| { | |
| "best_metric": 0.9129580069246931, | |
| "best_model_checkpoint": "./saved_models/cliroberta_sbdh_gpt4_v2_0/checkpoint-912", | |
| "epoch": 38.0, | |
| "eval_steps": 500, | |
| "global_step": 912, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.9901933670043945, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.575, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_acc_macro": 0.1223753452493468, | |
| "eval_acc_micro": 0.14617249740214502, | |
| "eval_auc_macro": 0.6179498256291489, | |
| "eval_auc_micro": 0.6115176410375945, | |
| "eval_f1_at_5": 0.24653561710044558, | |
| "eval_f1_at_8": 0.23250745850924381, | |
| "eval_f1_macro": 0.20626681869979227, | |
| "eval_f1_micro": 0.25506195225142775, | |
| "eval_loss": 0.4440538287162781, | |
| "eval_prec_at_5": 0.15890410958904108, | |
| "eval_prec_at_8": 0.13684360730593606, | |
| "eval_prec_macro": 0.16142693103921257, | |
| "eval_prec_micro": 0.15594974131559208, | |
| "eval_rec_at_5": 0.5496575342465754, | |
| "eval_rec_at_8": 0.772640791476408, | |
| "eval_rec_macro": 0.598680624031903, | |
| "eval_rec_micro": 0.6998341625206717, | |
| "eval_runtime": 3.2806, | |
| "eval_samples_per_second": 267.022, | |
| "eval_steps_per_second": 33.53, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.45090538263320923, | |
| "learning_rate": 5e-06, | |
| "loss": 0.416, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_acc_macro": 0.12473782113690822, | |
| "eval_acc_micro": 0.19038767923525782, | |
| "eval_auc_macro": 0.7812955931842048, | |
| "eval_auc_micro": 0.7120923874787535, | |
| "eval_f1_at_5": 0.28976519882341023, | |
| "eval_f1_at_8": 0.26129064546572484, | |
| "eval_f1_macro": 0.20647502666276477, | |
| "eval_f1_micro": 0.3198750836493277, | |
| "eval_loss": 0.35849881172180176, | |
| "eval_prec_at_5": 0.1860730593607306, | |
| "eval_prec_at_8": 0.15368150684931506, | |
| "eval_prec_macro": 0.2441853686080976, | |
| "eval_prec_micro": 0.2187976808056082, | |
| "eval_rec_at_5": 0.654490106544901, | |
| "eval_rec_at_8": 0.8715753424657534, | |
| "eval_rec_macro": 0.4255708847629977, | |
| "eval_rec_micro": 0.5945273631840303, | |
| "eval_runtime": 3.1796, | |
| "eval_samples_per_second": 275.508, | |
| "eval_steps_per_second": 34.596, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.27397865056991577, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3638, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_acc_macro": 0.08130874759470162, | |
| "eval_acc_micro": 0.24049000644743773, | |
| "eval_auc_macro": 0.8877837878947817, | |
| "eval_auc_micro": 0.7666590840482023, | |
| "eval_f1_at_5": 0.336822490351584, | |
| "eval_f1_at_8": 0.2698854516748229, | |
| "eval_f1_macro": 0.10904515062726224, | |
| "eval_f1_micro": 0.3877338877338474, | |
| "eval_loss": 0.3340158760547638, | |
| "eval_prec_at_5": 0.21575342465753425, | |
| "eval_prec_at_8": 0.158675799086758, | |
| "eval_prec_macro": 0.08635061077641497, | |
| "eval_prec_micro": 0.5194986072422675, | |
| "eval_rec_at_5": 0.7675038051750379, | |
| "eval_rec_at_8": 0.9022070015220699, | |
| "eval_rec_macro": 0.147953216374198, | |
| "eval_rec_micro": 0.309286898839112, | |
| "eval_runtime": 3.1991, | |
| "eval_samples_per_second": 273.823, | |
| "eval_steps_per_second": 34.384, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.2640039920806885, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3328, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_acc_macro": 0.2356290306850708, | |
| "eval_acc_micro": 0.36202531645567326, | |
| "eval_auc_macro": 0.9194506958206632, | |
| "eval_auc_micro": 0.8662944233628048, | |
| "eval_f1_at_5": 0.3860939867346734, | |
| "eval_f1_at_8": 0.2863186309798368, | |
| "eval_f1_macro": 0.31624180053900214, | |
| "eval_f1_micro": 0.5315985130111031, | |
| "eval_loss": 0.29880067706108093, | |
| "eval_prec_at_5": 0.24703196347031964, | |
| "eval_prec_at_8": 0.168236301369863, | |
| "eval_prec_macro": 0.38775967404251327, | |
| "eval_prec_micro": 0.6046511627906337, | |
| "eval_rec_at_5": 0.8833713850837139, | |
| "eval_rec_at_8": 0.9604261796042617, | |
| "eval_rec_macro": 0.31299638776931427, | |
| "eval_rec_micro": 0.4742951907130618, | |
| "eval_runtime": 3.2144, | |
| "eval_samples_per_second": 272.527, | |
| "eval_steps_per_second": 34.221, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.29883790016174316, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.2961, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_acc_macro": 0.45969839854913025, | |
| "eval_acc_micro": 0.5302557704304098, | |
| "eval_auc_macro": 0.9400078444800295, | |
| "eval_auc_micro": 0.9362775366665492, | |
| "eval_f1_at_5": 0.41206764411360836, | |
| "eval_f1_at_8": 0.2904319455576605, | |
| "eval_f1_macro": 0.5639799795799724, | |
| "eval_f1_micro": 0.6930289441499639, | |
| "eval_loss": 0.25566670298576355, | |
| "eval_prec_at_5": 0.26324200913242013, | |
| "eval_prec_at_8": 0.170662100456621, | |
| "eval_prec_macro": 0.5869010270714891, | |
| "eval_prec_micro": 0.6816359262228804, | |
| "eval_rec_at_5": 0.9480593607305936, | |
| "eval_rec_at_8": 0.9739345509893454, | |
| "eval_rec_macro": 0.5895666448012687, | |
| "eval_rec_micro": 0.7048092868987806, | |
| "eval_runtime": 3.2031, | |
| "eval_samples_per_second": 273.486, | |
| "eval_steps_per_second": 34.342, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 0.3095230162143707, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.2582, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_acc_macro": 0.5571172225508884, | |
| "eval_acc_micro": 0.599110546378615, | |
| "eval_auc_macro": 0.957377398100574, | |
| "eval_auc_micro": 0.9585021379241767, | |
| "eval_f1_at_5": 0.41988358958639926, | |
| "eval_f1_at_8": 0.2904319455576605, | |
| "eval_f1_macro": 0.6798913628949355, | |
| "eval_f1_micro": 0.7493047278505561, | |
| "eval_loss": 0.22171612083911896, | |
| "eval_prec_at_5": 0.2684931506849315, | |
| "eval_prec_at_8": 0.170662100456621, | |
| "eval_prec_macro": 0.7360995034850415, | |
| "eval_prec_micro": 0.7192982456139801, | |
| "eval_rec_at_5": 0.9627092846270927, | |
| "eval_rec_at_8": 0.9739345509893454, | |
| "eval_rec_macro": 0.7032528362779372, | |
| "eval_rec_micro": 0.7819237147594708, | |
| "eval_runtime": 3.2002, | |
| "eval_samples_per_second": 273.729, | |
| "eval_steps_per_second": 34.372, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.4028068780899048, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.2275, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_acc_macro": 0.6549888166459882, | |
| "eval_acc_micro": 0.6949270326615222, | |
| "eval_auc_macro": 0.9666862869707499, | |
| "eval_auc_micro": 0.9705963698236378, | |
| "eval_f1_at_5": 0.421956752081853, | |
| "eval_f1_at_8": 0.2911364149806836, | |
| "eval_f1_macro": 0.769457395477212, | |
| "eval_f1_micro": 0.8200082000819336, | |
| "eval_loss": 0.19522725045681, | |
| "eval_prec_at_5": 0.2698630136986302, | |
| "eval_prec_at_8": 0.17109018264840184, | |
| "eval_prec_macro": 0.8425529900891876, | |
| "eval_prec_micro": 0.8110300081102343, | |
| "eval_rec_at_5": 0.9668949771689498, | |
| "eval_rec_at_8": 0.9758371385083714, | |
| "eval_rec_macro": 0.7739718206004228, | |
| "eval_rec_micro": 0.8291873963515066, | |
| "eval_runtime": 3.1872, | |
| "eval_samples_per_second": 274.847, | |
| "eval_steps_per_second": 34.513, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.3119203448295593, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.2028, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_acc_macro": 0.6972915753471085, | |
| "eval_acc_micro": 0.7357664233576106, | |
| "eval_auc_macro": 0.9740916521582301, | |
| "eval_auc_micro": 0.9777448811533707, | |
| "eval_f1_at_5": 0.4238663640920515, | |
| "eval_f1_at_8": 0.2915919235644548, | |
| "eval_f1_macro": 0.8073127904534504, | |
| "eval_f1_micro": 0.8477712363329817, | |
| "eval_loss": 0.17399446666240692, | |
| "eval_prec_at_5": 0.2712328767123288, | |
| "eval_prec_at_8": 0.1713755707762557, | |
| "eval_prec_macro": 0.8738560982440289, | |
| "eval_prec_micro": 0.8600682593855921, | |
| "eval_rec_at_5": 0.9693683409436834, | |
| "eval_rec_at_8": 0.9767884322678843, | |
| "eval_rec_macro": 0.7867760719656439, | |
| "eval_rec_micro": 0.8358208955223188, | |
| "eval_runtime": 3.1788, | |
| "eval_samples_per_second": 275.577, | |
| "eval_steps_per_second": 34.604, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.3134922683238983, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.1824, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_acc_macro": 0.7304201087212143, | |
| "eval_acc_micro": 0.7586206896551179, | |
| "eval_auc_macro": 0.9803240639791783, | |
| "eval_auc_micro": 0.9830083410585158, | |
| "eval_f1_at_5": 0.42521790481745503, | |
| "eval_f1_at_8": 0.2922793659426448, | |
| "eval_f1_macro": 0.835259929842726, | |
| "eval_f1_micro": 0.8627450980391451, | |
| "eval_loss": 0.1565641313791275, | |
| "eval_prec_at_5": 0.2721461187214612, | |
| "eval_prec_at_8": 0.17180365296803654, | |
| "eval_prec_macro": 0.8537779731664906, | |
| "eval_prec_micro": 0.8502415458936513, | |
| "eval_rec_at_5": 0.9718417047184171, | |
| "eval_rec_at_8": 0.978310502283105, | |
| "eval_rec_macro": 0.838195231781537, | |
| "eval_rec_micro": 0.8756218905471911, | |
| "eval_runtime": 3.2483, | |
| "eval_samples_per_second": 269.681, | |
| "eval_steps_per_second": 33.864, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 0.4643474221229553, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.1651, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_acc_macro": 0.7537121442185194, | |
| "eval_acc_micro": 0.7786705624542893, | |
| "eval_auc_macro": 0.9820077423929909, | |
| "eval_auc_micro": 0.9843423829345286, | |
| "eval_f1_at_5": 0.4252361135724823, | |
| "eval_f1_at_8": 0.2915919235644548, | |
| "eval_f1_macro": 0.852437891908891, | |
| "eval_f1_micro": 0.8755646817247741, | |
| "eval_loss": 0.14288866519927979, | |
| "eval_prec_at_5": 0.2721461187214612, | |
| "eval_prec_at_8": 0.1713755707762557, | |
| "eval_prec_macro": 0.8628168198092824, | |
| "eval_prec_micro": 0.8673718470300352, | |
| "eval_rec_at_5": 0.9720319634703196, | |
| "eval_rec_at_8": 0.9767884322678843, | |
| "eval_rec_macro": 0.8531543520415448, | |
| "eval_rec_micro": 0.8839137645107061, | |
| "eval_runtime": 3.1569, | |
| "eval_samples_per_second": 277.491, | |
| "eval_steps_per_second": 34.845, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 0.41132673621177673, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.1507, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_acc_macro": 0.7725786729049252, | |
| "eval_acc_micro": 0.7965465465464867, | |
| "eval_auc_macro": 0.9845391365423227, | |
| "eval_auc_micro": 0.9861133832235769, | |
| "eval_f1_at_5": 0.42516324511829706, | |
| "eval_f1_at_8": 0.2915919235644548, | |
| "eval_f1_macro": 0.8654438329084839, | |
| "eval_f1_micro": 0.8867530296697962, | |
| "eval_loss": 0.13122519850730896, | |
| "eval_prec_at_5": 0.2721461187214612, | |
| "eval_prec_at_8": 0.1713755707762557, | |
| "eval_prec_macro": 0.890999273408568, | |
| "eval_prec_micro": 0.8938500421229238, | |
| "eval_rec_at_5": 0.9712709284627093, | |
| "eval_rec_at_8": 0.9767884322678843, | |
| "eval_rec_macro": 0.8506625538285683, | |
| "eval_rec_micro": 0.8797678275289486, | |
| "eval_runtime": 3.1904, | |
| "eval_samples_per_second": 274.571, | |
| "eval_steps_per_second": 34.478, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 0.5822508335113525, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.1381, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_acc_macro": 0.7941812566844084, | |
| "eval_acc_micro": 0.8098434004473669, | |
| "eval_auc_macro": 0.9868855804701235, | |
| "eval_auc_micro": 0.9890811185137426, | |
| "eval_f1_at_5": 0.4261995886925304, | |
| "eval_f1_at_8": 0.29205588274802374, | |
| "eval_f1_macro": 0.8806793746749714, | |
| "eval_f1_micro": 0.8949320148330534, | |
| "eval_loss": 0.12080539762973785, | |
| "eval_prec_at_5": 0.27283105022831056, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8843316154214099, | |
| "eval_prec_micro": 0.8894348894348165, | |
| "eval_rec_at_5": 0.9733637747336377, | |
| "eval_rec_at_8": 0.9779299847792998, | |
| "eval_rec_macro": 0.8816174783898175, | |
| "eval_rec_micro": 0.9004975124377362, | |
| "eval_runtime": 3.278, | |
| "eval_samples_per_second": 267.232, | |
| "eval_steps_per_second": 33.557, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 0.5692580342292786, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1278, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_acc_macro": 0.810678005842559, | |
| "eval_acc_micro": 0.8233082706766298, | |
| "eval_auc_macro": 0.9875718328324142, | |
| "eval_auc_micro": 0.989483950688566, | |
| "eval_f1_at_5": 0.4271994046239887, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.8918987653168854, | |
| "eval_f1_micro": 0.90309278350508, | |
| "eval_loss": 0.11230690032243729, | |
| "eval_prec_at_5": 0.27351598173515984, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.8939579787040955, | |
| "eval_prec_micro": 0.8982772764560378, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8924277193620916, | |
| "eval_rec_micro": 0.9079601990048998, | |
| "eval_runtime": 3.2773, | |
| "eval_samples_per_second": 267.291, | |
| "eval_steps_per_second": 33.564, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "grad_norm": 0.4341146647930145, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.1181, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_acc_macro": 0.8114177963739652, | |
| "eval_acc_micro": 0.8276661514682513, | |
| "eval_auc_macro": 0.9882489665360898, | |
| "eval_auc_micro": 0.9896063774543715, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.2915919235644548, | |
| "eval_f1_macro": 0.8910138454396187, | |
| "eval_f1_micro": 0.9057082452430524, | |
| "eval_loss": 0.10568831115961075, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1713755707762557, | |
| "eval_prec_macro": 0.9229102543764114, | |
| "eval_prec_micro": 0.924072476272569, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.9767884322678843, | |
| "eval_rec_macro": 0.8685981119086549, | |
| "eval_rec_micro": 0.8880597014924636, | |
| "eval_runtime": 3.2679, | |
| "eval_samples_per_second": 268.059, | |
| "eval_steps_per_second": 33.66, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 0.3292841613292694, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.1101, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_acc_macro": 0.8196380626953511, | |
| "eval_acc_micro": 0.8308880308879667, | |
| "eval_auc_macro": 0.9889504342095332, | |
| "eval_auc_micro": 0.9907443048387263, | |
| "eval_f1_at_5": 0.4271994046239887, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.8973412777794657, | |
| "eval_f1_micro": 0.9076339097426479, | |
| "eval_loss": 0.0997268408536911, | |
| "eval_prec_at_5": 0.27351598173515984, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9202983062064826, | |
| "eval_prec_micro": 0.923605150214513, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8785932935910168, | |
| "eval_rec_micro": 0.8922056384742212, | |
| "eval_runtime": 3.2426, | |
| "eval_samples_per_second": 270.154, | |
| "eval_steps_per_second": 33.923, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "grad_norm": 0.513087272644043, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.1023, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_acc_macro": 0.8235455386482644, | |
| "eval_acc_micro": 0.8347425057647321, | |
| "eval_auc_macro": 0.9885335456453125, | |
| "eval_auc_micro": 0.9898872283756374, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.2915919235644548, | |
| "eval_f1_macro": 0.8999181306237283, | |
| "eval_f1_micro": 0.9099287808964466, | |
| "eval_loss": 0.09510227292776108, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1713755707762557, | |
| "eval_prec_macro": 0.9168967179225254, | |
| "eval_prec_micro": 0.9195596951735038, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.9767884322678843, | |
| "eval_rec_macro": 0.8861519346844252, | |
| "eval_rec_micro": 0.9004975124377362, | |
| "eval_runtime": 3.3019, | |
| "eval_samples_per_second": 265.298, | |
| "eval_steps_per_second": 33.314, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "grad_norm": 0.44329148530960083, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0966, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_acc_macro": 0.8282861708589082, | |
| "eval_acc_micro": 0.8414164742108667, | |
| "eval_auc_macro": 0.9900771631031816, | |
| "eval_auc_micro": 0.9908838392748628, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.903024454767273, | |
| "eval_f1_micro": 0.9138795986621309, | |
| "eval_loss": 0.09051786363124847, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9150389365619009, | |
| "eval_prec_micro": 0.9215851602022831, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8934998678358861, | |
| "eval_rec_micro": 0.9063018242121967, | |
| "eval_runtime": 3.2033, | |
| "eval_samples_per_second": 273.471, | |
| "eval_steps_per_second": 34.34, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "grad_norm": 0.5752924084663391, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0907, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_acc_macro": 0.8284485888604599, | |
| "eval_acc_micro": 0.8406466512701432, | |
| "eval_auc_macro": 0.9899844969634316, | |
| "eval_auc_micro": 0.9914398385606178, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.9033529953273461, | |
| "eval_f1_micro": 0.9134253450438383, | |
| "eval_loss": 0.08615443110466003, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9148533164182675, | |
| "eval_prec_micro": 0.9215189873416944, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8936470636992689, | |
| "eval_rec_micro": 0.9054726368158453, | |
| "eval_runtime": 3.2471, | |
| "eval_samples_per_second": 269.782, | |
| "eval_steps_per_second": 33.877, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "grad_norm": 0.3273939788341522, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0856, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_acc_macro": 0.8365219798472802, | |
| "eval_acc_micro": 0.8460943542149384, | |
| "eval_auc_macro": 0.990006520319756, | |
| "eval_auc_micro": 0.991490537854463, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.908175633993403, | |
| "eval_f1_micro": 0.9166317553413559, | |
| "eval_loss": 0.08281172811985016, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9221522942691044, | |
| "eval_prec_micro": 0.9263336155799384, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8964813104500736, | |
| "eval_rec_micro": 0.9071310116085483, | |
| "eval_runtime": 3.2394, | |
| "eval_samples_per_second": 270.419, | |
| "eval_steps_per_second": 33.957, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "grad_norm": 0.4600836932659149, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.081, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_acc_macro": 0.8341810383521641, | |
| "eval_acc_micro": 0.8469778117826436, | |
| "eval_auc_macro": 0.9899728237387516, | |
| "eval_auc_micro": 0.9915228820436822, | |
| "eval_f1_at_5": 0.4278659031068435, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9068497360775286, | |
| "eval_f1_micro": 0.9171499585749033, | |
| "eval_loss": 0.07994194328784943, | |
| "eval_prec_at_5": 0.27397260273972607, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9087038321630772, | |
| "eval_prec_micro": 0.9163907284767453, | |
| "eval_rec_at_5": 0.9762176560121765, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9059843761789805, | |
| "eval_rec_micro": 0.9179104477611179, | |
| "eval_runtime": 3.1998, | |
| "eval_samples_per_second": 273.764, | |
| "eval_steps_per_second": 34.377, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "grad_norm": 0.5419149398803711, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0773, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_acc_macro": 0.8302896897341306, | |
| "eval_acc_micro": 0.8448947778643144, | |
| "eval_auc_macro": 0.9902784944925817, | |
| "eval_auc_micro": 0.9916112716737253, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.9041333779773183, | |
| "eval_f1_micro": 0.9159273341782074, | |
| "eval_loss": 0.07834368944168091, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9276589497355273, | |
| "eval_prec_micro": 0.9336778639103416, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.8839118707511296, | |
| "eval_rec_micro": 0.8988391376450332, | |
| "eval_runtime": 3.1453, | |
| "eval_samples_per_second": 278.513, | |
| "eval_steps_per_second": 34.973, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "grad_norm": 0.3567904829978943, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0729, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_acc_macro": 0.8348824235565738, | |
| "eval_acc_micro": 0.8433919022153672, | |
| "eval_auc_macro": 0.9903521725932115, | |
| "eval_auc_micro": 0.9919912936214408, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9076004004647569, | |
| "eval_f1_micro": 0.915043514297479, | |
| "eval_loss": 0.07567035406827927, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9101730173612985, | |
| "eval_prec_micro": 0.9146644573321528, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9057605150279991, | |
| "eval_rec_micro": 0.9154228855720634, | |
| "eval_runtime": 3.1937, | |
| "eval_samples_per_second": 274.293, | |
| "eval_steps_per_second": 34.443, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "grad_norm": 0.4837111234664917, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0698, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_acc_macro": 0.8374765617625478, | |
| "eval_acc_micro": 0.8441947565542438, | |
| "eval_auc_macro": 0.9904772253291115, | |
| "eval_auc_micro": 0.9921461536789155, | |
| "eval_f1_at_5": 0.4278659031068434, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9092344767866304, | |
| "eval_f1_micro": 0.9155158407797794, | |
| "eval_loss": 0.07319281995296478, | |
| "eval_prec_at_5": 0.273972602739726, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8914748226041258, | |
| "eval_prec_micro": 0.8972929936305017, | |
| "eval_rec_at_5": 0.9762176560121765, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9284071354304921, | |
| "eval_rec_micro": 0.9344941956881481, | |
| "eval_runtime": 3.1304, | |
| "eval_samples_per_second": 279.84, | |
| "eval_steps_per_second": 35.14, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "grad_norm": 0.39113664627075195, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0666, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_acc_macro": 0.8381100355802934, | |
| "eval_acc_micro": 0.844077961019427, | |
| "eval_auc_macro": 0.9903966871163578, | |
| "eval_auc_micro": 0.9918843706818725, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9097231670248208, | |
| "eval_f1_micro": 0.9154471544714704, | |
| "eval_loss": 0.07271166890859604, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8946829660987885, | |
| "eval_prec_micro": 0.8979266347686684, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9261163381279823, | |
| "eval_rec_micro": 0.9336650082917965, | |
| "eval_runtime": 3.2473, | |
| "eval_samples_per_second": 269.762, | |
| "eval_steps_per_second": 33.874, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "grad_norm": 0.623213529586792, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0635, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_acc_macro": 0.8378381654750737, | |
| "eval_acc_micro": 0.8448795180722255, | |
| "eval_auc_macro": 0.9903437445438174, | |
| "eval_auc_micro": 0.991977126331948, | |
| "eval_f1_at_5": 0.4275326584009282, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9095042466910366, | |
| "eval_f1_micro": 0.915918367346864, | |
| "eval_loss": 0.07025843113660812, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8980030893359562, | |
| "eval_prec_micro": 0.9019292604500883, | |
| "eval_rec_at_5": 0.9756468797564688, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.921910312969234, | |
| "eval_rec_micro": 0.9303482587063905, | |
| "eval_runtime": 3.2584, | |
| "eval_samples_per_second": 268.843, | |
| "eval_steps_per_second": 33.759, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "grad_norm": 0.5504964590072632, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0608, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_acc_macro": 0.8420968191195207, | |
| "eval_acc_micro": 0.8508705526115934, | |
| "eval_auc_macro": 0.9903203584733035, | |
| "eval_auc_micro": 0.9918962213076746, | |
| "eval_f1_at_5": 0.4271994046239887, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.911966406429931, | |
| "eval_f1_micro": 0.9194274028629105, | |
| "eval_loss": 0.06913089752197266, | |
| "eval_prec_at_5": 0.27351598173515984, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8987539781949079, | |
| "eval_prec_micro": 0.9071832122678848, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9260958603602073, | |
| "eval_rec_micro": 0.9320066334990935, | |
| "eval_runtime": 3.2691, | |
| "eval_samples_per_second": 267.962, | |
| "eval_steps_per_second": 33.648, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "grad_norm": 0.4138895571231842, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.059, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_acc_macro": 0.8343605198418258, | |
| "eval_acc_micro": 0.8466819221967318, | |
| "eval_auc_macro": 0.9905189753061885, | |
| "eval_auc_micro": 0.9920985729708075, | |
| "eval_f1_at_5": 0.4278659031068434, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9071167088535305, | |
| "eval_f1_micro": 0.9169764560098375, | |
| "eval_loss": 0.068180613219738, | |
| "eval_prec_at_5": 0.273972602739726, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.905623991446007, | |
| "eval_prec_micro": 0.913580246913505, | |
| "eval_rec_at_5": 0.9762176560121765, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9092956538553008, | |
| "eval_rec_micro": 0.9203980099501724, | |
| "eval_runtime": 3.2333, | |
| "eval_samples_per_second": 270.935, | |
| "eval_steps_per_second": 34.021, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "grad_norm": 0.4782184958457947, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0558, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_acc_macro": 0.8388794744893889, | |
| "eval_acc_micro": 0.8464419475654796, | |
| "eval_auc_macro": 0.9907869944842939, | |
| "eval_auc_micro": 0.9923112605180986, | |
| "eval_f1_at_5": 0.4281991387591175, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9097311444189383, | |
| "eval_f1_micro": 0.9168356997970858, | |
| "eval_loss": 0.06692954152822495, | |
| "eval_prec_at_5": 0.2742009132420091, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.8914388177432641, | |
| "eval_prec_micro": 0.8975377283557666, | |
| "eval_rec_at_5": 0.9767884322678843, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9295486988792282, | |
| "eval_rec_micro": 0.9369817578772025, | |
| "eval_runtime": 3.1471, | |
| "eval_samples_per_second": 278.355, | |
| "eval_steps_per_second": 34.953, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "grad_norm": 0.4504872262477875, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0538, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_acc_macro": 0.838778704972707, | |
| "eval_acc_micro": 0.849157733537454, | |
| "eval_auc_macro": 0.9902459134171048, | |
| "eval_auc_micro": 0.9921254819105988, | |
| "eval_f1_at_5": 0.4278476261962846, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.909828962756635, | |
| "eval_f1_micro": 0.9184265010351206, | |
| "eval_loss": 0.06644343584775925, | |
| "eval_prec_at_5": 0.27397260273972607, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9086581525308847, | |
| "eval_prec_micro": 0.9172870140611317, | |
| "eval_rec_at_5": 0.976027397260274, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9113050956999702, | |
| "eval_rec_micro": 0.9195688225538209, | |
| "eval_runtime": 3.1423, | |
| "eval_samples_per_second": 278.777, | |
| "eval_steps_per_second": 35.006, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "grad_norm": 0.4144255220890045, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0514, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_acc_macro": 0.8414238560509153, | |
| "eval_acc_micro": 0.8515981735159169, | |
| "eval_auc_macro": 0.9904232681518291, | |
| "eval_auc_micro": 0.9915424845826032, | |
| "eval_f1_at_5": 0.4278659031068435, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.9115867876146071, | |
| "eval_f1_micro": 0.9198520345252017, | |
| "eval_loss": 0.06602790951728821, | |
| "eval_prec_at_5": 0.27397260273972607, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9028158103948029, | |
| "eval_prec_micro": 0.9119804400977252, | |
| "eval_rec_at_5": 0.9762176560121765, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.9208453689121621, | |
| "eval_rec_micro": 0.9278606965173359, | |
| "eval_runtime": 3.2592, | |
| "eval_samples_per_second": 268.778, | |
| "eval_steps_per_second": 33.751, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "grad_norm": 0.5076408386230469, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0486, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_acc_macro": 0.8417559403467925, | |
| "eval_acc_micro": 0.8513097072418451, | |
| "eval_auc_macro": 0.9902567259413712, | |
| "eval_auc_micro": 0.9916356857449267, | |
| "eval_f1_at_5": 0.42747783226694747, | |
| "eval_f1_at_8": 0.2918154230125642, | |
| "eval_f1_macro": 0.9114967651862192, | |
| "eval_f1_micro": 0.9196837286724161, | |
| "eval_loss": 0.06532150506973267, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9135306848535077, | |
| "eval_prec_micro": 0.9231411862990039, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.9771689497716894, | |
| "eval_rec_macro": 0.9101782005755276, | |
| "eval_rec_micro": 0.9162520729684148, | |
| "eval_runtime": 3.154, | |
| "eval_samples_per_second": 277.742, | |
| "eval_steps_per_second": 34.876, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "grad_norm": 0.39073243737220764, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0473, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_acc_macro": 0.8365248835222155, | |
| "eval_acc_micro": 0.8455034588777213, | |
| "eval_auc_macro": 0.9902217568533902, | |
| "eval_auc_micro": 0.991443937273301, | |
| "eval_f1_at_5": 0.42747783226694747, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9082513467521097, | |
| "eval_f1_micro": 0.9162848812993822, | |
| "eval_loss": 0.06487824022769928, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9133935694619835, | |
| "eval_prec_micro": 0.9205020920501321, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.904152383416957, | |
| "eval_rec_micro": 0.9121061359866574, | |
| "eval_runtime": 3.241, | |
| "eval_samples_per_second": 270.288, | |
| "eval_steps_per_second": 33.94, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "grad_norm": 0.4108600318431854, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0452, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_acc_macro": 0.8342283813769025, | |
| "eval_acc_micro": 0.8460365853657892, | |
| "eval_auc_macro": 0.9900494801767596, | |
| "eval_auc_micro": 0.9917285304974518, | |
| "eval_f1_at_5": 0.4278476261962846, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9069117318096943, | |
| "eval_f1_micro": 0.9165978530139621, | |
| "eval_loss": 0.06424969434738159, | |
| "eval_prec_at_5": 0.27397260273972607, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9043399542883956, | |
| "eval_prec_micro": 0.912828947368346, | |
| "eval_rec_at_5": 0.976027397260274, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9102122008404995, | |
| "eval_rec_micro": 0.9203980099501724, | |
| "eval_runtime": 3.1728, | |
| "eval_samples_per_second": 276.1, | |
| "eval_steps_per_second": 34.67, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "grad_norm": 0.6398113369941711, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0441, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_acc_macro": 0.8402010441768392, | |
| "eval_acc_micro": 0.8504176157934054, | |
| "eval_auc_macro": 0.9899877083743055, | |
| "eval_auc_micro": 0.9911509684188842, | |
| "eval_f1_at_5": 0.42747783226694747, | |
| "eval_f1_at_8": 0.29182390543479947, | |
| "eval_f1_macro": 0.9105990068832329, | |
| "eval_f1_micro": 0.9191629052112499, | |
| "eval_loss": 0.06337956339120865, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.17151826484018265, | |
| "eval_prec_macro": 0.9012107317710064, | |
| "eval_prec_micro": 0.9098294069861161, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.977359208523592, | |
| "eval_rec_macro": 0.921097280138882, | |
| "eval_rec_micro": 0.9286898839136875, | |
| "eval_runtime": 3.2382, | |
| "eval_samples_per_second": 270.523, | |
| "eval_steps_per_second": 33.97, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "grad_norm": 0.5846818089485168, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0423, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_acc_macro": 0.8392593134167763, | |
| "eval_acc_micro": 0.8493788819875117, | |
| "eval_auc_macro": 0.9896914288641514, | |
| "eval_auc_micro": 0.9914127959671519, | |
| "eval_f1_at_5": 0.42747783226694747, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9100464733016533, | |
| "eval_f1_micro": 0.9185558354323325, | |
| "eval_loss": 0.06289488822221756, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9212797353863423, | |
| "eval_prec_micro": 0.9302721088434582, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9000358120655063, | |
| "eval_rec_micro": 0.9071310116085483, | |
| "eval_runtime": 3.1859, | |
| "eval_samples_per_second": 274.961, | |
| "eval_steps_per_second": 34.527, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "grad_norm": 0.47924453020095825, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0408, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_acc_macro": 0.839620505353177, | |
| "eval_acc_micro": 0.8521072796934213, | |
| "eval_auc_macro": 0.9895009786225338, | |
| "eval_auc_micro": 0.990984346838057, | |
| "eval_f1_at_5": 0.42747783226694747, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9100739843959514, | |
| "eval_f1_micro": 0.9201489449730311, | |
| "eval_loss": 0.06270425021648407, | |
| "eval_prec_at_5": 0.27374429223744295, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9077928816879504, | |
| "eval_prec_micro": 0.9182493806770505, | |
| "eval_rec_at_5": 0.975076103500761, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9134800253293897, | |
| "eval_rec_micro": 0.9220563847428754, | |
| "eval_runtime": 3.2485, | |
| "eval_samples_per_second": 269.663, | |
| "eval_steps_per_second": 33.862, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "grad_norm": 0.45999017357826233, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0389, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_acc_macro": 0.838151465593231, | |
| "eval_acc_micro": 0.8484848484847842, | |
| "eval_auc_macro": 0.9893868618238587, | |
| "eval_auc_micro": 0.9910452038111611, | |
| "eval_f1_at_5": 0.42714459980714947, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9091468843830479, | |
| "eval_f1_micro": 0.9180327868851706, | |
| "eval_loss": 0.0631389170885086, | |
| "eval_prec_at_5": 0.27351598173515984, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9002505510433269, | |
| "eval_prec_micro": 0.9076175040517903, | |
| "eval_rec_at_5": 0.9745053272450532, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9191609975742806, | |
| "eval_rec_micro": 0.9286898839136875, | |
| "eval_runtime": 3.1965, | |
| "eval_samples_per_second": 274.047, | |
| "eval_steps_per_second": 34.412, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "grad_norm": 0.5668067336082458, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.038, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_acc_macro": 0.8442263088885594, | |
| "eval_acc_micro": 0.8538985616956204, | |
| "eval_auc_macro": 0.989457040438969, | |
| "eval_auc_micro": 0.9912924631089128, | |
| "eval_f1_at_5": 0.42812596936069724, | |
| "eval_f1_at_8": 0.2920473967500829, | |
| "eval_f1_macro": 0.9129580069246931, | |
| "eval_f1_micro": 0.9211923233972297, | |
| "eval_loss": 0.062041375786066055, | |
| "eval_prec_at_5": 0.2742009132420091, | |
| "eval_prec_at_8": 0.1716609589041096, | |
| "eval_prec_macro": 0.9013584195385889, | |
| "eval_prec_micro": 0.907481898632268, | |
| "eval_rec_at_5": 0.976027397260274, | |
| "eval_rec_at_8": 0.9777397260273972, | |
| "eval_rec_macro": 0.9261603815096769, | |
| "eval_rec_micro": 0.9353233830844995, | |
| "eval_runtime": 3.186, | |
| "eval_samples_per_second": 274.957, | |
| "eval_steps_per_second": 34.527, | |
| "step": 912 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 960, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 40, | |
| "save_steps": 500, | |
| "total_flos": 1.533864671059968e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |