| { |
| "best_global_step": 2500, |
| "best_metric": 0.44659083423553114, |
| "best_model_checkpoint": "checkpoints/med-ruri-v3-70m-v2-from-ruri/checkpoint-2500", |
| "epoch": 0.4059100503328462, |
| "eval_steps": 500, |
| "global_step": 2500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.008118201006656925, |
| "grad_norm": 21.997957229614258, |
| "learning_rate": 1.590909090909091e-06, |
| "loss": 2.7372756958007813, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.01623640201331385, |
| "grad_norm": 21.73430824279785, |
| "learning_rate": 3.2142857142857147e-06, |
| "loss": 1.9263325500488282, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.024354603019970774, |
| "grad_norm": 20.49755859375, |
| "learning_rate": 4.837662337662339e-06, |
| "loss": 1.7200509643554687, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.0324728040266277, |
| "grad_norm": 21.036407470703125, |
| "learning_rate": 6.461038961038961e-06, |
| "loss": 1.5285859680175782, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.040591005033284625, |
| "grad_norm": 21.67646026611328, |
| "learning_rate": 8.084415584415586e-06, |
| "loss": 1.4216513061523437, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.04870920603994155, |
| "grad_norm": 30.95376205444336, |
| "learning_rate": 9.707792207792209e-06, |
| "loss": 1.382916259765625, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.05682740704659847, |
| "grad_norm": 22.470050811767578, |
| "learning_rate": 9.929926508289183e-06, |
| "loss": 1.316688995361328, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.0649456080532554, |
| "grad_norm": 31.66095733642578, |
| "learning_rate": 9.84447103059306e-06, |
| "loss": 1.284674072265625, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.07306380905991232, |
| "grad_norm": 20.316326141357422, |
| "learning_rate": 9.759015552896942e-06, |
| "loss": 1.18936767578125, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.08118201006656925, |
| "grad_norm": 23.299427032470703, |
| "learning_rate": 9.67356007520082e-06, |
| "loss": 1.1284925842285156, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.08118201006656925, |
| "eval_med-ir_cosine_accuracy@1": 0.2215, |
| "eval_med-ir_cosine_accuracy@10": 0.5745, |
| "eval_med-ir_cosine_accuracy@5": 0.4715, |
| "eval_med-ir_cosine_map@10": 0.32694821428571424, |
| "eval_med-ir_cosine_mrr@10": 0.3269482142857145, |
| "eval_med-ir_cosine_ndcg@10": 0.3859062398415415, |
| "eval_med-ir_cosine_precision@1": 0.2215, |
| "eval_med-ir_cosine_precision@10": 0.057449999999999994, |
| "eval_med-ir_cosine_precision@5": 0.0943, |
| "eval_med-ir_cosine_recall@1": 0.2215, |
| "eval_med-ir_cosine_recall@10": 0.5745, |
| "eval_med-ir_cosine_recall@5": 0.4715, |
| "eval_runtime": 53.6164, |
| "eval_samples_per_second": 0.0, |
| "eval_steps_per_second": 0.0, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.08930021107322618, |
| "grad_norm": 22.585927963256836, |
| "learning_rate": 9.5881045975047e-06, |
| "loss": 1.1239456176757812, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.0974184120798831, |
| "grad_norm": 23.06036949157715, |
| "learning_rate": 9.50264911980858e-06, |
| "loss": 1.0830574035644531, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.10553661308654003, |
| "grad_norm": 23.491195678710938, |
| "learning_rate": 9.41719364211246e-06, |
| "loss": 1.0685610961914063, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.11365481409319694, |
| "grad_norm": 21.600292205810547, |
| "learning_rate": 9.33173816441634e-06, |
| "loss": 1.018815460205078, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.12177301509985387, |
| "grad_norm": 28.022537231445312, |
| "learning_rate": 9.24628268672022e-06, |
| "loss": 0.9987781524658204, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.1298912161065108, |
| "grad_norm": 23.952239990234375, |
| "learning_rate": 9.160827209024098e-06, |
| "loss": 0.9764260864257812, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.13800941711316772, |
| "grad_norm": 22.52235984802246, |
| "learning_rate": 9.07537173132798e-06, |
| "loss": 0.9670017242431641, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.14612761811982464, |
| "grad_norm": 24.34895896911621, |
| "learning_rate": 8.989916253631858e-06, |
| "loss": 0.9470697021484376, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.15424581912648158, |
| "grad_norm": 22.89519500732422, |
| "learning_rate": 8.904460775935739e-06, |
| "loss": 0.9199741363525391, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.1623640201331385, |
| "grad_norm": 22.926559448242188, |
| "learning_rate": 8.819005298239618e-06, |
| "loss": 0.8991819000244141, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.1623640201331385, |
| "eval_med-ir_cosine_accuracy@1": 0.223, |
| "eval_med-ir_cosine_accuracy@10": 0.609, |
| "eval_med-ir_cosine_accuracy@5": 0.497, |
| "eval_med-ir_cosine_map@10": 0.3382640873015873, |
| "eval_med-ir_cosine_mrr@10": 0.3382640873015877, |
| "eval_med-ir_cosine_ndcg@10": 0.4028251022017164, |
| "eval_med-ir_cosine_precision@1": 0.223, |
| "eval_med-ir_cosine_precision@10": 0.060899999999999996, |
| "eval_med-ir_cosine_precision@5": 0.09939999999999999, |
| "eval_med-ir_cosine_recall@1": 0.223, |
| "eval_med-ir_cosine_recall@10": 0.609, |
| "eval_med-ir_cosine_recall@5": 0.497, |
| "eval_runtime": 53.4694, |
| "eval_samples_per_second": 0.0, |
| "eval_steps_per_second": 0.0, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.17048222113979541, |
| "grad_norm": 19.02985191345215, |
| "learning_rate": 8.733549820543497e-06, |
| "loss": 0.8809585571289062, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.17860042214645236, |
| "grad_norm": 23.444557189941406, |
| "learning_rate": 8.648094342847378e-06, |
| "loss": 0.875724868774414, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.18671862315310928, |
| "grad_norm": 25.26430320739746, |
| "learning_rate": 8.562638865151257e-06, |
| "loss": 0.8596250152587891, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.1948368241597662, |
| "grad_norm": 21.99104118347168, |
| "learning_rate": 8.477183387455136e-06, |
| "loss": 0.8121035766601562, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.2029550251664231, |
| "grad_norm": 25.18592071533203, |
| "learning_rate": 8.391727909759017e-06, |
| "loss": 0.8457485961914063, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.21107322617308005, |
| "grad_norm": 26.478687286376953, |
| "learning_rate": 8.306272432062896e-06, |
| "loss": 0.7970269012451172, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.21919142717973697, |
| "grad_norm": 22.500226974487305, |
| "learning_rate": 8.220816954366777e-06, |
| "loss": 0.781522216796875, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.22730962818639389, |
| "grad_norm": 25.17734146118164, |
| "learning_rate": 8.135361476670656e-06, |
| "loss": 0.7632185363769531, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.23542782919305083, |
| "grad_norm": 26.727651596069336, |
| "learning_rate": 8.049905998974535e-06, |
| "loss": 0.7543510437011719, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.24354603019970775, |
| "grad_norm": 22.24140739440918, |
| "learning_rate": 7.964450521278416e-06, |
| "loss": 0.7415306854248047, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.24354603019970775, |
| "eval_med-ir_cosine_accuracy@1": 0.231, |
| "eval_med-ir_cosine_accuracy@10": 0.6335, |
| "eval_med-ir_cosine_accuracy@5": 0.513, |
| "eval_med-ir_cosine_map@10": 0.35125734126984126, |
| "eval_med-ir_cosine_mrr@10": 0.3512573412698419, |
| "eval_med-ir_cosine_ndcg@10": 0.4185390527584874, |
| "eval_med-ir_cosine_precision@1": 0.231, |
| "eval_med-ir_cosine_precision@10": 0.06335, |
| "eval_med-ir_cosine_precision@5": 0.1026, |
| "eval_med-ir_cosine_recall@1": 0.231, |
| "eval_med-ir_cosine_recall@10": 0.6335, |
| "eval_med-ir_cosine_recall@5": 0.513, |
| "eval_runtime": 53.5243, |
| "eval_samples_per_second": 0.0, |
| "eval_steps_per_second": 0.0, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.25166423120636466, |
| "grad_norm": 30.241344451904297, |
| "learning_rate": 7.878995043582293e-06, |
| "loss": 0.731693115234375, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.2597824322130216, |
| "grad_norm": 24.415237426757812, |
| "learning_rate": 7.793539565886174e-06, |
| "loss": 0.747947769165039, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.2679006332196785, |
| "grad_norm": 22.14430809020996, |
| "learning_rate": 7.708084088190053e-06, |
| "loss": 0.7206668853759766, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.27601883422633544, |
| "grad_norm": 20.674978256225586, |
| "learning_rate": 7.622628610493933e-06, |
| "loss": 0.7117887115478516, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.2841370352329924, |
| "grad_norm": 25.1578426361084, |
| "learning_rate": 7.537173132797813e-06, |
| "loss": 0.7387435150146484, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.2922552362396493, |
| "grad_norm": 22.64226722717285, |
| "learning_rate": 7.451717655101693e-06, |
| "loss": 0.7028619384765625, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.3003734372463062, |
| "grad_norm": 28.462968826293945, |
| "learning_rate": 7.366262177405572e-06, |
| "loss": 0.7132760620117188, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.30849163825296316, |
| "grad_norm": 21.3095703125, |
| "learning_rate": 7.280806699709452e-06, |
| "loss": 0.6857410430908203, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.31660983925962005, |
| "grad_norm": 27.10430145263672, |
| "learning_rate": 7.1953512220133315e-06, |
| "loss": 0.6693791961669922, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.324728040266277, |
| "grad_norm": 28.126373291015625, |
| "learning_rate": 7.109895744317211e-06, |
| "loss": 0.6854069519042969, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.324728040266277, |
| "eval_med-ir_cosine_accuracy@1": 0.25, |
| "eval_med-ir_cosine_accuracy@10": 0.651, |
| "eval_med-ir_cosine_accuracy@5": 0.535, |
| "eval_med-ir_cosine_map@10": 0.36545555555555553, |
| "eval_med-ir_cosine_mrr@10": 0.3654555555555565, |
| "eval_med-ir_cosine_ndcg@10": 0.43330019996953195, |
| "eval_med-ir_cosine_precision@1": 0.25, |
| "eval_med-ir_cosine_precision@10": 0.06509999999999999, |
| "eval_med-ir_cosine_precision@5": 0.107, |
| "eval_med-ir_cosine_recall@1": 0.25, |
| "eval_med-ir_cosine_recall@10": 0.651, |
| "eval_med-ir_cosine_recall@5": 0.535, |
| "eval_runtime": 52.9599, |
| "eval_samples_per_second": 0.0, |
| "eval_steps_per_second": 0.0, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.33284624127293394, |
| "grad_norm": 25.855501174926758, |
| "learning_rate": 7.0244402666210905e-06, |
| "loss": 0.6711996459960937, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.34096444227959083, |
| "grad_norm": 27.489627838134766, |
| "learning_rate": 6.93898478892497e-06, |
| "loss": 0.6556209564208985, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.3490826432862478, |
| "grad_norm": 25.698232650756836, |
| "learning_rate": 6.85352931122885e-06, |
| "loss": 0.6434734344482422, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.3572008442929047, |
| "grad_norm": 22.26129913330078, |
| "learning_rate": 6.76807383353273e-06, |
| "loss": 0.6550411224365235, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.3653190452995616, |
| "grad_norm": 22.00887680053711, |
| "learning_rate": 6.68261835583661e-06, |
| "loss": 0.614831314086914, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.37343724630621855, |
| "grad_norm": 25.09625816345215, |
| "learning_rate": 6.597162878140489e-06, |
| "loss": 0.6087098693847657, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.38155544731287544, |
| "grad_norm": 24.351856231689453, |
| "learning_rate": 6.511707400444369e-06, |
| "loss": 0.6338196182250977, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.3896736483195324, |
| "grad_norm": 25.116064071655273, |
| "learning_rate": 6.426251922748249e-06, |
| "loss": 0.6470194244384766, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.39779184932618933, |
| "grad_norm": 26.57986831665039, |
| "learning_rate": 6.340796445052129e-06, |
| "loss": 0.5991011047363282, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.4059100503328462, |
| "grad_norm": 28.94403648376465, |
| "learning_rate": 6.255340967356008e-06, |
| "loss": 0.6004145812988281, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.4059100503328462, |
| "eval_med-ir_cosine_accuracy@1": 0.2515, |
| "eval_med-ir_cosine_accuracy@10": 0.668, |
| "eval_med-ir_cosine_accuracy@5": 0.5525, |
| "eval_med-ir_cosine_map@10": 0.3772464285714286, |
| "eval_med-ir_cosine_mrr@10": 0.377246428571429, |
| "eval_med-ir_cosine_ndcg@10": 0.44659083423553114, |
| "eval_med-ir_cosine_precision@1": 0.2515, |
| "eval_med-ir_cosine_precision@10": 0.0668, |
| "eval_med-ir_cosine_precision@5": 0.1105, |
| "eval_med-ir_cosine_recall@1": 0.2515, |
| "eval_med-ir_cosine_recall@10": 0.668, |
| "eval_med-ir_cosine_recall@5": 0.5525, |
| "eval_runtime": 53.1323, |
| "eval_samples_per_second": 0.0, |
| "eval_steps_per_second": 0.0, |
| "step": 2500 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 6159, |
| "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": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 192, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
