bhuvana-ak7's picture
Upload folder using huggingface_hub
c645722 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.3615982643283312,
"eval_steps": 500,
"global_step": 1000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0036159826432833123,
"grad_norm": 0.3235728144645691,
"learning_rate": 1.0000000000000002e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.043861389160156,
"logits/rejected": -18.043861389160156,
"logps/chosen": -0.6726381778717041,
"logps/rejected": -0.6726381778717041,
"loss": 6.9675,
"nll_loss": 1.6725690364837646,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06726382672786713,
"rewards/margins": 0.0,
"rewards/rejected": -0.06726382672786713,
"step": 10
},
{
"epoch": 0.0072319652865666245,
"grad_norm": 0.3509086072444916,
"learning_rate": 2.0000000000000003e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.013948440551758,
"logits/rejected": -18.013948440551758,
"logps/chosen": -0.6865767240524292,
"logps/rejected": -0.6865767240524292,
"loss": 6.7227,
"nll_loss": 1.6113628149032593,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06865767389535904,
"rewards/margins": 0.0,
"rewards/rejected": -0.06865767389535904,
"step": 20
},
{
"epoch": 0.010847947929849937,
"grad_norm": 0.38406211137771606,
"learning_rate": 3e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.367061614990234,
"logits/rejected": -18.367061614990234,
"logps/chosen": -0.6837120056152344,
"logps/rejected": -0.6837120056152344,
"loss": 6.3986,
"nll_loss": 1.5303384065628052,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06837119162082672,
"rewards/margins": 0.0,
"rewards/rejected": -0.06837119162082672,
"step": 30
},
{
"epoch": 0.014463930573133249,
"grad_norm": 0.4101921319961548,
"learning_rate": 4.000000000000001e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -17.99128532409668,
"logits/rejected": -17.99128532409668,
"logps/chosen": -0.6463029980659485,
"logps/rejected": -0.6463029980659485,
"loss": 7.0291,
"nll_loss": 1.6879545450210571,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06463029980659485,
"rewards/margins": 0.0,
"rewards/rejected": -0.06463029980659485,
"step": 40
},
{
"epoch": 0.01807991321641656,
"grad_norm": 0.444273978471756,
"learning_rate": 5e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.08310890197754,
"logits/rejected": -18.08310890197754,
"logps/chosen": -0.6336182355880737,
"logps/rejected": -0.6336182355880737,
"loss": 6.5453,
"nll_loss": 1.5670195817947388,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06336182355880737,
"rewards/margins": 0.0,
"rewards/rejected": -0.06336182355880737,
"step": 50
},
{
"epoch": 0.021695895859699874,
"grad_norm": 0.67641681432724,
"learning_rate": 6e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.269500732421875,
"logits/rejected": -18.269500732421875,
"logps/chosen": -0.6496673822402954,
"logps/rejected": -0.6496673822402954,
"loss": 6.733,
"nll_loss": 1.6139262914657593,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06496674567461014,
"rewards/margins": 0.0,
"rewards/rejected": -0.06496674567461014,
"step": 60
},
{
"epoch": 0.025311878502983186,
"grad_norm": 0.39636465907096863,
"learning_rate": 7e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.115657806396484,
"logits/rejected": -18.115657806396484,
"logps/chosen": -0.6373459696769714,
"logps/rejected": -0.6373459696769714,
"loss": 6.8825,
"nll_loss": 1.651307463645935,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06373460590839386,
"rewards/margins": 0.0,
"rewards/rejected": -0.06373460590839386,
"step": 70
},
{
"epoch": 0.028927861146266498,
"grad_norm": 0.5290302634239197,
"learning_rate": 8.000000000000001e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.138675689697266,
"logits/rejected": -18.138675689697266,
"logps/chosen": -0.6702518463134766,
"logps/rejected": -0.6702518463134766,
"loss": 6.4063,
"nll_loss": 1.5322693586349487,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06702519208192825,
"rewards/margins": 0.0,
"rewards/rejected": -0.06702519208192825,
"step": 80
},
{
"epoch": 0.03254384378954981,
"grad_norm": 0.2837754189968109,
"learning_rate": 9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.285091400146484,
"logits/rejected": -18.285091400146484,
"logps/chosen": -0.5695949196815491,
"logps/rejected": -0.5695949196815491,
"loss": 6.3599,
"nll_loss": 1.520659327507019,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.056959498673677444,
"rewards/margins": 0.0,
"rewards/rejected": -0.056959498673677444,
"step": 90
},
{
"epoch": 0.03615982643283312,
"grad_norm": 0.38816723227500916,
"learning_rate": 1e-05,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.38321304321289,
"logits/rejected": -18.38321304321289,
"logps/chosen": -0.6220490336418152,
"logps/rejected": -0.6220490336418152,
"loss": 6.3579,
"nll_loss": 1.5201700925827026,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.062204908579587936,
"rewards/margins": 0.0,
"rewards/rejected": -0.062204908579587936,
"step": 100
},
{
"epoch": 0.039775809076116435,
"grad_norm": 0.485514760017395,
"learning_rate": 9.88888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.24630355834961,
"logits/rejected": -18.24630355834961,
"logps/chosen": -0.58094322681427,
"logps/rejected": -0.58094322681427,
"loss": 6.8083,
"nll_loss": 1.6327617168426514,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.058094322681427,
"rewards/margins": 0.0,
"rewards/rejected": -0.058094322681427,
"step": 110
},
{
"epoch": 0.04339179171939975,
"grad_norm": 0.7200180292129517,
"learning_rate": 9.777777777777779e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.222978591918945,
"logits/rejected": -18.222978591918945,
"logps/chosen": -0.6140703558921814,
"logps/rejected": -0.6140703558921814,
"loss": 7.0165,
"nll_loss": 1.6848167181015015,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06140704080462456,
"rewards/margins": 0.0,
"rewards/rejected": -0.06140704080462456,
"step": 120
},
{
"epoch": 0.04700777436268306,
"grad_norm": 0.5130624175071716,
"learning_rate": 9.666666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -17.98419189453125,
"logits/rejected": -17.98419189453125,
"logps/chosen": -0.6171376705169678,
"logps/rejected": -0.6171376705169678,
"loss": 6.9743,
"nll_loss": 1.6742585897445679,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.06171376630663872,
"rewards/margins": 0.0,
"rewards/rejected": -0.06171376630663872,
"step": 130
},
{
"epoch": 0.05062375700596637,
"grad_norm": 0.6743359565734863,
"learning_rate": 9.555555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.43941879272461,
"logits/rejected": -18.43941879272461,
"logps/chosen": -0.5736243724822998,
"logps/rejected": -0.5736243724822998,
"loss": 6.4179,
"nll_loss": 1.5351518392562866,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.057362429797649384,
"rewards/margins": 0.0,
"rewards/rejected": -0.057362429797649384,
"step": 140
},
{
"epoch": 0.054239739649249684,
"grad_norm": 0.7007283568382263,
"learning_rate": 9.444444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.523662567138672,
"logits/rejected": -18.523662567138672,
"logps/chosen": -0.5047087073326111,
"logps/rejected": -0.5047087073326111,
"loss": 6.4734,
"nll_loss": 1.549041986465454,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.050470877438783646,
"rewards/margins": 0.0,
"rewards/rejected": -0.050470877438783646,
"step": 150
},
{
"epoch": 0.057855722292532996,
"grad_norm": 0.8055678009986877,
"learning_rate": 9.333333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.57369041442871,
"logits/rejected": -18.57369041442871,
"logps/chosen": -0.48697715997695923,
"logps/rejected": -0.48697715997695923,
"loss": 6.1125,
"nll_loss": 1.4588209390640259,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.04869771748781204,
"rewards/margins": 0.0,
"rewards/rejected": -0.04869771748781204,
"step": 160
},
{
"epoch": 0.06147170493581631,
"grad_norm": 0.5678962469100952,
"learning_rate": 9.222222222222224e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.546266555786133,
"logits/rejected": -18.546266555786133,
"logps/chosen": -0.4713471531867981,
"logps/rejected": -0.4713471531867981,
"loss": 6.3675,
"nll_loss": 1.5225670337677002,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.04713470861315727,
"rewards/margins": 0.0,
"rewards/rejected": -0.04713470861315727,
"step": 170
},
{
"epoch": 0.06508768757909962,
"grad_norm": 0.9227916598320007,
"learning_rate": 9.111111111111112e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.552295684814453,
"logits/rejected": -18.552295684814453,
"logps/chosen": -0.5135122537612915,
"logps/rejected": -0.5135122537612915,
"loss": 6.3925,
"nll_loss": 1.5288182497024536,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.05135122686624527,
"rewards/margins": 0.0,
"rewards/rejected": -0.05135122686624527,
"step": 180
},
{
"epoch": 0.06870367022238293,
"grad_norm": 1.3030140399932861,
"learning_rate": 9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.453067779541016,
"logits/rejected": -18.453067779541016,
"logps/chosen": -0.4881868362426758,
"logps/rejected": -0.4881868362426758,
"loss": 6.5908,
"nll_loss": 1.5783830881118774,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.0488186851143837,
"rewards/margins": 0.0,
"rewards/rejected": -0.0488186851143837,
"step": 190
},
{
"epoch": 0.07231965286566625,
"grad_norm": 1.0002509355545044,
"learning_rate": 8.888888888888888e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.477296829223633,
"logits/rejected": -18.477296829223633,
"logps/chosen": -0.46165475249290466,
"logps/rejected": -0.46165475249290466,
"loss": 6.6221,
"nll_loss": 1.5862023830413818,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.046165481209754944,
"rewards/margins": 0.0,
"rewards/rejected": -0.046165481209754944,
"step": 200
},
{
"epoch": 0.07593563550894955,
"grad_norm": 0.7885683178901672,
"learning_rate": 8.777777777777778e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.564895629882812,
"logits/rejected": -18.564895629882812,
"logps/chosen": -0.41595011949539185,
"logps/rejected": -0.41595011949539185,
"loss": 6.2789,
"nll_loss": 1.500412940979004,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.04159501940011978,
"rewards/margins": 0.0,
"rewards/rejected": -0.04159501940011978,
"step": 210
},
{
"epoch": 0.07955161815223287,
"grad_norm": 0.6379426717758179,
"learning_rate": 8.666666666666668e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.577289581298828,
"logits/rejected": -18.577289581298828,
"logps/chosen": -0.43426617980003357,
"logps/rejected": -0.43426617980003357,
"loss": 6.4009,
"nll_loss": 1.5309035778045654,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.043426621705293655,
"rewards/margins": 0.0,
"rewards/rejected": -0.043426621705293655,
"step": 220
},
{
"epoch": 0.08316760079551618,
"grad_norm": 0.5890640020370483,
"learning_rate": 8.555555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.817012786865234,
"logits/rejected": -18.817012786865234,
"logps/chosen": -0.38255172967910767,
"logps/rejected": -0.38255172967910767,
"loss": 6.021,
"nll_loss": 1.4359278678894043,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.038255173712968826,
"rewards/margins": 0.0,
"rewards/rejected": -0.038255173712968826,
"step": 230
},
{
"epoch": 0.0867835834387995,
"grad_norm": 0.5675965547561646,
"learning_rate": 8.444444444444446e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.88204002380371,
"logits/rejected": -18.88204002380371,
"logps/chosen": -0.357022225856781,
"logps/rejected": -0.357022225856781,
"loss": 6.3796,
"nll_loss": 1.5255934000015259,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.03570222482085228,
"rewards/margins": 0.0,
"rewards/rejected": -0.03570222482085228,
"step": 240
},
{
"epoch": 0.0903995660820828,
"grad_norm": 0.284084677696228,
"learning_rate": 8.344444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.659196853637695,
"logits/rejected": -18.659196853637695,
"logps/chosen": -0.35400137305259705,
"logps/rejected": -0.35400137305259705,
"loss": 6.1544,
"nll_loss": 1.4692747592926025,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.03540014103055,
"rewards/margins": 0.0,
"rewards/rejected": -0.03540014103055,
"step": 250
},
{
"epoch": 0.09401554872536612,
"grad_norm": 0.6424693465232849,
"learning_rate": 8.233333333333335e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.762996673583984,
"logits/rejected": -18.762996673583984,
"logps/chosen": -0.40258026123046875,
"logps/rejected": -0.40258026123046875,
"loss": 6.3857,
"nll_loss": 1.5271098613739014,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.040258027613162994,
"rewards/margins": 0.0,
"rewards/rejected": -0.040258027613162994,
"step": 260
},
{
"epoch": 0.09763153136864942,
"grad_norm": 0.6682632565498352,
"learning_rate": 8.122222222222223e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.719924926757812,
"logits/rejected": -18.719924926757812,
"logps/chosen": -0.3123939633369446,
"logps/rejected": -0.3123939633369446,
"loss": 5.7456,
"nll_loss": 1.36708664894104,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.031239395961165428,
"rewards/margins": 0.0,
"rewards/rejected": -0.031239395961165428,
"step": 270
},
{
"epoch": 0.10124751401193274,
"grad_norm": 0.9669603109359741,
"learning_rate": 8.011111111111113e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.799579620361328,
"logits/rejected": -18.799579620361328,
"logps/chosen": -0.3817462623119354,
"logps/rejected": -0.3817462623119354,
"loss": 6.2838,
"nll_loss": 1.5016276836395264,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.03817462921142578,
"rewards/margins": 0.0,
"rewards/rejected": -0.03817462921142578,
"step": 280
},
{
"epoch": 0.10486349665521605,
"grad_norm": 0.7264003157615662,
"learning_rate": 7.9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.090185165405273,
"logits/rejected": -19.090185165405273,
"logps/chosen": -0.3294925093650818,
"logps/rejected": -0.3294925093650818,
"loss": 5.6699,
"nll_loss": 1.3481695652008057,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.03294925019145012,
"rewards/margins": 0.0,
"rewards/rejected": -0.03294925019145012,
"step": 290
},
{
"epoch": 0.10847947929849937,
"grad_norm": 0.564611554145813,
"learning_rate": 7.788888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.767759323120117,
"logits/rejected": -18.767759323120117,
"logps/chosen": -0.26777949929237366,
"logps/rejected": -0.26777949929237366,
"loss": 6.5104,
"nll_loss": 1.5582915544509888,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.026777952909469604,
"rewards/margins": 0.0,
"rewards/rejected": -0.026777952909469604,
"step": 300
},
{
"epoch": 0.11209546194178267,
"grad_norm": 0.6952475905418396,
"learning_rate": 7.677777777777778e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.896703720092773,
"logits/rejected": -18.896703720092773,
"logps/chosen": -0.24664482474327087,
"logps/rejected": -0.24664482474327087,
"loss": 5.9777,
"nll_loss": 1.4251067638397217,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02466448023915291,
"rewards/margins": 0.0,
"rewards/rejected": -0.02466448023915291,
"step": 310
},
{
"epoch": 0.11571144458506599,
"grad_norm": 0.7094094753265381,
"learning_rate": 7.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.993297576904297,
"logits/rejected": -18.993297576904297,
"logps/chosen": -0.27139216661453247,
"logps/rejected": -0.27139216661453247,
"loss": 5.848,
"nll_loss": 1.3926928043365479,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.027139216661453247,
"rewards/margins": 0.0,
"rewards/rejected": -0.027139216661453247,
"step": 320
},
{
"epoch": 0.1193274272283493,
"grad_norm": 0.8386672735214233,
"learning_rate": 7.455555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.818639755249023,
"logits/rejected": -18.818639755249023,
"logps/chosen": -0.2643309533596039,
"logps/rejected": -0.2643309533596039,
"loss": 6.2247,
"nll_loss": 1.4868563413619995,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.026433095335960388,
"rewards/margins": 0.0,
"rewards/rejected": -0.026433095335960388,
"step": 330
},
{
"epoch": 0.12294340987163262,
"grad_norm": 0.6337667107582092,
"learning_rate": 7.344444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.07388687133789,
"logits/rejected": -19.07388687133789,
"logps/chosen": -0.23179857432842255,
"logps/rejected": -0.23179857432842255,
"loss": 5.7871,
"nll_loss": 1.3774592876434326,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.023179858922958374,
"rewards/margins": 0.0,
"rewards/rejected": -0.023179858922958374,
"step": 340
},
{
"epoch": 0.12655939251491594,
"grad_norm": 1.056504726409912,
"learning_rate": 7.233333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.8520450592041,
"logits/rejected": -18.8520450592041,
"logps/chosen": -0.24602051079273224,
"logps/rejected": -0.24602051079273224,
"loss": 5.8268,
"nll_loss": 1.3873755931854248,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.024602051824331284,
"rewards/margins": 0.0,
"rewards/rejected": -0.024602051824331284,
"step": 350
},
{
"epoch": 0.13017537515819924,
"grad_norm": 0.4462037682533264,
"learning_rate": 7.122222222222222e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.04658317565918,
"logits/rejected": -19.04658317565918,
"logps/chosen": -0.2291136234998703,
"logps/rejected": -0.2291136234998703,
"loss": 5.8142,
"nll_loss": 1.384232521057129,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02291136048734188,
"rewards/margins": 0.0,
"rewards/rejected": -0.02291136048734188,
"step": 360
},
{
"epoch": 0.13379135780148255,
"grad_norm": 0.5579663515090942,
"learning_rate": 7.011111111111112e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -18.912765502929688,
"logits/rejected": -18.912765502929688,
"logps/chosen": -0.26528915762901306,
"logps/rejected": -0.26528915762901306,
"loss": 6.0316,
"nll_loss": 1.4385900497436523,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.026528915390372276,
"rewards/margins": 0.0,
"rewards/rejected": -0.026528915390372276,
"step": 370
},
{
"epoch": 0.13740734044476585,
"grad_norm": 1.2433960437774658,
"learning_rate": 6.9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.217700958251953,
"logits/rejected": -19.217700958251953,
"logps/chosen": -0.21606405079364777,
"logps/rejected": -0.21606405079364777,
"loss": 5.7944,
"nll_loss": 1.3792803287506104,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.021606406196951866,
"rewards/margins": 0.0,
"rewards/rejected": -0.021606406196951866,
"step": 380
},
{
"epoch": 0.14102332308804919,
"grad_norm": 0.7657280564308167,
"learning_rate": 6.788888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.175457000732422,
"logits/rejected": -19.175457000732422,
"logps/chosen": -0.2413448989391327,
"logps/rejected": -0.2413448989391327,
"loss": 5.7839,
"nll_loss": 1.3766554594039917,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02413449063897133,
"rewards/margins": 0.0,
"rewards/rejected": -0.02413449063897133,
"step": 390
},
{
"epoch": 0.1446393057313325,
"grad_norm": 0.7070457935333252,
"learning_rate": 6.677777777777779e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.15035629272461,
"logits/rejected": -19.15035629272461,
"logps/chosen": -0.25096631050109863,
"logps/rejected": -0.25096631050109863,
"loss": 6.2307,
"nll_loss": 1.4883568286895752,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.025096634402871132,
"rewards/margins": 0.0,
"rewards/rejected": -0.025096634402871132,
"step": 400
},
{
"epoch": 0.1482552883746158,
"grad_norm": 0.43025216460227966,
"learning_rate": 6.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.02223014831543,
"logits/rejected": -19.02223014831543,
"logps/chosen": -0.24732474982738495,
"logps/rejected": -0.24732474982738495,
"loss": 6.2865,
"nll_loss": 1.5023012161254883,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.024732474237680435,
"rewards/margins": 0.0,
"rewards/rejected": -0.024732474237680435,
"step": 410
},
{
"epoch": 0.1518712710178991,
"grad_norm": 1.3701528310775757,
"learning_rate": 6.455555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.089460372924805,
"logits/rejected": -19.089460372924805,
"logps/chosen": -0.23749932646751404,
"logps/rejected": -0.23749932646751404,
"loss": 6.2238,
"nll_loss": 1.4866377115249634,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.023749932646751404,
"rewards/margins": 0.0,
"rewards/rejected": -0.023749932646751404,
"step": 420
},
{
"epoch": 0.15548725366118243,
"grad_norm": 0.9421939253807068,
"learning_rate": 6.3444444444444454e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.231048583984375,
"logits/rejected": -19.231048583984375,
"logps/chosen": -0.19140982627868652,
"logps/rejected": -0.19140982627868652,
"loss": 6.0814,
"nll_loss": 1.451047420501709,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01914098486304283,
"rewards/margins": 0.0,
"rewards/rejected": -0.01914098486304283,
"step": 430
},
{
"epoch": 0.15910323630446574,
"grad_norm": 0.9060840010643005,
"learning_rate": 6.2333333333333335e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.107379913330078,
"logits/rejected": -19.107379913330078,
"logps/chosen": -0.2809773087501526,
"logps/rejected": -0.2809773087501526,
"loss": 6.3671,
"nll_loss": 1.522457480430603,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.0280977301299572,
"rewards/margins": 0.0,
"rewards/rejected": -0.0280977301299572,
"step": 440
},
{
"epoch": 0.16271921894774904,
"grad_norm": 0.5965518355369568,
"learning_rate": 6.1222222222222224e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.34296417236328,
"logits/rejected": -19.34296417236328,
"logps/chosen": -0.19397786259651184,
"logps/rejected": -0.19397786259651184,
"loss": 5.7975,
"nll_loss": 1.3800629377365112,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.019397784024477005,
"rewards/margins": 0.0,
"rewards/rejected": -0.019397784024477005,
"step": 450
},
{
"epoch": 0.16633520159103235,
"grad_norm": 0.9895502328872681,
"learning_rate": 6.011111111111112e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.35286521911621,
"logits/rejected": -19.35286521911621,
"logps/chosen": -0.20646443963050842,
"logps/rejected": -0.20646443963050842,
"loss": 5.8574,
"nll_loss": 1.3950278759002686,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02064644545316696,
"rewards/margins": 0.0,
"rewards/rejected": -0.02064644545316696,
"step": 460
},
{
"epoch": 0.16995118423431568,
"grad_norm": 1.6251282691955566,
"learning_rate": 5.9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.29703140258789,
"logits/rejected": -19.29703140258789,
"logps/chosen": -0.23044386506080627,
"logps/rejected": -0.23044386506080627,
"loss": 6.0743,
"nll_loss": 1.4492676258087158,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.023044386878609657,
"rewards/margins": 0.0,
"rewards/rejected": -0.023044386878609657,
"step": 470
},
{
"epoch": 0.173567166877599,
"grad_norm": 0.7959076762199402,
"learning_rate": 5.788888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.089004516601562,
"logits/rejected": -19.089004516601562,
"logps/chosen": -0.1713530272245407,
"logps/rejected": -0.1713530272245407,
"loss": 6.0034,
"nll_loss": 1.4315412044525146,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01713530346751213,
"rewards/margins": 0.0,
"rewards/rejected": -0.01713530346751213,
"step": 480
},
{
"epoch": 0.1771831495208823,
"grad_norm": 0.6285834908485413,
"learning_rate": 5.677777777777779e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.485876083374023,
"logits/rejected": -19.485876083374023,
"logps/chosen": -0.18361307680606842,
"logps/rejected": -0.18361307680606842,
"loss": 5.6858,
"nll_loss": 1.3521363735198975,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.018361307680606842,
"rewards/margins": 0.0,
"rewards/rejected": -0.018361307680606842,
"step": 490
},
{
"epoch": 0.1807991321641656,
"grad_norm": 0.9556539058685303,
"learning_rate": 5.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.300710678100586,
"logits/rejected": -19.300710678100586,
"logps/chosen": -0.17280462384223938,
"logps/rejected": -0.17280462384223938,
"loss": 6.0335,
"nll_loss": 1.439051866531372,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01728046126663685,
"rewards/margins": 0.0,
"rewards/rejected": -0.01728046126663685,
"step": 500
},
{
"epoch": 0.18441511480744893,
"grad_norm": 0.7111396789550781,
"learning_rate": 5.455555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.141691207885742,
"logits/rejected": -19.141691207885742,
"logps/chosen": -0.17267338931560516,
"logps/rejected": -0.17267338931560516,
"loss": 6.1707,
"nll_loss": 1.4733521938323975,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017267340794205666,
"rewards/margins": 0.0,
"rewards/rejected": -0.017267340794205666,
"step": 510
},
{
"epoch": 0.18803109745073224,
"grad_norm": 1.1565580368041992,
"learning_rate": 5.344444444444446e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.17026138305664,
"logits/rejected": -19.17026138305664,
"logps/chosen": -0.25084125995635986,
"logps/rejected": -0.25084125995635986,
"loss": 6.1347,
"nll_loss": 1.464364767074585,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.025084126740694046,
"rewards/margins": 0.0,
"rewards/rejected": -0.025084126740694046,
"step": 520
},
{
"epoch": 0.19164708009401554,
"grad_norm": 0.8986483216285706,
"learning_rate": 5.233333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.375362396240234,
"logits/rejected": -19.375362396240234,
"logps/chosen": -0.24914030730724335,
"logps/rejected": -0.24914030730724335,
"loss": 6.1778,
"nll_loss": 1.475141167640686,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.024914031848311424,
"rewards/margins": 0.0,
"rewards/rejected": -0.024914031848311424,
"step": 530
},
{
"epoch": 0.19526306273729885,
"grad_norm": 1.2064629793167114,
"learning_rate": 5.122222222222223e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.600208282470703,
"logits/rejected": -19.600208282470703,
"logps/chosen": -0.18557456135749817,
"logps/rejected": -0.18557456135749817,
"loss": 5.4563,
"nll_loss": 1.2947726249694824,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.018557453528046608,
"rewards/margins": 0.0,
"rewards/rejected": -0.018557453528046608,
"step": 540
},
{
"epoch": 0.19887904538058218,
"grad_norm": 0.6842054128646851,
"learning_rate": 5.011111111111111e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.55367660522461,
"logits/rejected": -19.55367660522461,
"logps/chosen": -0.19609031081199646,
"logps/rejected": -0.19609031081199646,
"loss": 6.0472,
"nll_loss": 1.442474365234375,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.019609034061431885,
"rewards/margins": 0.0,
"rewards/rejected": -0.019609034061431885,
"step": 550
},
{
"epoch": 0.2024950280238655,
"grad_norm": 0.5375315546989441,
"learning_rate": 4.9000000000000005e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.148658752441406,
"logits/rejected": -19.148658752441406,
"logps/chosen": -0.17001472413539886,
"logps/rejected": -0.17001472413539886,
"loss": 5.6006,
"nll_loss": 1.3308324813842773,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017001474276185036,
"rewards/margins": 0.0,
"rewards/rejected": -0.017001474276185036,
"step": 560
},
{
"epoch": 0.2061110106671488,
"grad_norm": 1.169607400894165,
"learning_rate": 4.7888888888888894e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.166362762451172,
"logits/rejected": -19.166362762451172,
"logps/chosen": -0.19456727802753448,
"logps/rejected": -0.19456727802753448,
"loss": 6.1838,
"nll_loss": 1.476626992225647,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01945672743022442,
"rewards/margins": 0.0,
"rewards/rejected": -0.01945672743022442,
"step": 570
},
{
"epoch": 0.2097269933104321,
"grad_norm": 1.1300163269042969,
"learning_rate": 4.677777777777778e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.370128631591797,
"logits/rejected": -19.370128631591797,
"logps/chosen": -0.20290282368659973,
"logps/rejected": -0.20290282368659973,
"loss": 5.8201,
"nll_loss": 1.385710597038269,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.020290281623601913,
"rewards/margins": 0.0,
"rewards/rejected": -0.020290281623601913,
"step": 580
},
{
"epoch": 0.21334297595371543,
"grad_norm": 1.181201696395874,
"learning_rate": 4.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.285415649414062,
"logits/rejected": -19.285415649414062,
"logps/chosen": -0.18727798759937286,
"logps/rejected": -0.18727798759937286,
"loss": 5.8618,
"nll_loss": 1.3961261510849,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.018727798014879227,
"rewards/margins": 0.0,
"rewards/rejected": -0.018727798014879227,
"step": 590
},
{
"epoch": 0.21695895859699874,
"grad_norm": 0.8772425651550293,
"learning_rate": 4.455555555555555e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.32810401916504,
"logits/rejected": -19.32810401916504,
"logps/chosen": -0.1583147644996643,
"logps/rejected": -0.1583147644996643,
"loss": 5.9621,
"nll_loss": 1.4212169647216797,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.0158314760774374,
"rewards/margins": 0.0,
"rewards/rejected": -0.0158314760774374,
"step": 600
},
{
"epoch": 0.22057494124028204,
"grad_norm": 0.9875885248184204,
"learning_rate": 4.344444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.371606826782227,
"logits/rejected": -19.371606826782227,
"logps/chosen": -0.15842756628990173,
"logps/rejected": -0.15842756628990173,
"loss": 5.9785,
"nll_loss": 1.4253188371658325,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.015842756256461143,
"rewards/margins": 0.0,
"rewards/rejected": -0.015842756256461143,
"step": 610
},
{
"epoch": 0.22419092388356535,
"grad_norm": 0.8926348090171814,
"learning_rate": 4.233333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.365840911865234,
"logits/rejected": -19.365840911865234,
"logps/chosen": -0.2225889265537262,
"logps/rejected": -0.2225889265537262,
"loss": 5.9332,
"nll_loss": 1.413989782333374,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02225889265537262,
"rewards/margins": 0.0,
"rewards/rejected": -0.02225889265537262,
"step": 620
},
{
"epoch": 0.22780690652684868,
"grad_norm": 1.2021228075027466,
"learning_rate": 4.122222222222222e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.250947952270508,
"logits/rejected": -19.250947952270508,
"logps/chosen": -0.20725078880786896,
"logps/rejected": -0.20725078880786896,
"loss": 5.8929,
"nll_loss": 1.4039117097854614,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.020725077018141747,
"rewards/margins": 0.0,
"rewards/rejected": -0.020725077018141747,
"step": 630
},
{
"epoch": 0.23142288917013198,
"grad_norm": 0.6190844774246216,
"learning_rate": 4.011111111111111e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.514095306396484,
"logits/rejected": -19.514095306396484,
"logps/chosen": -0.14317741990089417,
"logps/rejected": -0.14317741990089417,
"loss": 5.655,
"nll_loss": 1.3444368839263916,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.014317741617560387,
"rewards/margins": 0.0,
"rewards/rejected": -0.014317741617560387,
"step": 640
},
{
"epoch": 0.2350388718134153,
"grad_norm": 0.6763967871665955,
"learning_rate": 3.900000000000001e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.486291885375977,
"logits/rejected": -19.486291885375977,
"logps/chosen": -0.19646582007408142,
"logps/rejected": -0.19646582007408142,
"loss": 6.1549,
"nll_loss": 1.4694058895111084,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.019646581262350082,
"rewards/margins": 0.0,
"rewards/rejected": -0.019646581262350082,
"step": 650
},
{
"epoch": 0.2386548544566986,
"grad_norm": 1.276583194732666,
"learning_rate": 3.7888888888888893e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.337284088134766,
"logits/rejected": -19.337284088134766,
"logps/chosen": -0.16177110373973846,
"logps/rejected": -0.16177110373973846,
"loss": 6.0415,
"nll_loss": 1.441070318222046,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.016177110373973846,
"rewards/margins": 0.0,
"rewards/rejected": -0.016177110373973846,
"step": 660
},
{
"epoch": 0.24227083709998193,
"grad_norm": 1.4721999168395996,
"learning_rate": 3.6777777777777778e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.851795196533203,
"logits/rejected": -19.851795196533203,
"logps/chosen": -0.19354157149791718,
"logps/rejected": -0.19354157149791718,
"loss": 6.0983,
"nll_loss": 1.455251932144165,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.019354157149791718,
"rewards/margins": 0.0,
"rewards/rejected": -0.019354157149791718,
"step": 670
},
{
"epoch": 0.24588681974326523,
"grad_norm": 1.205200433731079,
"learning_rate": 3.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.199329376220703,
"logits/rejected": -19.199329376220703,
"logps/chosen": -0.2025156319141388,
"logps/rejected": -0.2025156319141388,
"loss": 6.0431,
"nll_loss": 1.441453218460083,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.0202515609562397,
"rewards/margins": 0.0,
"rewards/rejected": -0.0202515609562397,
"step": 680
},
{
"epoch": 0.24950280238654854,
"grad_norm": 0.4968813359737396,
"learning_rate": 3.455555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.286746978759766,
"logits/rejected": -19.286746978759766,
"logps/chosen": -0.20702144503593445,
"logps/rejected": -0.20702144503593445,
"loss": 6.0379,
"nll_loss": 1.4401479959487915,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.020702145993709564,
"rewards/margins": 0.0,
"rewards/rejected": -0.020702145993709564,
"step": 690
},
{
"epoch": 0.25311878502983187,
"grad_norm": 0.49326029419898987,
"learning_rate": 3.3444444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.63970184326172,
"logits/rejected": -19.63970184326172,
"logps/chosen": -0.1862333118915558,
"logps/rejected": -0.1862333118915558,
"loss": 5.6812,
"nll_loss": 1.350982666015625,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.018623333424329758,
"rewards/margins": 0.0,
"rewards/rejected": -0.018623333424329758,
"step": 700
},
{
"epoch": 0.25673476767311515,
"grad_norm": 1.1381062269210815,
"learning_rate": 3.2333333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.339035034179688,
"logits/rejected": -19.339035034179688,
"logps/chosen": -0.22874228656291962,
"logps/rejected": -0.22874228656291962,
"loss": 6.5147,
"nll_loss": 1.5593668222427368,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.022874226793646812,
"rewards/margins": 0.0,
"rewards/rejected": -0.022874226793646812,
"step": 710
},
{
"epoch": 0.2603507503163985,
"grad_norm": 1.3999911546707153,
"learning_rate": 3.1222222222222228e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.305469512939453,
"logits/rejected": -19.305469512939453,
"logps/chosen": -0.1623169630765915,
"logps/rejected": -0.1623169630765915,
"loss": 5.9032,
"nll_loss": 1.4064949750900269,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01623169705271721,
"rewards/margins": 0.0,
"rewards/rejected": -0.01623169705271721,
"step": 720
},
{
"epoch": 0.2639667329596818,
"grad_norm": 0.6383623480796814,
"learning_rate": 3.0111111111111113e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.534765243530273,
"logits/rejected": -19.534765243530273,
"logps/chosen": -0.15669655799865723,
"logps/rejected": -0.15669655799865723,
"loss": 5.723,
"nll_loss": 1.3614407777786255,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.015669656917452812,
"rewards/margins": 0.0,
"rewards/rejected": -0.015669656917452812,
"step": 730
},
{
"epoch": 0.2675827156029651,
"grad_norm": 1.1800575256347656,
"learning_rate": 2.9e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.569406509399414,
"logits/rejected": -19.569406509399414,
"logps/chosen": -0.17918285727500916,
"logps/rejected": -0.17918285727500916,
"loss": 5.7004,
"nll_loss": 1.355780839920044,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017918284982442856,
"rewards/margins": 0.0,
"rewards/rejected": -0.017918284982442856,
"step": 740
},
{
"epoch": 0.2711986982462484,
"grad_norm": 0.5818102359771729,
"learning_rate": 2.788888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.516826629638672,
"logits/rejected": -19.516826629638672,
"logps/chosen": -0.14605708420276642,
"logps/rejected": -0.14605708420276642,
"loss": 5.8445,
"nll_loss": 1.391806960105896,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.014605708420276642,
"rewards/margins": 0.0,
"rewards/rejected": -0.014605708420276642,
"step": 750
},
{
"epoch": 0.2748146808895317,
"grad_norm": 0.6804122924804688,
"learning_rate": 2.677777777777778e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.40943145751953,
"logits/rejected": -19.40943145751953,
"logps/chosen": -0.17038683593273163,
"logps/rejected": -0.17038683593273163,
"loss": 5.8438,
"nll_loss": 1.391636848449707,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017038684338331223,
"rewards/margins": 0.0,
"rewards/rejected": -0.017038684338331223,
"step": 760
},
{
"epoch": 0.27843066353281504,
"grad_norm": 0.7628147006034851,
"learning_rate": 2.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.423494338989258,
"logits/rejected": -19.423494338989258,
"logps/chosen": -0.20526845753192902,
"logps/rejected": -0.20526845753192902,
"loss": 6.2205,
"nll_loss": 1.4858171939849854,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.02052684687077999,
"rewards/margins": 0.0,
"rewards/rejected": -0.02052684687077999,
"step": 770
},
{
"epoch": 0.28204664617609837,
"grad_norm": 0.6365695595741272,
"learning_rate": 2.455555555555556e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.535381317138672,
"logits/rejected": -19.535381317138672,
"logps/chosen": -0.138637512922287,
"logps/rejected": -0.138637512922287,
"loss": 5.848,
"nll_loss": 1.392683744430542,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.013863752596080303,
"rewards/margins": 0.0,
"rewards/rejected": -0.013863752596080303,
"step": 780
},
{
"epoch": 0.28566262881938165,
"grad_norm": 0.4407297968864441,
"learning_rate": 2.3444444444444448e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.312089920043945,
"logits/rejected": -19.312089920043945,
"logps/chosen": -0.21358434855937958,
"logps/rejected": -0.21358434855937958,
"loss": 5.8644,
"nll_loss": 1.396776795387268,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.021358435973525047,
"rewards/margins": 0.0,
"rewards/rejected": -0.021358435973525047,
"step": 790
},
{
"epoch": 0.289278611462665,
"grad_norm": 1.0635472536087036,
"learning_rate": 2.2333333333333333e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.445270538330078,
"logits/rejected": -19.445270538330078,
"logps/chosen": -0.14682866632938385,
"logps/rejected": -0.14682866632938385,
"loss": 5.5576,
"nll_loss": 1.3200775384902954,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.014682864770293236,
"rewards/margins": 0.0,
"rewards/rejected": -0.014682864770293236,
"step": 800
},
{
"epoch": 0.2928945941059483,
"grad_norm": 0.6877497434616089,
"learning_rate": 2.1222222222222226e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.4804744720459,
"logits/rejected": -19.4804744720459,
"logps/chosen": -0.20306849479675293,
"logps/rejected": -0.20306849479675293,
"loss": 6.2889,
"nll_loss": 1.5029165744781494,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.020306849852204323,
"rewards/margins": 0.0,
"rewards/rejected": -0.020306849852204323,
"step": 810
},
{
"epoch": 0.2965105767492316,
"grad_norm": 0.7547898888587952,
"learning_rate": 2.011111111111111e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.336137771606445,
"logits/rejected": -19.336137771606445,
"logps/chosen": -0.18079710006713867,
"logps/rejected": -0.18079710006713867,
"loss": 5.777,
"nll_loss": 1.3749300241470337,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.018079709261655807,
"rewards/margins": 0.0,
"rewards/rejected": -0.018079709261655807,
"step": 820
},
{
"epoch": 0.3001265593925149,
"grad_norm": 1.389726996421814,
"learning_rate": 1.9000000000000002e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.307830810546875,
"logits/rejected": -19.307830810546875,
"logps/chosen": -0.15548691153526306,
"logps/rejected": -0.15548691153526306,
"loss": 6.3126,
"nll_loss": 1.5088294744491577,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.015548691153526306,
"rewards/margins": 0.0,
"rewards/rejected": -0.015548691153526306,
"step": 830
},
{
"epoch": 0.3037425420357982,
"grad_norm": 0.6280723810195923,
"learning_rate": 1.788888888888889e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.246450424194336,
"logits/rejected": -19.246450424194336,
"logps/chosen": -0.19246172904968262,
"logps/rejected": -0.19246172904968262,
"loss": 6.0085,
"nll_loss": 1.4328036308288574,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01924617402255535,
"rewards/margins": 0.0,
"rewards/rejected": -0.01924617402255535,
"step": 840
},
{
"epoch": 0.30735852467908154,
"grad_norm": 0.9361159205436707,
"learning_rate": 1.6777777777777779e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.374317169189453,
"logits/rejected": -19.374317169189453,
"logps/chosen": -0.20890088379383087,
"logps/rejected": -0.20890088379383087,
"loss": 5.9353,
"nll_loss": 1.4145065546035767,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.020890090614557266,
"rewards/margins": 0.0,
"rewards/rejected": -0.020890090614557266,
"step": 850
},
{
"epoch": 0.31097450732236487,
"grad_norm": 0.8236832022666931,
"learning_rate": 1.566666666666667e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.475786209106445,
"logits/rejected": -19.475786209106445,
"logps/chosen": -0.1413622796535492,
"logps/rejected": -0.1413622796535492,
"loss": 5.9494,
"nll_loss": 1.4180256128311157,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.014136227779090405,
"rewards/margins": 0.0,
"rewards/rejected": -0.014136227779090405,
"step": 860
},
{
"epoch": 0.31459048996564815,
"grad_norm": 0.7201619148254395,
"learning_rate": 1.4555555555555557e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.682842254638672,
"logits/rejected": -19.682842254638672,
"logps/chosen": -0.18202020227909088,
"logps/rejected": -0.18202020227909088,
"loss": 5.9195,
"nll_loss": 1.410556674003601,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01820201985538006,
"rewards/margins": 0.0,
"rewards/rejected": -0.01820201985538006,
"step": 870
},
{
"epoch": 0.3182064726089315,
"grad_norm": 0.886303722858429,
"learning_rate": 1.3444444444444446e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.420122146606445,
"logits/rejected": -19.420122146606445,
"logps/chosen": -0.17583726346492767,
"logps/rejected": -0.17583726346492767,
"loss": 5.5973,
"nll_loss": 1.3300113677978516,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017583727836608887,
"rewards/margins": 0.0,
"rewards/rejected": -0.017583727836608887,
"step": 880
},
{
"epoch": 0.3218224552522148,
"grad_norm": 0.8513539433479309,
"learning_rate": 1.2444444444444445e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.468677520751953,
"logits/rejected": -19.468677520751953,
"logps/chosen": -0.16836388409137726,
"logps/rejected": -0.16836388409137726,
"loss": 6.062,
"nll_loss": 1.4461798667907715,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.016836389899253845,
"rewards/margins": 0.0,
"rewards/rejected": -0.016836389899253845,
"step": 890
},
{
"epoch": 0.3254384378954981,
"grad_norm": 1.0691941976547241,
"learning_rate": 1.1333333333333334e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.57880210876465,
"logits/rejected": -19.57880210876465,
"logps/chosen": -0.16049222648143768,
"logps/rejected": -0.16049222648143768,
"loss": 5.9859,
"nll_loss": 1.4271578788757324,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01604922115802765,
"rewards/margins": 0.0,
"rewards/rejected": -0.01604922115802765,
"step": 900
},
{
"epoch": 0.3290544205387814,
"grad_norm": 1.168095350265503,
"learning_rate": 1.0222222222222223e-06,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.2463436126709,
"logits/rejected": -19.2463436126709,
"logps/chosen": -0.1521718055009842,
"logps/rejected": -0.1521718055009842,
"loss": 5.71,
"nll_loss": 1.3581907749176025,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.015217180363833904,
"rewards/margins": 0.0,
"rewards/rejected": -0.015217180363833904,
"step": 910
},
{
"epoch": 0.3326704031820647,
"grad_norm": 0.862614631652832,
"learning_rate": 9.111111111111113e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.247844696044922,
"logits/rejected": -19.247844696044922,
"logps/chosen": -0.19881650805473328,
"logps/rejected": -0.19881650805473328,
"loss": 5.9249,
"nll_loss": 1.4119006395339966,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.019881650805473328,
"rewards/margins": 0.0,
"rewards/rejected": -0.019881650805473328,
"step": 920
},
{
"epoch": 0.33628638582534803,
"grad_norm": 0.9760550260543823,
"learning_rate": 8.000000000000001e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.753040313720703,
"logits/rejected": -19.753040313720703,
"logps/chosen": -0.17076358199119568,
"logps/rejected": -0.17076358199119568,
"loss": 5.8841,
"nll_loss": 1.401706337928772,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.017076358199119568,
"rewards/margins": 0.0,
"rewards/rejected": -0.017076358199119568,
"step": 930
},
{
"epoch": 0.33990236846863137,
"grad_norm": 0.8302357196807861,
"learning_rate": 6.88888888888889e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.467557907104492,
"logits/rejected": -19.467557907104492,
"logps/chosen": -0.15472976863384247,
"logps/rejected": -0.15472976863384247,
"loss": 5.5121,
"nll_loss": 1.308708906173706,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.015472980216145515,
"rewards/margins": 0.0,
"rewards/rejected": -0.015472980216145515,
"step": 940
},
{
"epoch": 0.34351835111191464,
"grad_norm": 1.4063400030136108,
"learning_rate": 5.777777777777778e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.383520126342773,
"logits/rejected": -19.383520126342773,
"logps/chosen": -0.16120824217796326,
"logps/rejected": -0.16120824217796326,
"loss": 6.0657,
"nll_loss": 1.4471219778060913,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.016120824962854385,
"rewards/margins": 0.0,
"rewards/rejected": -0.016120824962854385,
"step": 950
},
{
"epoch": 0.347134333755198,
"grad_norm": 1.351473331451416,
"learning_rate": 4.666666666666667e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.597942352294922,
"logits/rejected": -19.597942352294922,
"logps/chosen": -0.1628998965024948,
"logps/rejected": -0.1628998965024948,
"loss": 5.8956,
"nll_loss": 1.4045830965042114,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01628998853266239,
"rewards/margins": 0.0,
"rewards/rejected": -0.01628998853266239,
"step": 960
},
{
"epoch": 0.3507503163984813,
"grad_norm": 0.688176691532135,
"learning_rate": 3.555555555555556e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.574413299560547,
"logits/rejected": -19.574413299560547,
"logps/chosen": -0.12833839654922485,
"logps/rejected": -0.12833839654922485,
"loss": 5.6608,
"nll_loss": 1.3458738327026367,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.012833841145038605,
"rewards/margins": 0.0,
"rewards/rejected": -0.012833841145038605,
"step": 970
},
{
"epoch": 0.3543662990417646,
"grad_norm": 0.605377733707428,
"learning_rate": 2.444444444444445e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -20.091583251953125,
"logits/rejected": -20.091583251953125,
"logps/chosen": -0.1359986811876297,
"logps/rejected": -0.1359986811876297,
"loss": 5.6425,
"nll_loss": 1.3413007259368896,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.013599867932498455,
"rewards/margins": 0.0,
"rewards/rejected": -0.013599867932498455,
"step": 980
},
{
"epoch": 0.3579822816850479,
"grad_norm": 1.1121692657470703,
"learning_rate": 1.3333333333333336e-07,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.476083755493164,
"logits/rejected": -19.476083755493164,
"logps/chosen": -0.1606629192829132,
"logps/rejected": -0.1606629192829132,
"loss": 5.9177,
"nll_loss": 1.4101154804229736,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.01606629230082035,
"rewards/margins": 0.0,
"rewards/rejected": -0.01606629230082035,
"step": 990
},
{
"epoch": 0.3615982643283312,
"grad_norm": 0.8480359315872192,
"learning_rate": 2.2222222222222224e-08,
"log_odds_chosen": 0.0,
"log_odds_ratio": -0.6931472420692444,
"logits/chosen": -19.61067771911621,
"logits/rejected": -19.61067771911621,
"logps/chosen": -0.14116862416267395,
"logps/rejected": -0.14116862416267395,
"loss": 5.7495,
"nll_loss": 1.3680593967437744,
"rewards/accuracies": 0.0,
"rewards/chosen": -0.014116862788796425,
"rewards/margins": 0.0,
"rewards/rejected": -0.014116862788796425,
"step": 1000
}
],
"logging_steps": 10,
"max_steps": 1000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"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": 4,
"trial_name": null,
"trial_params": null
}