{ "best_metric": 0.7314804685814131, "best_model_checkpoint": "../saved_model/cino-small-v2_tncc-document_v3/checkpoint-1386", "epoch": 40.0, "eval_steps": 500, "global_step": 9240, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_accuracy": 0.5728260869565217, "eval_loss": 1.4385737180709839, "eval_macro-f1": 0.3838604227373153, "eval_macro-precision": 0.4180870918370973, "eval_macro-recall": 0.401101362786273, "eval_runtime": 3.7275, "eval_samples_per_second": 246.815, "eval_steps_per_second": 7.78, "eval_weighted-f1": 0.5305606310926694, "eval_weighted-precision": 0.5297428752450334, "eval_weighted-recall": 0.5728260869565217, "step": 231 }, { "epoch": 2.0, "eval_accuracy": 0.6804347826086956, "eval_loss": 0.9704039692878723, "eval_macro-f1": 0.5670979693101942, "eval_macro-precision": 0.6248049410073685, "eval_macro-recall": 0.5626229599238465, "eval_runtime": 3.649, "eval_samples_per_second": 252.121, "eval_steps_per_second": 7.947, "eval_weighted-f1": 0.659306294880126, "eval_weighted-precision": 0.6612128534838714, "eval_weighted-recall": 0.6804347826086956, "step": 462 }, { "epoch": 2.16, "learning_rate": 2.7056277056277056e-05, "loss": 1.5683, "step": 500 }, { "epoch": 3.0, "eval_accuracy": 0.7228260869565217, "eval_loss": 0.8554874658584595, "eval_macro-f1": 0.6439019323461226, "eval_macro-precision": 0.7591831945218918, "eval_macro-recall": 0.647150694134015, "eval_runtime": 3.6619, "eval_samples_per_second": 251.235, "eval_steps_per_second": 7.919, "eval_weighted-f1": 0.7074240890463663, "eval_weighted-precision": 0.7364652520925159, "eval_weighted-recall": 0.7228260869565217, "step": 693 }, { "epoch": 4.0, "eval_accuracy": 0.7217391304347827, "eval_loss": 0.8149855136871338, "eval_macro-f1": 0.692605745410065, "eval_macro-precision": 0.7440547514586789, "eval_macro-recall": 0.6752749193421123, "eval_runtime": 3.6631, "eval_samples_per_second": 251.153, "eval_steps_per_second": 7.917, "eval_weighted-f1": 0.7209306678998675, "eval_weighted-precision": 0.7414800435950141, "eval_weighted-recall": 0.7217391304347827, "step": 924 }, { "epoch": 4.33, "learning_rate": 4.9543049543049543e-05, "loss": 0.8117, "step": 1000 }, { "epoch": 5.0, "eval_accuracy": 0.7380434782608696, "eval_loss": 0.7939425110816956, "eval_macro-f1": 0.6992459555314743, "eval_macro-precision": 0.7591820184313646, "eval_macro-recall": 0.6754292333806964, "eval_runtime": 3.6533, "eval_samples_per_second": 251.83, "eval_steps_per_second": 7.938, "eval_weighted-f1": 0.7380000414393602, "eval_weighted-precision": 0.7564326290768578, "eval_weighted-recall": 0.7380434782608696, "step": 1155 }, { "epoch": 6.0, "eval_accuracy": 0.758695652173913, "eval_loss": 0.7805317640304565, "eval_macro-f1": 0.7314804685814131, "eval_macro-precision": 0.7488860778713532, "eval_macro-recall": 0.7209236372848382, "eval_runtime": 3.6563, "eval_samples_per_second": 251.62, "eval_steps_per_second": 7.932, "eval_weighted-f1": 0.7573646414610792, "eval_weighted-precision": 0.7607162388274387, "eval_weighted-recall": 0.758695652173913, "step": 1386 }, { "epoch": 6.49, "learning_rate": 4.653679653679654e-05, "loss": 0.5138, "step": 1500 }, { "epoch": 7.0, "eval_accuracy": 0.7271739130434782, "eval_loss": 0.9128503799438477, "eval_macro-f1": 0.6932349275628739, "eval_macro-precision": 0.709503845500428, "eval_macro-recall": 0.6965007036577074, "eval_runtime": 3.6549, "eval_samples_per_second": 251.715, "eval_steps_per_second": 7.934, "eval_weighted-f1": 0.7248736247795886, "eval_weighted-precision": 0.7328289572198895, "eval_weighted-recall": 0.7271739130434782, "step": 1617 }, { "epoch": 8.0, "eval_accuracy": 0.7217391304347827, "eval_loss": 0.9858090877532959, "eval_macro-f1": 0.689163447413493, "eval_macro-precision": 0.7090289269200102, "eval_macro-recall": 0.6776571823481318, "eval_runtime": 3.6807, "eval_samples_per_second": 249.952, "eval_steps_per_second": 7.879, "eval_weighted-f1": 0.7165836364208036, "eval_weighted-precision": 0.7206730807638884, "eval_weighted-recall": 0.7217391304347827, "step": 1848 }, { "epoch": 8.66, "learning_rate": 4.3530543530543535e-05, "loss": 0.2896, "step": 2000 }, { "epoch": 9.0, "eval_accuracy": 0.7206521739130435, "eval_loss": 1.0959097146987915, "eval_macro-f1": 0.6881695092418435, "eval_macro-precision": 0.7242135136847904, "eval_macro-recall": 0.6739069406781741, "eval_runtime": 3.6639, "eval_samples_per_second": 251.1, "eval_steps_per_second": 7.915, "eval_weighted-f1": 0.721541258098875, "eval_weighted-precision": 0.7354744485281628, "eval_weighted-recall": 0.7206521739130435, "step": 2079 }, { "epoch": 10.0, "eval_accuracy": 0.7402173913043478, "eval_loss": 1.0843907594680786, "eval_macro-f1": 0.7110726162631839, "eval_macro-precision": 0.7142340249185155, "eval_macro-recall": 0.7171490983289037, "eval_runtime": 3.6707, "eval_samples_per_second": 250.635, "eval_steps_per_second": 7.9, "eval_weighted-f1": 0.7389745804216191, "eval_weighted-precision": 0.7454086875631871, "eval_weighted-recall": 0.7402173913043478, "step": 2310 }, { "epoch": 10.82, "learning_rate": 4.0524290524290525e-05, "loss": 0.1697, "step": 2500 }, { "epoch": 11.0, "eval_accuracy": 0.7445652173913043, "eval_loss": 1.233216643333435, "eval_macro-f1": 0.7174385595847431, "eval_macro-precision": 0.7222220197958554, "eval_macro-recall": 0.7223862957961207, "eval_runtime": 3.6639, "eval_samples_per_second": 251.1, "eval_steps_per_second": 7.915, "eval_weighted-f1": 0.7415516921744333, "eval_weighted-precision": 0.7454222933021499, "eval_weighted-recall": 0.7445652173913043, "step": 2541 }, { "epoch": 12.0, "eval_accuracy": 0.725, "eval_loss": 1.3816906213760376, "eval_macro-f1": 0.7016354693909223, "eval_macro-precision": 0.7189672090787159, "eval_macro-recall": 0.6935770369995492, "eval_runtime": 3.6765, "eval_samples_per_second": 250.236, "eval_steps_per_second": 7.888, "eval_weighted-f1": 0.724330602886681, "eval_weighted-precision": 0.7285735967072182, "eval_weighted-recall": 0.725, "step": 2772 }, { "epoch": 12.99, "learning_rate": 3.751803751803752e-05, "loss": 0.1103, "step": 3000 }, { "epoch": 13.0, "eval_accuracy": 0.7489130434782608, "eval_loss": 1.3391927480697632, "eval_macro-f1": 0.7238074657296613, "eval_macro-precision": 0.7446967874522551, "eval_macro-recall": 0.712629424508482, "eval_runtime": 3.6667, "eval_samples_per_second": 250.906, "eval_steps_per_second": 7.909, "eval_weighted-f1": 0.7501963340512701, "eval_weighted-precision": 0.7590165879471408, "eval_weighted-recall": 0.7489130434782608, "step": 3003 }, { "epoch": 14.0, "eval_accuracy": 0.7402173913043478, "eval_loss": 1.5429991483688354, "eval_macro-f1": 0.7240452372603429, "eval_macro-precision": 0.7331291046121945, "eval_macro-recall": 0.7243088646296482, "eval_runtime": 3.6687, "eval_samples_per_second": 250.767, "eval_steps_per_second": 7.905, "eval_weighted-f1": 0.7380284868051685, "eval_weighted-precision": 0.7450380621081162, "eval_weighted-recall": 0.7402173913043478, "step": 3234 }, { "epoch": 15.0, "eval_accuracy": 0.741304347826087, "eval_loss": 1.5235261917114258, "eval_macro-f1": 0.7183262530521685, "eval_macro-precision": 0.7374055253765689, "eval_macro-recall": 0.710378384840802, "eval_runtime": 3.6757, "eval_samples_per_second": 250.291, "eval_steps_per_second": 7.89, "eval_weighted-f1": 0.7413765884828194, "eval_weighted-precision": 0.7457027535340536, "eval_weighted-recall": 0.741304347826087, "step": 3465 }, { "epoch": 15.15, "learning_rate": 3.451178451178451e-05, "loss": 0.076, "step": 3500 }, { "epoch": 16.0, "eval_accuracy": 0.7369565217391304, "eval_loss": 1.601302981376648, "eval_macro-f1": 0.7146795994306041, "eval_macro-precision": 0.731834113865428, "eval_macro-recall": 0.7046419450264447, "eval_runtime": 3.6629, "eval_samples_per_second": 251.17, "eval_steps_per_second": 7.917, "eval_weighted-f1": 0.7361455920467442, "eval_weighted-precision": 0.7414209945981161, "eval_weighted-recall": 0.7369565217391304, "step": 3696 }, { "epoch": 17.0, "eval_accuracy": 0.7521739130434782, "eval_loss": 1.6069446802139282, "eval_macro-f1": 0.72890524872289, "eval_macro-precision": 0.7550074105527117, "eval_macro-recall": 0.7143027398745371, "eval_runtime": 3.6875, "eval_samples_per_second": 249.492, "eval_steps_per_second": 7.864, "eval_weighted-f1": 0.750626634711916, "eval_weighted-precision": 0.7542096149679086, "eval_weighted-recall": 0.7521739130434782, "step": 3927 }, { "epoch": 17.32, "learning_rate": 3.1505531505531506e-05, "loss": 0.0599, "step": 4000 }, { "epoch": 18.0, "eval_accuracy": 0.7380434782608696, "eval_loss": 1.6708563566207886, "eval_macro-f1": 0.7113858264976506, "eval_macro-precision": 0.721360196572633, "eval_macro-recall": 0.7084908417098105, "eval_runtime": 3.6673, "eval_samples_per_second": 250.863, "eval_steps_per_second": 7.908, "eval_weighted-f1": 0.7362916788435598, "eval_weighted-precision": 0.7410606549056714, "eval_weighted-recall": 0.7380434782608696, "step": 4158 }, { "epoch": 19.0, "eval_accuracy": 0.7304347826086957, "eval_loss": 1.7036793231964111, "eval_macro-f1": 0.706200706354292, "eval_macro-precision": 0.725013063692726, "eval_macro-recall": 0.6960154772814597, "eval_runtime": 3.6618, "eval_samples_per_second": 251.244, "eval_steps_per_second": 7.92, "eval_weighted-f1": 0.7272025757986477, "eval_weighted-precision": 0.7299415345713408, "eval_weighted-recall": 0.7304347826086957, "step": 4389 }, { "epoch": 19.48, "learning_rate": 2.8499278499278502e-05, "loss": 0.0527, "step": 4500 }, { "epoch": 20.0, "eval_accuracy": 0.7456521739130435, "eval_loss": 1.6668744087219238, "eval_macro-f1": 0.7224498874030757, "eval_macro-precision": 0.7310054057580726, "eval_macro-recall": 0.7165591792061506, "eval_runtime": 3.6616, "eval_samples_per_second": 251.257, "eval_steps_per_second": 7.92, "eval_weighted-f1": 0.7462188604522836, "eval_weighted-precision": 0.748502249382744, "eval_weighted-recall": 0.7456521739130435, "step": 4620 }, { "epoch": 21.0, "eval_accuracy": 0.7369565217391304, "eval_loss": 1.714359998703003, "eval_macro-f1": 0.7010216405318593, "eval_macro-precision": 0.7089052674512984, "eval_macro-recall": 0.698099718977543, "eval_runtime": 3.6583, "eval_samples_per_second": 251.485, "eval_steps_per_second": 7.927, "eval_weighted-f1": 0.7359926399774661, "eval_weighted-precision": 0.7389162747277125, "eval_weighted-recall": 0.7369565217391304, "step": 4851 }, { "epoch": 21.65, "learning_rate": 2.5493025493025498e-05, "loss": 0.041, "step": 5000 }, { "epoch": 22.0, "eval_accuracy": 0.741304347826087, "eval_loss": 1.8112448453903198, "eval_macro-f1": 0.7129367870323943, "eval_macro-precision": 0.7109062719547, "eval_macro-recall": 0.7246584013705878, "eval_runtime": 3.6643, "eval_samples_per_second": 251.071, "eval_steps_per_second": 7.914, "eval_weighted-f1": 0.7373824826939279, "eval_weighted-precision": 0.742304135787633, "eval_weighted-recall": 0.741304347826087, "step": 5082 }, { "epoch": 23.0, "eval_accuracy": 0.7358695652173913, "eval_loss": 1.776945948600769, "eval_macro-f1": 0.7100717867057288, "eval_macro-precision": 0.7175705473737001, "eval_macro-recall": 0.7091991070542707, "eval_runtime": 3.6635, "eval_samples_per_second": 251.123, "eval_steps_per_second": 7.916, "eval_weighted-f1": 0.7345006810309644, "eval_weighted-precision": 0.7390098845189911, "eval_weighted-recall": 0.7358695652173913, "step": 5313 }, { "epoch": 23.81, "learning_rate": 2.2486772486772487e-05, "loss": 0.0412, "step": 5500 }, { "epoch": 24.0, "eval_accuracy": 0.75, "eval_loss": 1.747084140777588, "eval_macro-f1": 0.7227708998896428, "eval_macro-precision": 0.7279946870184277, "eval_macro-recall": 0.7269702794261114, "eval_runtime": 3.6637, "eval_samples_per_second": 251.112, "eval_steps_per_second": 7.915, "eval_weighted-f1": 0.7458470196881466, "eval_weighted-precision": 0.7496528420569815, "eval_weighted-recall": 0.75, "step": 5544 }, { "epoch": 25.0, "eval_accuracy": 0.7532608695652174, "eval_loss": 1.7346147298812866, "eval_macro-f1": 0.7300620988998959, "eval_macro-precision": 0.7254323915116926, "eval_macro-recall": 0.7396215021682684, "eval_runtime": 3.6654, "eval_samples_per_second": 250.999, "eval_steps_per_second": 7.912, "eval_weighted-f1": 0.7504535503896388, "eval_weighted-precision": 0.7526891843637886, "eval_weighted-recall": 0.7532608695652174, "step": 5775 }, { "epoch": 25.97, "learning_rate": 1.9480519480519483e-05, "loss": 0.0326, "step": 6000 }, { "epoch": 26.0, "eval_accuracy": 0.7456521739130435, "eval_loss": 1.773017168045044, "eval_macro-f1": 0.7210970682285488, "eval_macro-precision": 0.7283775906058515, "eval_macro-recall": 0.719270945602435, "eval_runtime": 3.6601, "eval_samples_per_second": 251.357, "eval_steps_per_second": 7.923, "eval_weighted-f1": 0.7454697580769286, "eval_weighted-precision": 0.7489157944224945, "eval_weighted-recall": 0.7456521739130435, "step": 6006 }, { "epoch": 27.0, "eval_accuracy": 0.741304347826087, "eval_loss": 1.7493728399276733, "eval_macro-f1": 0.7025880602354205, "eval_macro-precision": 0.7130648727750518, "eval_macro-recall": 0.6993768549946884, "eval_runtime": 3.6688, "eval_samples_per_second": 250.762, "eval_steps_per_second": 7.904, "eval_weighted-f1": 0.7414220307733658, "eval_weighted-precision": 0.7452401075563395, "eval_weighted-recall": 0.741304347826087, "step": 6237 }, { "epoch": 28.0, "eval_accuracy": 0.7380434782608696, "eval_loss": 1.7450823783874512, "eval_macro-f1": 0.7026259912227975, "eval_macro-precision": 0.7214901347374018, "eval_macro-recall": 0.6911837077117138, "eval_runtime": 3.7325, "eval_samples_per_second": 246.486, "eval_steps_per_second": 7.77, "eval_weighted-f1": 0.736513945546438, "eval_weighted-precision": 0.7385934880642558, "eval_weighted-recall": 0.7380434782608696, "step": 6468 }, { "epoch": 28.14, "learning_rate": 1.6474266474266476e-05, "loss": 0.0257, "step": 6500 }, { "epoch": 29.0, "eval_accuracy": 0.7402173913043478, "eval_loss": 1.7633217573165894, "eval_macro-f1": 0.7134642108516674, "eval_macro-precision": 0.7190433251313625, "eval_macro-recall": 0.710258338303699, "eval_runtime": 3.6817, "eval_samples_per_second": 249.884, "eval_steps_per_second": 7.877, "eval_weighted-f1": 0.7398481546483754, "eval_weighted-precision": 0.7415064106206414, "eval_weighted-recall": 0.7402173913043478, "step": 6699 }, { "epoch": 30.0, "eval_accuracy": 0.7347826086956522, "eval_loss": 1.820716381072998, "eval_macro-f1": 0.7078683429969447, "eval_macro-precision": 0.7372603591067731, "eval_macro-recall": 0.6893626457978533, "eval_runtime": 3.7068, "eval_samples_per_second": 248.19, "eval_steps_per_second": 7.823, "eval_weighted-f1": 0.7322350095276792, "eval_weighted-precision": 0.7372777947320607, "eval_weighted-recall": 0.7347826086956522, "step": 6930 }, { "epoch": 30.3, "learning_rate": 1.3468013468013468e-05, "loss": 0.0262, "step": 7000 }, { "epoch": 31.0, "eval_accuracy": 0.7510869565217392, "eval_loss": 1.76492178440094, "eval_macro-f1": 0.731348188671936, "eval_macro-precision": 0.7382460346691814, "eval_macro-recall": 0.7298883924645497, "eval_runtime": 3.6813, "eval_samples_per_second": 249.913, "eval_steps_per_second": 7.878, "eval_weighted-f1": 0.7508384750022834, "eval_weighted-precision": 0.7539289140986796, "eval_weighted-recall": 0.7510869565217392, "step": 7161 }, { "epoch": 32.0, "eval_accuracy": 0.7358695652173913, "eval_loss": 1.8048666715621948, "eval_macro-f1": 0.7135206623010025, "eval_macro-precision": 0.7206599735879954, "eval_macro-recall": 0.713146029256634, "eval_runtime": 3.6665, "eval_samples_per_second": 250.92, "eval_steps_per_second": 7.909, "eval_weighted-f1": 0.7342617186640648, "eval_weighted-precision": 0.7381810820989364, "eval_weighted-recall": 0.7358695652173913, "step": 7392 }, { "epoch": 32.47, "learning_rate": 1.0461760461760463e-05, "loss": 0.0195, "step": 7500 }, { "epoch": 33.0, "eval_accuracy": 0.7434782608695653, "eval_loss": 1.7627431154251099, "eval_macro-f1": 0.72002488158224, "eval_macro-precision": 0.727980412915839, "eval_macro-recall": 0.7171781195809488, "eval_runtime": 3.6606, "eval_samples_per_second": 251.327, "eval_steps_per_second": 7.922, "eval_weighted-f1": 0.7428966222335778, "eval_weighted-precision": 0.7455932383430586, "eval_weighted-recall": 0.7434782608695653, "step": 7623 }, { "epoch": 34.0, "eval_accuracy": 0.7445652173913043, "eval_loss": 1.8427157402038574, "eval_macro-f1": 0.7238120517774087, "eval_macro-precision": 0.7310564585025658, "eval_macro-recall": 0.7232388744393482, "eval_runtime": 3.6585, "eval_samples_per_second": 251.468, "eval_steps_per_second": 7.927, "eval_weighted-f1": 0.7436940578232905, "eval_weighted-precision": 0.7475437346184277, "eval_weighted-recall": 0.7445652173913043, "step": 7854 }, { "epoch": 34.63, "learning_rate": 7.455507455507456e-06, "loss": 0.0188, "step": 8000 }, { "epoch": 35.0, "eval_accuracy": 0.7445652173913043, "eval_loss": 1.82598078250885, "eval_macro-f1": 0.7226558273088113, "eval_macro-precision": 0.7280808646863082, "eval_macro-recall": 0.721750060320641, "eval_runtime": 3.9179, "eval_samples_per_second": 234.822, "eval_steps_per_second": 7.402, "eval_weighted-f1": 0.7433397528033767, "eval_weighted-precision": 0.7455193887868321, "eval_weighted-recall": 0.7445652173913043, "step": 8085 }, { "epoch": 36.0, "eval_accuracy": 0.7521739130434782, "eval_loss": 1.8167239427566528, "eval_macro-f1": 0.7311172141769898, "eval_macro-precision": 0.7387132761496166, "eval_macro-recall": 0.7260195259385496, "eval_runtime": 3.6724, "eval_samples_per_second": 250.519, "eval_steps_per_second": 7.897, "eval_weighted-f1": 0.7512808559429369, "eval_weighted-precision": 0.7518176004978223, "eval_weighted-recall": 0.7521739130434782, "step": 8316 }, { "epoch": 36.8, "learning_rate": 4.44925444925445e-06, "loss": 0.0176, "step": 8500 }, { "epoch": 37.0, "eval_accuracy": 0.75, "eval_loss": 1.8308371305465698, "eval_macro-f1": 0.7308495034985188, "eval_macro-precision": 0.7403845771754249, "eval_macro-recall": 0.7242967832014863, "eval_runtime": 3.6615, "eval_samples_per_second": 251.261, "eval_steps_per_second": 7.92, "eval_weighted-f1": 0.7478384491335044, "eval_weighted-precision": 0.7485903968276678, "eval_weighted-recall": 0.75, "step": 8547 }, { "epoch": 38.0, "eval_accuracy": 0.7467391304347826, "eval_loss": 1.807198405265808, "eval_macro-f1": 0.7291086447475449, "eval_macro-precision": 0.7329302540393156, "eval_macro-recall": 0.7271188451475986, "eval_runtime": 3.6625, "eval_samples_per_second": 251.195, "eval_steps_per_second": 7.918, "eval_weighted-f1": 0.7458071950555136, "eval_weighted-precision": 0.7463050676859623, "eval_weighted-recall": 0.7467391304347826, "step": 8778 }, { "epoch": 38.96, "learning_rate": 1.4430014430014432e-06, "loss": 0.0164, "step": 9000 }, { "epoch": 39.0, "eval_accuracy": 0.7456521739130435, "eval_loss": 1.8073784112930298, "eval_macro-f1": 0.7256401482067129, "eval_macro-precision": 0.733018922323657, "eval_macro-recall": 0.7204178880954032, "eval_runtime": 3.673, "eval_samples_per_second": 250.478, "eval_steps_per_second": 7.896, "eval_weighted-f1": 0.7445134806621682, "eval_weighted-precision": 0.7450872143093246, "eval_weighted-recall": 0.7456521739130435, "step": 9009 }, { "epoch": 40.0, "eval_accuracy": 0.741304347826087, "eval_loss": 1.818052887916565, "eval_macro-f1": 0.7215539592969167, "eval_macro-precision": 0.7317731195834322, "eval_macro-recall": 0.7133028256075337, "eval_runtime": 3.6643, "eval_samples_per_second": 251.069, "eval_steps_per_second": 7.914, "eval_weighted-f1": 0.7401045633293094, "eval_weighted-precision": 0.740661997113788, "eval_weighted-recall": 0.741304347826087, "step": 9240 } ], "logging_steps": 500, "max_steps": 9240, "num_input_tokens_seen": 0, "num_train_epochs": 40, "save_steps": 500, "total_flos": 3.902655553929216e+16, "train_batch_size": 32, "trial_name": null, "trial_params": null }