594849, "rewards/margins": 0.21954509615898132, "rewards/rejected": -1.449176549911499, "step": 5030 }, { "epoch": 2.694765010871383, "grad_norm": 14.596781404528606, "learning_rate": 3.1083815592824416e-08, "logits/chosen": -0.23647980391979218, "logits/rejected": -0.12656892836093903, "logps/chosen": -1.2724589109420776, "logps/rejected": -1.484289526939392, "loss": 1.6234, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2724589109420776, "rewards/margins": 0.2118304967880249, "rewards/rejected": -1.484289526939392, "step": 5035 }, { "epoch": 2.697441043652785, "grad_norm": 11.529255673618911, "learning_rate": 3.054553457834053e-08, "logits/chosen": -0.01275326032191515, "logits/rejected": -0.06487753987312317, "logps/chosen": -1.1942696571350098, "logps/rejected": -1.5052051544189453, "loss": 1.5246, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1942696571350098, "rewards/margins": 0.31093549728393555, "rewards/rejected": -1.5052051544189453, "step": 5040 }, { "epoch": 2.700117076434186, "grad_norm": 8.754850568076293, "learning_rate": 3.0011808213139036e-08, "logits/chosen": -0.1731742024421692, "logits/rejected": -0.14932702481746674, "logps/chosen": -1.1981568336486816, "logps/rejected": -1.4055202007293701, "loss": 1.5489, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1981568336486816, "rewards/margins": 0.20736345648765564, "rewards/rejected": -1.4055202007293701, "step": 5045 }, { "epoch": 2.702793109215588, "grad_norm": 8.859370600902391, "learning_rate": 2.948264167543568e-08, "logits/chosen": -0.20417948067188263, "logits/rejected": -0.16572143137454987, "logps/chosen": -1.1094834804534912, "logps/rejected": -1.4431778192520142, "loss": 1.4449, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -1.1094834804534912, "rewards/margins": 0.33369430899620056, "rewards/rejected": -1.4431778192520142, "step": 5050 }, { "epoch": 2.7054691419969896, "grad_norm": 8.30519925014008, "learning_rate": 2.8958040099206216e-08, "logits/chosen": -0.30399882793426514, "logits/rejected": -0.23754310607910156, "logps/chosen": -1.1653364896774292, "logps/rejected": -1.4400817155838013, "loss": 1.5383, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.1653364896774292, "rewards/margins": 0.2747451663017273, "rewards/rejected": -1.4400817155838013, "step": 5055 }, { "epoch": 2.708145174778391, "grad_norm": 11.704168087582307, "learning_rate": 2.843800857413775e-08, "logits/chosen": -0.18940813839435577, "logits/rejected": -0.14188504219055176, "logps/chosen": -1.2219481468200684, "logps/rejected": -1.4544917345046997, "loss": 1.5719, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2219481468200684, "rewards/margins": 0.23254361748695374, "rewards/rejected": -1.4544917345046997, "step": 5060 }, { "epoch": 2.7108212075597926, "grad_norm": 9.32374694815667, "learning_rate": 2.7922552145578203e-08, "logits/chosen": -0.2245454341173172, "logits/rejected": -0.03208258002996445, "logps/chosen": -1.2869073152542114, "logps/rejected": -1.5468248128890991, "loss": 1.6126, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2869073152542114, "rewards/margins": 0.2599175274372101, "rewards/rejected": -1.5468248128890991, "step": 5065 }, { "epoch": 2.7134972403411943, "grad_norm": 7.559388414982109, "learning_rate": 2.7411675814488277e-08, "logits/chosen": -0.10575256496667862, "logits/rejected": 0.030589479953050613, "logps/chosen": -1.1588886976242065, "logps/rejected": -1.376731514930725, "loss": 1.5377, "rewards/accuracies": 0.625, "rewards/chosen": -1.1588886976242065, "rewards/margins": 0.21784265339374542, "rewards/rejected": -1.376731514930725, "step": 5070 }, { "epoch": 2.7161732731225956, "grad_norm": 10.956395790032781, "learning_rate": 2.690538453739216e-08, "logits/chosen": -0.18339474499225616, "logits/rejected": -0.12223565578460693, "logps/chosen": -1.2097294330596924, "logps/rejected": -1.341022253036499, "loss": 1.6031, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2097294330596924, "rewards/margins": 0.13129296898841858, "rewards/rejected": -1.341022253036499, "step": 5075 }, { "epoch": 2.7188493059039973, "grad_norm": 7.363109836236882, "learning_rate": 2.6403683226330298e-08, "logits/chosen": -0.2401486337184906, "logits/rejected": -0.13127775490283966, "logps/chosen": -1.2772992849349976, "logps/rejected": -1.4677261114120483, "loss": 1.6174, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2772992849349976, "rewards/margins": 0.1904267817735672, "rewards/rejected": -1.4677261114120483, "step": 5080 }, { "epoch": 2.721525338685399, "grad_norm": 10.174103894659988, "learning_rate": 2.5906576748810804e-08, "logits/chosen": -0.3042605221271515, "logits/rejected": -0.20611527562141418, "logps/chosen": -1.1317962408065796, "logps/rejected": -1.5567015409469604, "loss": 1.4574, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.1317962408065796, "rewards/margins": 0.42490530014038086, "rewards/rejected": -1.5567015409469604, "step": 5085 }, { "epoch": 2.7242013714668003, "grad_norm": 11.486051703605545, "learning_rate": 2.5414069927763016e-08, "logits/chosen": -0.28841036558151245, "logits/rejected": -0.16781942546367645, "logps/chosen": -1.2877984046936035, "logps/rejected": -1.5316404104232788, "loss": 1.6117, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2877984046936035, "rewards/margins": 0.24384205043315887, "rewards/rejected": -1.5316404104232788, "step": 5090 }, { "epoch": 2.726877404248202, "grad_norm": 6.742605643992969, "learning_rate": 2.4926167541490185e-08, "logits/chosen": -0.31649142503738403, "logits/rejected": -0.1508931666612625, "logps/chosen": -1.2179322242736816, "logps/rejected": -1.5233149528503418, "loss": 1.5515, "rewards/accuracies": 0.625, "rewards/chosen": -1.2179322242736816, "rewards/margins": 0.30538275837898254, "rewards/rejected": -1.5233149528503418, "step": 5095 }, { "epoch": 2.7295534370296037, "grad_norm": 9.89213418989225, "learning_rate": 2.4442874323623574e-08, "logits/chosen": -0.11926839500665665, "logits/rejected": -0.006692640483379364, "logps/chosen": -1.1790974140167236, "logps/rejected": -1.5305302143096924, "loss": 1.5145, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.1790974140167236, "rewards/margins": 0.35143283009529114, "rewards/rejected": -1.5305302143096924, "step": 5100 }, { "epoch": 2.7322294698110055, "grad_norm": 10.34193039811828, "learning_rate": 2.396419496307589e-08, "logits/chosen": -0.18230049312114716, "logits/rejected": -0.044870562851428986, "logps/chosen": -1.2445478439331055, "logps/rejected": -1.4135087728500366, "loss": 1.5967, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2445478439331055, "rewards/margins": 0.16896098852157593, "rewards/rejected": -1.4135087728500366, "step": 5105 }, { "epoch": 2.7349055025924067, "grad_norm": 8.068889265098178, "learning_rate": 2.349013410399653e-08, "logits/chosen": -0.2620542645454407, "logits/rejected": -0.1357966959476471, "logps/chosen": -1.204101324081421, "logps/rejected": -1.4400665760040283, "loss": 1.5632, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.204101324081421, "rewards/margins": 0.23596513271331787, "rewards/rejected": -1.4400665760040283, "step": 5110 }, { "epoch": 2.7375815353738084, "grad_norm": 8.335383192965962, "learning_rate": 2.3020696345725954e-08, "logits/chosen": -0.2606256902217865, "logits/rejected": -0.08713199198246002, "logps/chosen": -1.2145345211029053, "logps/rejected": -1.5258738994598389, "loss": 1.539, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.2145345211029053, "rewards/margins": 0.31133946776390076, "rewards/rejected": -1.5258738994598389, "step": 5115 }, { "epoch": 2.7402575681552097, "grad_norm": 10.889216618276286, "learning_rate": 2.2555886242751398e-08, "logits/chosen": -0.20491114258766174, "logits/rejected": -0.16528639197349548, "logps/chosen": -1.3192756175994873, "logps/rejected": -1.5630297660827637, "loss": 1.6264, "rewards/accuracies": 0.59375, "rewards/chosen": -1.3192756175994873, "rewards/margins": 0.24375399947166443, "rewards/rejected": -1.5630297660827637, "step": 5120 }, { "epoch": 2.7429336009366114, "grad_norm": 15.88467586253665, "learning_rate": 2.2095708304662453e-08, "logits/chosen": -0.29381468892097473, "logits/rejected": -0.09942570328712463, "logps/chosen": -1.1907604932785034, "logps/rejected": -1.4625751972198486, "loss": 1.5411, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.1907604932785034, "rewards/margins": 0.2718146741390228, "rewards/rejected": -1.4625751972198486, "step": 5125 }, { "epoch": 2.745609633718013, "grad_norm": 6.842095804156989, "learning_rate": 2.16401669961076e-08, "logits/chosen": -0.3690885305404663, "logits/rejected": -0.2036101520061493, "logps/chosen": -1.2096027135849, "logps/rejected": -1.5022227764129639, "loss": 1.5295, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.2096027135849, "rewards/margins": 0.2926200032234192, "rewards/rejected": -1.5022227764129639, "step": 5130 }, { "epoch": 2.748285666499415, "grad_norm": 8.580770840370676, "learning_rate": 2.1189266736750532e-08, "logits/chosen": -0.130879208445549, "logits/rejected": -0.0750630795955658, "logps/chosen": -1.209316372871399, "logps/rejected": -1.454590916633606, "loss": 1.5721, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.209316372871399, "rewards/margins": 0.24527454376220703, "rewards/rejected": -1.454590916633606, "step": 5135 }, { "epoch": 2.750961699280816, "grad_norm": 6.706691055358692, "learning_rate": 2.0743011901227623e-08, "logits/chosen": -0.1535293161869049, "logits/rejected": -0.04885602742433548, "logps/chosen": -1.2943406105041504, "logps/rejected": -1.476296305656433, "loss": 1.6434, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -1.2943406105041504, "rewards/margins": 0.1819557249546051, "rewards/rejected": -1.476296305656433, "step": 5140 }, { "epoch": 2.753637732062218, "grad_norm": 11.940224560512375, "learning_rate": 2.030140681910508e-08, "logits/chosen": -0.22967776656150818, "logits/rejected": -0.07449609041213989, "logps/chosen": -1.2044155597686768, "logps/rejected": -1.4070556163787842, "loss": 1.5614, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -1.2044155597686768, "rewards/margins": 0.20264002680778503, "rewards/rejected": -1.4070556163787842, "step": 5145 }, { "epoch": 2.756313764843619, "grad_norm": 5.597851916698418, "learning_rate": 1.986445577483753e-08, "logits/chosen": -0.2449430227279663, "logits/rejected": -0.14069703221321106, "logps/chosen": -1.1979897022247314, "logps/rejected": -1.4494991302490234, "loss": 1.552, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.1979897022247314, "rewards/margins": 0.25150924921035767, "rewards/rejected": -1.4494991302490234, "step": 5150 }, { "epoch": 2.758989797625021, "grad_norm": 7.59879161071765, "learning_rate": 1.9432163007725765e-08, "logits/chosen": -0.28897973895072937, "logits/rejected": -0.20519177615642548, "logps/chosen": -1.268553376197815, "logps/rejected": -1.505704402923584, "loss": 1.5986, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.268553376197815, "rewards/margins": 0.23715105652809143, "rewards/rejected": -1.505704402923584, "step": 5155 }, { "epoch": 2.7616658304064226, "grad_norm": 9.142398559940261, "learning_rate": 1.9004532711876297e-08, "logits/chosen": -0.2665271461009979, "logits/rejected": -0.2148573398590088, "logps/chosen": -1.186708927154541, "logps/rejected": -1.5509085655212402, "loss": 1.5012, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.186708927154541, "rewards/margins": 0.364199697971344, "rewards/rejected": -1.5509085655212402, "step": 5160 }, { "epoch": 2.7643418631878243, "grad_norm": 8.065384544004745, "learning_rate": 1.8581569036159928e-08, "logits/chosen": -0.22522492706775665, "logits/rejected": -0.06659802049398422, "logps/chosen": -1.1889328956604004, "logps/rejected": -1.4411547183990479, "loss": 1.531, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1889328956604004, "rewards/margins": 0.2522217631340027, "rewards/rejected": -1.4411547183990479, "step": 5165 }, { "epoch": 2.7670178959692255, "grad_norm": 8.878700812965635, "learning_rate": 1.8163276084172285e-08, "logits/chosen": -0.1980402022600174, "logits/rejected": -0.09475599229335785, "logps/chosen": -1.2845584154129028, "logps/rejected": -1.4850971698760986, "loss": 1.6296, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2845584154129028, "rewards/margins": 0.2005387544631958, "rewards/rejected": -1.4850971698760986, "step": 5170 }, { "epoch": 2.7696939287506273, "grad_norm": 8.518784025504653, "learning_rate": 1.7749657914193194e-08, "logits/chosen": -0.254823237657547, "logits/rejected": -0.16935142874717712, "logps/chosen": -1.2913539409637451, "logps/rejected": -1.4431110620498657, "loss": 1.6446, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2913539409637451, "rewards/margins": 0.15175721049308777, "rewards/rejected": -1.4431110620498657, "step": 5175 }, { "epoch": 2.7723699615320285, "grad_norm": 6.991399757164978, "learning_rate": 1.7340718539148203e-08, "logits/chosen": -0.15637443959712982, "logits/rejected": -0.12771818041801453, "logps/chosen": -1.2784687280654907, "logps/rejected": -1.5572850704193115, "loss": 1.5896, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2784687280654907, "rewards/margins": 0.27881646156311035, "rewards/rejected": -1.5572850704193115, "step": 5180 }, { "epoch": 2.7750459943134302, "grad_norm": 12.336643207237364, "learning_rate": 1.6936461926568724e-08, "logits/chosen": -0.1913735717535019, "logits/rejected": -0.09848286211490631, "logps/chosen": -1.137644648551941, "logps/rejected": -1.4215614795684814, "loss": 1.5127, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.137644648551941, "rewards/margins": 0.2839166224002838, "rewards/rejected": -1.4215614795684814, "step": 5185 }, { "epoch": 2.777722027094832, "grad_norm": 8.694579253035036, "learning_rate": 1.6536891998554346e-08, "logits/chosen": -0.29581698775291443, "logits/rejected": -0.16581565141677856, "logps/chosen": -1.2244768142700195, "logps/rejected": -1.4132734537124634, "loss": 1.566, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.2244768142700195, "rewards/margins": 0.18879663944244385, "rewards/rejected": -1.4132734537124634, "step": 5190 }, { "epoch": 2.7803980598762337, "grad_norm": 12.053115358743863, "learning_rate": 1.6142012631734093e-08, "logits/chosen": -0.19457034766674042, "logits/rejected": -0.0800018310546875, "logps/chosen": -1.228502631187439, "logps/rejected": -1.4483073949813843, "loss": 1.5805, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -1.228502631187439, "rewards/margins": 0.21980471909046173, "rewards/rejected": -1.4483073949813843, "step": 5195 }, { "epoch": 2.783074092657635, "grad_norm": 9.889001908094752, "learning_rate": 1.575182765722949e-08, "logits/chosen": -0.27219057083129883, "logits/rejected": -0.15498098731040955, "logps/chosen": -1.175596833229065, "logps/rejected": -1.422145128250122, "loss": 1.5671, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.175596833229065, "rewards/margins": 0.24654832482337952, "rewards/rejected": -1.422145128250122, "step": 5200 }, { "epoch": 2.783074092657635, "eval_logits/chosen": 0.12440230697393417, "eval_logits/rejected": 0.20024247467517853, "eval_logps/chosen": -1.3029695749282837, "eval_logps/rejected": -1.499582052230835, "eval_loss": 1.6474473476409912, "eval_rewards/accuracies": 0.5704748034477234, "eval_rewards/chosen": -1.3029695749282837, "eval_rewards/margins": 0.19661255180835724, "eval_rewards/rejected": -1.499582052230835, "eval_runtime": 40.378, "eval_samples_per_second": 33.31, "eval_steps_per_second": 8.346, "step": 5200 }, { "epoch": 2.7857501254390367, "grad_norm": 7.894284148032132, "learning_rate": 1.536634086061672e-08, "logits/chosen": -0.11715634167194366, "logits/rejected": -0.09793325513601303, "logps/chosen": -1.2184431552886963, "logps/rejected": -1.4105503559112549, "loss": 1.5915, "rewards/accuracies": 0.46875, "rewards/chosen": -1.2184431552886963, "rewards/margins": 0.19210708141326904, "rewards/rejected": -1.4105503559112549, "step": 5205 }, { "epoch": 2.788426158220438, "grad_norm": 8.626425862347766, "learning_rate": 1.4985555981890495e-08, "logits/chosen": -0.1889527142047882, "logits/rejected": -0.12425367534160614, "logps/chosen": -1.2231550216674805, "logps/rejected": -1.4383323192596436, "loss": 1.5682, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.2231550216674805, "rewards/margins": 0.21517720818519592, "rewards/rejected": -1.4383323192596436, "step": 5210 }, { "epoch": 2.7911021910018396, "grad_norm": 7.747896620645018, "learning_rate": 1.4609476715427226e-08, "logits/chosen": -0.2118125855922699, "logits/rejected": -0.13611485064029694, "logps/chosen": -1.2079906463623047, "logps/rejected": -1.5039584636688232, "loss": 1.5452, "rewards/accuracies": 0.5625, "rewards/chosen": -1.2079906463623047, "rewards/margins": 0.2959679663181305, "rewards/rejected": -1.5039584636688232, "step": 5215 }, { "epoch": 2.7937782237832414, "grad_norm": 10.303972437834066, "learning_rate": 1.4238106709949792e-08, "logits/chosen": -0.22145023941993713, "logits/rejected": -0.1696382611989975, "logps/chosen": -1.198301911354065, "logps/rejected": -1.5057936906814575, "loss": 1.5284, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.198301911354065, "rewards/margins": 0.3074919283390045, "rewards/rejected": -1.5057936906814575, "step": 5220 }, { "epoch": 2.796454256564643, "grad_norm": 9.121670917332018, "learning_rate": 1.3871449568491511e-08, "logits/chosen": -0.16502538323402405, "logits/rejected": -0.062241099774837494, "logps/chosen": -1.2805616855621338, "logps/rejected": -1.5200989246368408, "loss": 1.5943, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2805616855621338, "rewards/margins": 0.2395373284816742, "rewards/rejected": -1.5200989246368408, "step": 5225 }, { "epoch": 2.7991302893460444, "grad_norm": 9.755347750909337, "learning_rate": 1.3509508848361606e-08, "logits/chosen": -0.31038230657577515, "logits/rejected": -0.20442113280296326, "logps/chosen": -1.2425405979156494, "logps/rejected": -1.4383920431137085, "loss": 1.59, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2425405979156494, "rewards/margins": 0.19585171341896057, "rewards/rejected": -1.4383920431137085, "step": 5230 }, { "epoch": 2.801806322127446, "grad_norm": 6.164015248610251, "learning_rate": 1.3152288061110517e-08, "logits/chosen": -0.26571184396743774, "logits/rejected": -0.19366225600242615, "logps/chosen": -1.251062035560608, "logps/rejected": -1.454511284828186, "loss": 1.5877, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.251062035560608, "rewards/margins": 0.20344917476177216, "rewards/rejected": -1.454511284828186, "step": 5235 }, { "epoch": 2.804482354908848, "grad_norm": 9.239101149943929, "learning_rate": 1.2799790672495814e-08, "logits/chosen": -0.23363038897514343, "logits/rejected": -0.05663818120956421, "logps/chosen": -1.243467092514038, "logps/rejected": -1.462083101272583, "loss": 1.6168, "rewards/accuracies": 0.59375, "rewards/chosen": -1.243467092514038, "rewards/margins": 0.2186162769794464, "rewards/rejected": -1.462083101272583, "step": 5240 }, { "epoch": 2.807158387690249, "grad_norm": 6.829649870494868, "learning_rate": 1.2452020102448835e-08, "logits/chosen": -0.13945211470127106, "logits/rejected": -0.10049142688512802, "logps/chosen": -1.194481372833252, "logps/rejected": -1.3906782865524292, "loss": 1.5771, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.194481372833252, "rewards/margins": 0.1961970031261444, "rewards/rejected": -1.3906782865524292, "step": 5245 }, { "epoch": 2.8098344204716508, "grad_norm": 7.432169595362053, "learning_rate": 1.2108979725041103e-08, "logits/chosen": -0.23462161421775818, "logits/rejected": -0.1215369701385498, "logps/chosen": -1.270316481590271, "logps/rejected": -1.473505973815918, "loss": 1.6098, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.270316481590271, "rewards/margins": 0.20318952202796936, "rewards/rejected": -1.473505973815918, "step": 5250 }, { "epoch": 2.8125104532530525, "grad_norm": 8.257547457316058, "learning_rate": 1.1770672868451958e-08, "logits/chosen": -0.1968560516834259, "logits/rejected": -0.013767095282673836, "logps/chosen": -1.2501423358917236, "logps/rejected": -1.419557809829712, "loss": 1.5999, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2501423358917236, "rewards/margins": 0.1694156527519226, "rewards/rejected": -1.419557809829712, "step": 5255 }, { "epoch": 2.8151864860344538, "grad_norm": 10.65316125130127, "learning_rate": 1.1437102814935872e-08, "logits/chosen": -0.2315382957458496, "logits/rejected": -0.17422154545783997, "logps/chosen": -1.2226743698120117, "logps/rejected": -1.5226083993911743, "loss": 1.58, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2226743698120117, "rewards/margins": 0.29993388056755066, "rewards/rejected": -1.5226083993911743, "step": 5260 }, { "epoch": 2.8178625188158555, "grad_norm": 8.085369009106172, "learning_rate": 1.1108272800791018e-08, "logits/chosen": -0.312086284160614, "logits/rejected": -0.1401098519563675, "logps/chosen": -1.405895709991455, "logps/rejected": -1.5206575393676758, "loss": 1.7369, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.405895709991455, "rewards/margins": 0.11476186662912369, "rewards/rejected": -1.5206575393676758, "step": 5265 }, { "epoch": 2.820538551597257, "grad_norm": 6.393843195919607, "learning_rate": 1.078418601632769e-08, "logits/chosen": -0.19875682890415192, "logits/rejected": -0.0823943018913269, "logps/chosen": -1.1760179996490479, "logps/rejected": -1.4837948083877563, "loss": 1.5043, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1760179996490479, "rewards/margins": 0.3077767491340637, "rewards/rejected": -1.4837948083877563, "step": 5270 }, { "epoch": 2.8232145843786585, "grad_norm": 8.06384468638755, "learning_rate": 1.0464845605837159e-08, "logits/chosen": -0.20249497890472412, "logits/rejected": -0.07072342932224274, "logps/chosen": -1.254334807395935, "logps/rejected": -1.4947919845581055, "loss": 1.5855, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.254334807395935, "rewards/margins": 0.24045705795288086, "rewards/rejected": -1.4947919845581055, "step": 5275 }, { "epoch": 2.82589061716006, "grad_norm": 9.025134402265097, "learning_rate": 1.0150254667561642e-08, "logits/chosen": -0.19225776195526123, "logits/rejected": -0.060486339032649994, "logps/chosen": -1.291597604751587, "logps/rejected": -1.489046573638916, "loss": 1.6364, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.291597604751587, "rewards/margins": 0.19744889438152313, "rewards/rejected": -1.489046573638916, "step": 5280 }, { "epoch": 2.828566649941462, "grad_norm": 7.938475769626625, "learning_rate": 9.840416253663719e-09, "logits/chosen": -0.2338906079530716, "logits/rejected": -0.1550368070602417, "logps/chosen": -1.194307565689087, "logps/rejected": -1.4473832845687866, "loss": 1.5363, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.194307565689087, "rewards/margins": 0.253075510263443, "rewards/rejected": -1.4473832845687866, "step": 5285 }, { "epoch": 2.8312426827228636, "grad_norm": 7.118025400739444, "learning_rate": 9.535333370197074e-09, "logits/chosen": -0.21142733097076416, "logits/rejected": -0.0969664454460144, "logps/chosen": -1.2613712549209595, "logps/rejected": -1.4549555778503418, "loss": 1.6146, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2613712549209595, "rewards/margins": 0.19358427822589874, "rewards/rejected": -1.4549555778503418, "step": 5290 }, { "epoch": 2.833918715504265, "grad_norm": 6.376414171175719, "learning_rate": 9.23500897707713e-09, "logits/chosen": -0.2588057518005371, "logits/rejected": -0.10172237455844879, "logps/chosen": -1.32537043094635, "logps/rejected": -1.5747493505477905, "loss": 1.6308, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.32537043094635, "rewards/margins": 0.24937906861305237, "rewards/rejected": -1.5747493505477905, "step": 5295 }, { "epoch": 2.8365947482856666, "grad_norm": 7.875069396310308, "learning_rate": 8.939445988052574e-09, "logits/chosen": -0.20552654564380646, "logits/rejected": -0.17443294823169708, "logps/chosen": -1.2336318492889404, "logps/rejected": -1.5586997270584106, "loss": 1.5621, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2336318492889404, "rewards/margins": 0.3250678479671478, "rewards/rejected": -1.5586997270584106, "step": 5300 }, { "epoch": 2.839270781067068, "grad_norm": 8.252376839450696, "learning_rate": 8.648647270676656e-09, "logits/chosen": -0.18448421359062195, "logits/rejected": -0.06468196958303452, "logps/chosen": -1.285400152206421, "logps/rejected": -1.5381015539169312, "loss": 1.6323, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.285400152206421, "rewards/margins": 0.25270140171051025, "rewards/rejected": -1.5381015539169312, "step": 5305 }, { "epoch": 2.8419468138484696, "grad_norm": 5.856217296203612, "learning_rate": 8.362615646279991e-09, "logits/chosen": -0.3467407822608948, "logits/rejected": -0.13658985495567322, "logps/chosen": -1.1790553331375122, "logps/rejected": -1.4979755878448486, "loss": 1.5298, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.1790553331375122, "rewards/margins": 0.3189201354980469, "rewards/rejected": -1.4979755878448486, "step": 5310 }, { "epoch": 2.8446228466298713, "grad_norm": 9.410102772135225, "learning_rate": 8.081353889942466e-09, "logits/chosen": -0.13126492500305176, "logits/rejected": -0.05035420134663582, "logps/chosen": -1.2153863906860352, "logps/rejected": -1.3907142877578735, "loss": 1.5759, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2153863906860352, "rewards/margins": 0.175327867269516, "rewards/rejected": -1.3907142877578735, "step": 5315 }, { "epoch": 2.847298879411273, "grad_norm": 8.792712405679417, "learning_rate": 7.804864730467042e-09, "logits/chosen": -0.11626704037189484, "logits/rejected": -0.05582579970359802, "logps/chosen": -1.2154616117477417, "logps/rejected": -1.3626997470855713, "loss": 1.5908, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2154616117477417, "rewards/margins": 0.14723792672157288, "rewards/rejected": -1.3626997470855713, "step": 5320 }, { "epoch": 2.8499749121926743, "grad_norm": 7.961338709116618, "learning_rate": 7.533150850352665e-09, "logits/chosen": -0.1864510327577591, "logits/rejected": -0.0834638699889183, "logps/chosen": -1.262117862701416, "logps/rejected": -1.5629605054855347, "loss": 1.5815, "rewards/accuracies": 0.59375, "rewards/chosen": -1.262117862701416, "rewards/margins": 0.30084237456321716, "rewards/rejected": -1.5629605054855347, "step": 5325 }, { "epoch": 2.852650944974076, "grad_norm": 9.561337144875273, "learning_rate": 7.2662148857686175e-09, "logits/chosen": -0.13462045788764954, "logits/rejected": -0.08371740579605103, "logps/chosen": -1.1950032711029053, "logps/rejected": -1.4544382095336914, "loss": 1.5301, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1950032711029053, "rewards/margins": 0.2594349682331085, "rewards/rejected": -1.4544382095336914, "step": 5330 }, { "epoch": 2.8553269777554773, "grad_norm": 13.869066735304004, "learning_rate": 7.0040594265287635e-09, "logits/chosen": -0.10755538940429688, "logits/rejected": -0.13098764419555664, "logps/chosen": -1.2394194602966309, "logps/rejected": -1.4161888360977173, "loss": 1.597, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.2394194602966309, "rewards/margins": 0.17676934599876404, "rewards/rejected": -1.4161888360977173, "step": 5335 }, { "epoch": 2.858003010536879, "grad_norm": 8.208179342259479, "learning_rate": 6.746687016066566e-09, "logits/chosen": -0.14634302258491516, "logits/rejected": -0.13666020333766937, "logps/chosen": -1.258013129234314, "logps/rejected": -1.4033854007720947, "loss": 1.6307, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.258013129234314, "rewards/margins": 0.1453721970319748, "rewards/rejected": -1.4033854007720947, "step": 5340 }, { "epoch": 2.8606790433182807, "grad_norm": 6.851778857771071, "learning_rate": 6.494100151410276e-09, "logits/chosen": -0.2954311966896057, "logits/rejected": -0.1517902910709381, "logps/chosen": -1.1930725574493408, "logps/rejected": -1.4258384704589844, "loss": 1.5191, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1930725574493408, "rewards/margins": 0.2327660322189331, "rewards/rejected": -1.4258384704589844, "step": 5345 }, { "epoch": 2.8633550760996824, "grad_norm": 10.521025339927364, "learning_rate": 6.246301283158728e-09, "logits/chosen": -0.1592158079147339, "logits/rejected": -0.19058796763420105, "logps/chosen": -1.3086292743682861, "logps/rejected": -1.5055660009384155, "loss": 1.6591, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.3086292743682861, "rewards/margins": 0.1969369798898697, "rewards/rejected": -1.5055660009384155, "step": 5350 }, { "epoch": 2.8660311088810837, "grad_norm": 9.041618099717345, "learning_rate": 6.0032928154576944e-09, "logits/chosen": -0.22521352767944336, "logits/rejected": -0.15095236897468567, "logps/chosen": -1.2401598691940308, "logps/rejected": -1.4540021419525146, "loss": 1.6007, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.2401598691940308, "rewards/margins": 0.21384210884571075, "rewards/rejected": -1.4540021419525146, "step": 5355 }, { "epoch": 2.8687071416624854, "grad_norm": 9.742275839581838, "learning_rate": 5.76507710597629e-09, "logits/chosen": -0.20284557342529297, "logits/rejected": -0.045843563973903656, "logps/chosen": -1.2402071952819824, "logps/rejected": -1.4867188930511475, "loss": 1.6002, "rewards/accuracies": 0.59375, "rewards/chosen": -1.2402071952819824, "rewards/margins": 0.24651166796684265, "rewards/rejected": -1.4867188930511475, "step": 5360 }, { "epoch": 2.8713831744438867, "grad_norm": 7.235815472685509, "learning_rate": 5.531656465884438e-09, "logits/chosen": -0.2580950856208801, "logits/rejected": -0.12209419161081314, "logps/chosen": -1.2364237308502197, "logps/rejected": -1.5134259462356567, "loss": 1.5656, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2364237308502197, "rewards/margins": 0.2770020663738251, "rewards/rejected": -1.5134259462356567, "step": 5365 }, { "epoch": 2.8740592072252884, "grad_norm": 9.834877063853899, "learning_rate": 5.303033159830217e-09, "logits/chosen": -0.14675696194171906, "logits/rejected": -0.1237391009926796, "logps/chosen": -1.2326931953430176, "logps/rejected": -1.3666555881500244, "loss": 1.6062, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -1.2326931953430176, "rewards/margins": 0.1339622437953949, "rewards/rejected": -1.3666555881500244, "step": 5370 }, { "epoch": 2.87673524000669, "grad_norm": 8.08309568080615, "learning_rate": 5.079209405917939e-09, "logits/chosen": -0.20791010558605194, "logits/rejected": -0.12605224549770355, "logps/chosen": -1.1827991008758545, "logps/rejected": -1.5293145179748535, "loss": 1.5095, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.1827991008758545, "rewards/margins": 0.34651538729667664, "rewards/rejected": -1.5293145179748535, "step": 5375 }, { "epoch": 2.879411272788092, "grad_norm": 6.139482732449664, "learning_rate": 4.860187375686664e-09, "logits/chosen": -0.2154158651828766, "logits/rejected": -0.040366265922784805, "logps/chosen": -1.3586286306381226, "logps/rejected": -1.548743724822998, "loss": 1.6622, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.3586286306381226, "rewards/margins": 0.19011513888835907, "rewards/rejected": -1.548743724822998, "step": 5380 }, { "epoch": 2.882087305569493, "grad_norm": 7.257550059938905, "learning_rate": 4.64596919408905e-09, "logits/chosen": -0.1530151516199112, "logits/rejected": -0.09397026151418686, "logps/chosen": -1.217520833015442, "logps/rejected": -1.400549292564392, "loss": 1.5824, "rewards/accuracies": 0.59375, "rewards/chosen": -1.217520833015442, "rewards/margins": 0.18302853405475616, "rewards/rejected": -1.400549292564392, "step": 5385 }, { "epoch": 2.884763338350895, "grad_norm": 6.777289153854856, "learning_rate": 4.436556939470814e-09, "logits/chosen": -0.16746576130390167, "logits/rejected": -0.07902796566486359, "logps/chosen": -1.2627239227294922, "logps/rejected": -1.4349424839019775, "loss": 1.6059, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2627239227294922, "rewards/margins": 0.17221860587596893, "rewards/rejected": -1.4349424839019775, "step": 5390 }, { "epoch": 2.887439371132296, "grad_norm": 6.679859669428982, "learning_rate": 4.23195264355064e-09, "logits/chosen": -0.327754408121109, "logits/rejected": -0.16632018983364105, "logps/chosen": -1.1831705570220947, "logps/rejected": -1.4538441896438599, "loss": 1.5111, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.1831705570220947, "rewards/margins": 0.2706735134124756, "rewards/rejected": -1.4538441896438599, "step": 5395 }, { "epoch": 2.890115403913698, "grad_norm": 7.606677602141569, "learning_rate": 4.032158291400245e-09, "logits/chosen": -0.2535339891910553, "logits/rejected": -0.049510832875967026, "logps/chosen": -1.2447563409805298, "logps/rejected": -1.6381620168685913, "loss": 1.5732, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2447563409805298, "rewards/margins": 0.3934057354927063, "rewards/rejected": -1.6381620168685913, "step": 5400 }, { "epoch": 2.8927914366950995, "grad_norm": 9.116747966192884, "learning_rate": 3.837175821425398e-09, "logits/chosen": -0.16068437695503235, "logits/rejected": -0.13778674602508545, "logps/chosen": -1.300722360610962, "logps/rejected": -1.568656325340271, "loss": 1.6196, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.300722360610962, "rewards/margins": 0.2679339051246643, "rewards/rejected": -1.568656325340271, "step": 5405 }, { "epoch": 2.8954674694765012, "grad_norm": 7.7744056104123445, "learning_rate": 3.6470071253467683e-09, "logits/chosen": -0.1354900598526001, "logits/rejected": -0.029869627207517624, "logps/chosen": -1.2225590944290161, "logps/rejected": -1.5437707901000977, "loss": 1.5594, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2225590944290161, "rewards/margins": 0.321211576461792, "rewards/rejected": -1.5437707901000977, "step": 5410 }, { "epoch": 2.8981435022579025, "grad_norm": 8.490174611217292, "learning_rate": 3.461654048181939e-09, "logits/chosen": -0.2366805523633957, "logits/rejected": -0.10289554297924042, "logps/chosen": -1.252111554145813, "logps/rejected": -1.4506523609161377, "loss": 1.5945, "rewards/accuracies": 0.625, "rewards/chosen": -1.252111554145813, "rewards/margins": 0.1985408365726471, "rewards/rejected": -1.4506523609161377, "step": 5415 }, { "epoch": 2.9008195350393042, "grad_norm": 9.58980198103724, "learning_rate": 3.281118388227255e-09, "logits/chosen": -0.17817652225494385, "logits/rejected": -0.1203083023428917, "logps/chosen": -1.1674425601959229, "logps/rejected": -1.3943111896514893, "loss": 1.541, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1674425601959229, "rewards/margins": 0.2268688678741455, "rewards/rejected": -1.3943111896514893, "step": 5420 }, { "epoch": 2.903495567820706, "grad_norm": 13.051780801828894, "learning_rate": 3.1054018970405048e-09, "logits/chosen": -0.18536776304244995, "logits/rejected": -0.08422654867172241, "logps/chosen": -1.272070288658142, "logps/rejected": -1.519740343093872, "loss": 1.6125, "rewards/accuracies": 0.5625, "rewards/chosen": -1.272070288658142, "rewards/margins": 0.24767009913921356, "rewards/rejected": -1.519740343093872, "step": 5425 }, { "epoch": 2.906171600602107, "grad_norm": 9.637390983771256, "learning_rate": 2.9345062794238207e-09, "logits/chosen": -0.19817817211151123, "logits/rejected": -0.05495418980717659, "logps/chosen": -1.2120387554168701, "logps/rejected": -1.5326441526412964, "loss": 1.5384, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2120387554168701, "rewards/margins": 0.3206053376197815, "rewards/rejected": -1.5326441526412964, "step": 5430 }, { "epoch": 2.908847633383509, "grad_norm": 12.442063962181845, "learning_rate": 2.7684331934072492e-09, "logits/chosen": -0.3037939667701721, "logits/rejected": -0.21871428191661835, "logps/chosen": -1.2116082906723022, "logps/rejected": -1.531022310256958, "loss": 1.5294, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2116082906723022, "rewards/margins": 0.31941401958465576, "rewards/rejected": -1.531022310256958, "step": 5435 }, { "epoch": 2.9115236661649107, "grad_norm": 8.255143974951732, "learning_rate": 2.6071842502326526e-09, "logits/chosen": -0.2509545385837555, "logits/rejected": -0.16142788529396057, "logps/chosen": -1.2278988361358643, "logps/rejected": -1.5453009605407715, "loss": 1.5674, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.2278988361358643, "rewards/margins": 0.3174021244049072, "rewards/rejected": -1.5453009605407715, "step": 5440 }, { "epoch": 2.9141996989463124, "grad_norm": 7.084364648224411, "learning_rate": 2.450761014337888e-09, "logits/chosen": -0.03262624889612198, "logits/rejected": -0.03374551981687546, "logps/chosen": -1.2018756866455078, "logps/rejected": -1.586195707321167, "loss": 1.5345, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2018756866455078, "rewards/margins": 0.3843199908733368, "rewards/rejected": -1.586195707321167, "step": 5445 }, { "epoch": 2.9168757317277136, "grad_norm": 9.392045491922225, "learning_rate": 2.299165003341985e-09, "logits/chosen": -0.13462188839912415, "logits/rejected": -0.04474419727921486, "logps/chosen": -1.2435557842254639, "logps/rejected": -1.5604596138000488, "loss": 1.5655, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.2435557842254639, "rewards/margins": 0.31690365076065063, "rewards/rejected": -1.5604596138000488, "step": 5450 }, { "epoch": 2.9195517645091154, "grad_norm": 7.59640684513388, "learning_rate": 2.1523976880299945e-09, "logits/chosen": -0.2498139888048172, "logits/rejected": -0.11051896959543228, "logps/chosen": -1.2221424579620361, "logps/rejected": -1.3954873085021973, "loss": 1.5822, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.2221424579620361, "rewards/margins": 0.17334489524364471, "rewards/rejected": -1.3954873085021973, "step": 5455 }, { "epoch": 2.9222277972905166, "grad_norm": 8.232145092021927, "learning_rate": 2.010460492339161e-09, "logits/chosen": -0.24411647021770477, "logits/rejected": -0.17508909106254578, "logps/chosen": -1.2442398071289062, "logps/rejected": -1.5127445459365845, "loss": 1.5682, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2442398071289062, "rewards/margins": 0.26850467920303345, "rewards/rejected": -1.5127445459365845, "step": 5460 }, { "epoch": 2.9249038300719183, "grad_norm": 8.596987983047574, "learning_rate": 1.8733547933446614e-09, "logits/chosen": -0.2940111756324768, "logits/rejected": -0.14695706963539124, "logps/chosen": -1.29569411277771, "logps/rejected": -1.4681915044784546, "loss": 1.643, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.29569411277771, "rewards/margins": 0.17249757051467896, "rewards/rejected": -1.4681915044784546, "step": 5465 }, { "epoch": 2.92757986285332, "grad_norm": 10.207032165158024, "learning_rate": 1.7410819212467231e-09, "logits/chosen": -0.19064298272132874, "logits/rejected": -0.12294765561819077, "logps/chosen": -1.1765105724334717, "logps/rejected": -1.425095796585083, "loss": 1.5242, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.1765105724334717, "rewards/margins": 0.2485850751399994, "rewards/rejected": -1.425095796585083, "step": 5470 }, { "epoch": 2.9302558956347218, "grad_norm": 8.050551279955041, "learning_rate": 1.613643159357192e-09, "logits/chosen": -0.14887556433677673, "logits/rejected": -0.1944231539964676, "logps/chosen": -1.1479804515838623, "logps/rejected": -1.3909088373184204, "loss": 1.5207, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.1479804515838623, "rewards/margins": 0.2429281771183014, "rewards/rejected": -1.3909088373184204, "step": 5475 }, { "epoch": 2.932931928416123, "grad_norm": 7.206610777472823, "learning_rate": 1.4910397440875967e-09, "logits/chosen": -0.1818329095840454, "logits/rejected": -0.0986550822854042, "logps/chosen": -1.241981863975525, "logps/rejected": -1.4175071716308594, "loss": 1.6009, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.241981863975525, "rewards/margins": 0.17552533745765686, "rewards/rejected": -1.4175071716308594, "step": 5480 }, { "epoch": 2.9356079611975248, "grad_norm": 8.150550298484777, "learning_rate": 1.3732728649368253e-09, "logits/chosen": -0.12400376796722412, "logits/rejected": 0.01435978151857853, "logps/chosen": -1.219214916229248, "logps/rejected": -1.4300227165222168, "loss": 1.5769, "rewards/accuracies": 0.625, "rewards/chosen": -1.219214916229248, "rewards/margins": 0.21080783009529114, "rewards/rejected": -1.4300227165222168, "step": 5485 }, { "epoch": 2.938283993978926, "grad_norm": 9.444641265654619, "learning_rate": 1.260343664479524e-09, "logits/chosen": -0.18277186155319214, "logits/rejected": -0.15881676971912384, "logps/chosen": -1.2164661884307861, "logps/rejected": -1.3838369846343994, "loss": 1.5771, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2164661884307861, "rewards/margins": 0.1673709601163864, "rewards/rejected": -1.3838369846343994, "step": 5490 }, { "epoch": 2.9409600267603278, "grad_norm": 7.254483347505405, "learning_rate": 1.1522532383554384e-09, "logits/chosen": -0.28208598494529724, "logits/rejected": -0.11577792465686798, "logps/chosen": -1.179441213607788, "logps/rejected": -1.4718883037567139, "loss": 1.5135, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.179441213607788, "rewards/margins": 0.2924470007419586, "rewards/rejected": -1.4718883037567139, "step": 5495 }, { "epoch": 2.9436360595417295, "grad_norm": 7.158672078935296, "learning_rate": 1.049002635258256e-09, "logits/chosen": -0.2083779126405716, "logits/rejected": -0.11806578934192657, "logps/chosen": -1.2659683227539062, "logps/rejected": -1.39592707157135, "loss": 1.6229, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.2659683227539062, "rewards/margins": 0.12995865941047668, "rewards/rejected": -1.39592707157135, "step": 5500 }, { "epoch": 2.946312092323131, "grad_norm": 6.341158898811177, "learning_rate": 9.505928569258358e-10, "logits/chosen": -0.12015247344970703, "logits/rejected": -0.12586413323879242, "logps/chosen": -1.2438194751739502, "logps/rejected": -1.4896745681762695, "loss": 1.5735, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2438194751739502, "rewards/margins": 0.2458551824092865, "rewards/rejected": -1.4896745681762695, "step": 5505 }, { "epoch": 2.9489881251045325, "grad_norm": 9.149303002417405, "learning_rate": 8.57024858130273e-10, "logits/chosen": -0.23201246559619904, "logits/rejected": -0.15525290369987488, "logps/chosen": -1.2421350479125977, "logps/rejected": -1.585129737854004, "loss": 1.5495, "rewards/accuracies": 0.65625, "rewards/chosen": -1.2421350479125977, "rewards/margins": 0.3429946005344391, "rewards/rejected": -1.585129737854004, "step": 5510 }, { "epoch": 2.951664157885934, "grad_norm": 9.70660458120073, "learning_rate": 7.682995466686826e-10, "logits/chosen": -0.28649234771728516, "logits/rejected": -0.1822827011346817, "logps/chosen": -1.207561731338501, "logps/rejected": -1.5450680255889893, "loss": 1.5349, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.207561731338501, "rewards/margins": 0.3375062346458435, "rewards/rejected": -1.5450680255889893, "step": 5515 }, { "epoch": 2.9543401906673354, "grad_norm": 9.893690419887266, "learning_rate": 6.844177833543741e-10, "logits/chosen": -0.1666533648967743, "logits/rejected": -0.12191154062747955, "logps/chosen": -1.1799771785736084, "logps/rejected": -1.418250322341919, "loss": 1.5562, "rewards/accuracies": 0.59375, "rewards/chosen": -1.1799771785736084, "rewards/margins": 0.23827309906482697, "rewards/rejected": -1.418250322341919, "step": 5520 }, { "epoch": 2.957016223448737, "grad_norm": 6.773672161944208, "learning_rate": 6.053803820087467e-10, "logits/chosen": -0.18774524331092834, "logits/rejected": -0.08795052021741867, "logps/chosen": -1.2257516384124756, "logps/rejected": -1.5606324672698975, "loss": 1.5534, "rewards/accuracies": 0.625, "rewards/chosen": -1.2257516384124756, "rewards/margins": 0.33488088846206665, "rewards/rejected": -1.5606324672698975, "step": 5525 }, { "epoch": 2.959692256230139, "grad_norm": 12.468963843370322, "learning_rate": 5.311881094528514e-10, "logits/chosen": -0.21507851779460907, "logits/rejected": -0.02370876632630825, "logps/chosen": -1.3210976123809814, "logps/rejected": -1.4538319110870361, "loss": 1.6688, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.3210976123809814, "rewards/margins": 0.13273414969444275, "rewards/rejected": -1.4538319110870361, "step": 5530 }, { "epoch": 2.9623682890115406, "grad_norm": 11.22212558598113, "learning_rate": 4.6184168550050806e-10, "logits/chosen": -0.2066337615251541, "logits/rejected": -0.17999473214149475, "logps/chosen": -1.2044970989227295, "logps/rejected": -1.3566124439239502, "loss": 1.5863, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2044970989227295, "rewards/margins": 0.1521153599023819, "rewards/rejected": -1.3566124439239502, "step": 5535 }, { "epoch": 2.965044321792942, "grad_norm": 9.236625587079201, "learning_rate": 3.973417829510328e-10, "logits/chosen": -0.3020160496234894, "logits/rejected": -0.16865511238574982, "logps/chosen": -1.278668999671936, "logps/rejected": -1.4230049848556519, "loss": 1.6192, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -1.278668999671936, "rewards/margins": 0.1443360149860382, "rewards/rejected": -1.4230049848556519, "step": 5540 }, { "epoch": 2.9677203545743436, "grad_norm": 7.604616929280045, "learning_rate": 3.3768902758274377e-10, "logits/chosen": -0.18155820667743683, "logits/rejected": -0.09487228095531464, "logps/chosen": -1.1723692417144775, "logps/rejected": -1.377142310142517, "loss": 1.5527, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.1723692417144775, "rewards/margins": 0.20477309823036194, "rewards/rejected": -1.377142310142517, "step": 5545 }, { "epoch": 2.970396387355745, "grad_norm": 7.8461369749242715, "learning_rate": 2.8288399814691e-10, "logits/chosen": -0.10973820835351944, "logits/rejected": -0.012431502342224121, "logps/chosen": -1.296041488647461, "logps/rejected": -1.5153841972351074, "loss": 1.6154, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.296041488647461, "rewards/margins": 0.21934261918067932, "rewards/rejected": -1.5153841972351074, "step": 5550 }, { "epoch": 2.9730724201371466, "grad_norm": 9.744055832864698, "learning_rate": 2.3292722636220066e-10, "logits/chosen": -0.21309657394886017, "logits/rejected": -0.040102262049913406, "logps/chosen": -1.3197863101959229, "logps/rejected": -1.5416650772094727, "loss": 1.6391, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.3197863101959229, "rewards/margins": 0.22187869250774384, "rewards/rejected": -1.5416650772094727, "step": 5555 }, { "epoch": 2.9757484529185483, "grad_norm": 7.960303028084376, "learning_rate": 1.8781919690946668e-10, "logits/chosen": -0.1571412831544876, "logits/rejected": -0.15997639298439026, "logps/chosen": -1.2541515827178955, "logps/rejected": -1.3549838066101074, "loss": 1.62, "rewards/accuracies": 0.53125, "rewards/chosen": -1.2541515827178955, "rewards/margins": 0.10083191096782684, "rewards/rejected": -1.3549838066101074, "step": 5560 }, { "epoch": 2.97842448569995, "grad_norm": 11.345542567123752, "learning_rate": 1.4756034742696711e-10, "logits/chosen": -0.249413400888443, "logits/rejected": -0.22129006683826447, "logps/chosen": -1.2185876369476318, "logps/rejected": -1.3922250270843506, "loss": 1.5827, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -1.2185876369476318, "rewards/margins": 0.17363746464252472, "rewards/rejected": -1.3922250270843506, "step": 5565 }, { "epoch": 2.9811005184813513, "grad_norm": 9.709369528026375, "learning_rate": 1.12151068506261e-10, "logits/chosen": -0.16392943263053894, "logits/rejected": -0.06352823972702026, "logps/chosen": -1.1880154609680176, "logps/rejected": -1.525194764137268, "loss": 1.5293, "rewards/accuracies": 0.65625, "rewards/chosen": -1.1880154609680176, "rewards/margins": 0.3371793329715729, "rewards/rejected": -1.525194764137268, "step": 5570 }, { "epoch": 2.983776551262753, "grad_norm": 9.717767351843722, "learning_rate": 8.159170368826629e-11, "logits/chosen": -0.2174181491136551, "logits/rejected": -0.09884576499462128, "logps/chosen": -1.2130773067474365, "logps/rejected": -1.4968039989471436, "loss": 1.5547, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.2130773067474365, "rewards/margins": 0.2837267816066742, "rewards/rejected": -1.4968039989471436, "step": 5575 }, { "epoch": 2.9864525840441547, "grad_norm": 8.268016813574327, "learning_rate": 5.588254946015114e-11, "logits/chosen": -0.2827422618865967, "logits/rejected": -0.07571440190076828, "logps/chosen": -1.1469650268554688, "logps/rejected": -1.4948405027389526, "loss": 1.5042, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.1469650268554688, "rewards/margins": 0.3478756844997406, "rewards/rejected": -1.4948405027389526, "step": 5580 }, { "epoch": 2.989128616825556, "grad_norm": 9.069282097255376, "learning_rate": 3.502385525216978e-11, "logits/chosen": -0.2544441521167755, "logits/rejected": -0.11987463384866714, "logps/chosen": -1.2516387701034546, "logps/rejected": -1.493402361869812, "loss": 1.5747, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.2516387701034546, "rewards/margins": 0.24176368117332458, "rewards/rejected": -1.493402361869812, "step": 5585 }, { "epoch": 2.9918046496069577, "grad_norm": 6.916859793628514, "learning_rate": 1.901582343555308e-11, "logits/chosen": -0.1889420449733734, "logits/rejected": -0.15718932449817657, "logps/chosen": -1.2972419261932373, "logps/rejected": -1.4586095809936523, "loss": 1.6364, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -1.2972419261932373, "rewards/margins": 0.1613677740097046, "rewards/rejected": -1.4586095809936523, "step": 5590 }, { "epoch": 2.9944806823883594, "grad_norm": 9.315142123250766, "learning_rate": 7.858609320232634e-12, "logits/chosen": -0.20134368538856506, "logits/rejected": -0.09880489856004715, "logps/chosen": -1.1670615673065186, "logps/rejected": -1.4292680025100708, "loss": 1.5248, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.1670615673065186, "rewards/margins": 0.262206494808197, "rewards/rejected": -1.4292680025100708, "step": 5595 }, { "epoch": 2.9971567151697607, "grad_norm": 9.052815076284157, "learning_rate": 1.5523211535639624e-12, "logits/chosen": -0.19403724372386932, "logits/rejected": -0.12277773767709732, "logps/chosen": -1.2054804563522339, "logps/rejected": -1.5248081684112549, "loss": 1.5485, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -1.2054804563522339, "rewards/margins": 0.3193278908729553, "rewards/rejected": -1.5248081684112549, "step": 5600 }, { "epoch": 2.9971567151697607, "eval_logits/chosen": 0.07478887587785721, "eval_logits/rejected": 0.14641378819942474, "eval_logps/chosen": -1.3031429052352905, "eval_logps/rejected": -1.499267339706421, "eval_loss": 1.647899866104126, "eval_rewards/accuracies": 0.5712166428565979, "eval_rewards/chosen": -1.3031429052352905, "eval_rewards/margins": 0.19612456858158112, "eval_rewards/rejected": -1.499267339706421, "eval_runtime": 40.482, "eval_samples_per_second": 33.225, "eval_steps_per_second": 8.325, "step": 5600 }, { "epoch": 2.999297541394882, "step": 5604, "total_flos": 0.0, "train_loss": 1.6264377747324006, "train_runtime": 30342.8146, "train_samples_per_second": 5.911, "train_steps_per_second": 0.185 } ], "logging_steps": 5, "max_steps": 5604, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 1000000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }