| { | |
| "best_metric": 0.6838614548013121, | |
| "best_model_checkpoint": "../saved_model/cino-small-v2_tncc-title_v3/checkpoint-7424", | |
| "epoch": 40.0, | |
| "eval_steps": 500, | |
| "global_step": 9280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.49838187702265374, | |
| "eval_loss": 1.6120452880859375, | |
| "eval_macro-f1": 0.3080200150906938, | |
| "eval_macro-precision": 0.4470989896292958, | |
| "eval_macro-recall": 0.33407156196923765, | |
| "eval_runtime": 3.5999, | |
| "eval_samples_per_second": 257.505, | |
| "eval_steps_per_second": 8.056, | |
| "eval_weighted-f1": 0.423917385683886, | |
| "eval_weighted-precision": 0.5166350613267663, | |
| "eval_weighted-recall": 0.49838187702265374, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6472491909385113, | |
| "eval_loss": 1.1052128076553345, | |
| "eval_macro-f1": 0.5457350988248497, | |
| "eval_macro-precision": 0.6404982007168077, | |
| "eval_macro-recall": 0.559508539878825, | |
| "eval_runtime": 3.6454, | |
| "eval_samples_per_second": 254.294, | |
| "eval_steps_per_second": 7.955, | |
| "eval_weighted-f1": 0.6324317601702024, | |
| "eval_weighted-precision": 0.6658002868686018, | |
| "eval_weighted-recall": 0.6472491909385113, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.6939655172413796e-05, | |
| "loss": 1.7073, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6709816612729234, | |
| "eval_loss": 1.0001471042633057, | |
| "eval_macro-f1": 0.6065301675158895, | |
| "eval_macro-precision": 0.654418913280353, | |
| "eval_macro-recall": 0.6099139180853289, | |
| "eval_runtime": 3.6176, | |
| "eval_samples_per_second": 256.245, | |
| "eval_steps_per_second": 8.016, | |
| "eval_weighted-f1": 0.6605809984207812, | |
| "eval_weighted-precision": 0.6778557170715117, | |
| "eval_weighted-recall": 0.6709816612729234, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.686084142394822, | |
| "eval_loss": 0.9795448780059814, | |
| "eval_macro-f1": 0.645770505361501, | |
| "eval_macro-precision": 0.6624511940296568, | |
| "eval_macro-recall": 0.6454397105147797, | |
| "eval_runtime": 3.6312, | |
| "eval_samples_per_second": 255.29, | |
| "eval_steps_per_second": 7.986, | |
| "eval_weighted-f1": 0.680474769885252, | |
| "eval_weighted-precision": 0.6882611664755186, | |
| "eval_weighted-recall": 0.686084142394822, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 4.9568965517241384e-05, | |
| "loss": 0.8684, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.6947141316073355, | |
| "eval_loss": 0.9905449748039246, | |
| "eval_macro-f1": 0.6674507510628628, | |
| "eval_macro-precision": 0.6625099174020581, | |
| "eval_macro-recall": 0.6829958649269056, | |
| "eval_runtime": 3.6247, | |
| "eval_samples_per_second": 255.745, | |
| "eval_steps_per_second": 8.001, | |
| "eval_weighted-f1": 0.6952993996868073, | |
| "eval_weighted-precision": 0.7036333325841285, | |
| "eval_weighted-recall": 0.6947141316073355, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7044228694714132, | |
| "eval_loss": 1.0819956064224243, | |
| "eval_macro-f1": 0.6835424949893771, | |
| "eval_macro-precision": 0.6831322585201048, | |
| "eval_macro-recall": 0.6897387284921295, | |
| "eval_runtime": 3.6099, | |
| "eval_samples_per_second": 256.793, | |
| "eval_steps_per_second": 8.033, | |
| "eval_weighted-f1": 0.7063455583024486, | |
| "eval_weighted-precision": 0.7133818379200453, | |
| "eval_weighted-recall": 0.7044228694714132, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 4.6575670498084294e-05, | |
| "loss": 0.4904, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.6850053937432579, | |
| "eval_loss": 1.2797751426696777, | |
| "eval_macro-f1": 0.6529772810537373, | |
| "eval_macro-precision": 0.6632776375522283, | |
| "eval_macro-recall": 0.6621596382156031, | |
| "eval_runtime": 3.621, | |
| "eval_samples_per_second": 256.009, | |
| "eval_steps_per_second": 8.009, | |
| "eval_weighted-f1": 0.6864725766536225, | |
| "eval_weighted-precision": 0.7003185776965812, | |
| "eval_weighted-recall": 0.6850053937432579, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.6796116504854369, | |
| "eval_loss": 1.3903510570526123, | |
| "eval_macro-f1": 0.6510499118775824, | |
| "eval_macro-precision": 0.6651601590963526, | |
| "eval_macro-recall": 0.6479225440423547, | |
| "eval_runtime": 3.6152, | |
| "eval_samples_per_second": 256.415, | |
| "eval_steps_per_second": 8.022, | |
| "eval_weighted-f1": 0.6783723538205079, | |
| "eval_weighted-precision": 0.6881154476659207, | |
| "eval_weighted-recall": 0.6796116504854369, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 4.3582375478927204e-05, | |
| "loss": 0.2376, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.674217907227616, | |
| "eval_loss": 1.4255810976028442, | |
| "eval_macro-f1": 0.6424901861691795, | |
| "eval_macro-precision": 0.6705435579848175, | |
| "eval_macro-recall": 0.6378243354162482, | |
| "eval_runtime": 3.6179, | |
| "eval_samples_per_second": 256.225, | |
| "eval_steps_per_second": 8.016, | |
| "eval_weighted-f1": 0.6736702968697759, | |
| "eval_weighted-precision": 0.6879912323928888, | |
| "eval_weighted-recall": 0.674217907227616, | |
| "step": 2088 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.6882416396979504, | |
| "eval_loss": 1.5364353656768799, | |
| "eval_macro-f1": 0.6654116450501577, | |
| "eval_macro-precision": 0.6574262835666708, | |
| "eval_macro-recall": 0.6801476025508416, | |
| "eval_runtime": 3.6024, | |
| "eval_samples_per_second": 257.33, | |
| "eval_steps_per_second": 8.05, | |
| "eval_weighted-f1": 0.6895751725299344, | |
| "eval_weighted-precision": 0.6945151299816443, | |
| "eval_weighted-recall": 0.6882416396979504, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 4.058908045977012e-05, | |
| "loss": 0.1495, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.6828478964401294, | |
| "eval_loss": 1.6653459072113037, | |
| "eval_macro-f1": 0.6512014541822767, | |
| "eval_macro-precision": 0.6627229410666052, | |
| "eval_macro-recall": 0.6485507026145255, | |
| "eval_runtime": 3.5928, | |
| "eval_samples_per_second": 258.014, | |
| "eval_steps_per_second": 8.072, | |
| "eval_weighted-f1": 0.6800769417795847, | |
| "eval_weighted-precision": 0.6882902408612807, | |
| "eval_weighted-recall": 0.6828478964401294, | |
| "step": 2552 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.6839266450916937, | |
| "eval_loss": 1.6987706422805786, | |
| "eval_macro-f1": 0.6479532709170456, | |
| "eval_macro-precision": 0.6580973534101414, | |
| "eval_macro-recall": 0.6474043097812547, | |
| "eval_runtime": 3.6004, | |
| "eval_samples_per_second": 257.471, | |
| "eval_steps_per_second": 8.055, | |
| "eval_weighted-f1": 0.6841454877516878, | |
| "eval_weighted-precision": 0.6902044355228595, | |
| "eval_weighted-recall": 0.6839266450916937, | |
| "step": 2784 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "learning_rate": 3.759578544061303e-05, | |
| "loss": 0.1058, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.6817691477885652, | |
| "eval_loss": 1.747767448425293, | |
| "eval_macro-f1": 0.6556363951964088, | |
| "eval_macro-precision": 0.6700908993068831, | |
| "eval_macro-recall": 0.6561971975750561, | |
| "eval_runtime": 3.6016, | |
| "eval_samples_per_second": 257.383, | |
| "eval_steps_per_second": 8.052, | |
| "eval_weighted-f1": 0.6814628702483411, | |
| "eval_weighted-precision": 0.6867012225910729, | |
| "eval_weighted-recall": 0.6817691477885652, | |
| "step": 3016 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.674217907227616, | |
| "eval_loss": 1.918135166168213, | |
| "eval_macro-f1": 0.6397376565535934, | |
| "eval_macro-precision": 0.6549903289689117, | |
| "eval_macro-recall": 0.6473498155004918, | |
| "eval_runtime": 3.5879, | |
| "eval_samples_per_second": 258.368, | |
| "eval_steps_per_second": 8.083, | |
| "eval_weighted-f1": 0.6765985742439555, | |
| "eval_weighted-precision": 0.6901672277432537, | |
| "eval_weighted-recall": 0.674217907227616, | |
| "step": 3248 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.6957928802588996, | |
| "eval_loss": 1.9204330444335938, | |
| "eval_macro-f1": 0.6658200325646808, | |
| "eval_macro-precision": 0.6663228409772431, | |
| "eval_macro-recall": 0.6692964824595501, | |
| "eval_runtime": 3.5954, | |
| "eval_samples_per_second": 257.831, | |
| "eval_steps_per_second": 8.066, | |
| "eval_weighted-f1": 0.6957701345209284, | |
| "eval_weighted-precision": 0.69903863020904, | |
| "eval_weighted-recall": 0.6957928802588996, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 15.09, | |
| "learning_rate": 3.460249042145594e-05, | |
| "loss": 0.0775, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.6752966558791802, | |
| "eval_loss": 1.9924219846725464, | |
| "eval_macro-f1": 0.6392465953749347, | |
| "eval_macro-precision": 0.6561467817801693, | |
| "eval_macro-recall": 0.6402901277480958, | |
| "eval_runtime": 3.6181, | |
| "eval_samples_per_second": 256.214, | |
| "eval_steps_per_second": 8.015, | |
| "eval_weighted-f1": 0.673631619187917, | |
| "eval_weighted-precision": 0.6858049526730299, | |
| "eval_weighted-recall": 0.6752966558791802, | |
| "step": 3712 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.686084142394822, | |
| "eval_loss": 2.0478932857513428, | |
| "eval_macro-f1": 0.6463214194781717, | |
| "eval_macro-precision": 0.6570967483791249, | |
| "eval_macro-recall": 0.6503589081998865, | |
| "eval_runtime": 3.5962, | |
| "eval_samples_per_second": 257.771, | |
| "eval_steps_per_second": 8.064, | |
| "eval_weighted-f1": 0.6852865958011676, | |
| "eval_weighted-precision": 0.6939600166349821, | |
| "eval_weighted-recall": 0.686084142394822, | |
| "step": 3944 | |
| }, | |
| { | |
| "epoch": 17.24, | |
| "learning_rate": 3.160919540229885e-05, | |
| "loss": 0.0687, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.674217907227616, | |
| "eval_loss": 2.0522632598876953, | |
| "eval_macro-f1": 0.6405085398516451, | |
| "eval_macro-precision": 0.6662302886679464, | |
| "eval_macro-recall": 0.6393535717017708, | |
| "eval_runtime": 3.6131, | |
| "eval_samples_per_second": 256.565, | |
| "eval_steps_per_second": 8.026, | |
| "eval_weighted-f1": 0.674060546264426, | |
| "eval_weighted-precision": 0.6894424888472697, | |
| "eval_weighted-recall": 0.674217907227616, | |
| "step": 4176 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.6828478964401294, | |
| "eval_loss": 1.9823521375656128, | |
| "eval_macro-f1": 0.651754402283885, | |
| "eval_macro-precision": 0.6440801367882046, | |
| "eval_macro-recall": 0.6683888090630882, | |
| "eval_runtime": 3.6037, | |
| "eval_samples_per_second": 257.233, | |
| "eval_steps_per_second": 8.047, | |
| "eval_weighted-f1": 0.6821812966643793, | |
| "eval_weighted-precision": 0.68760298706152, | |
| "eval_weighted-recall": 0.6828478964401294, | |
| "step": 4408 | |
| }, | |
| { | |
| "epoch": 19.4, | |
| "learning_rate": 2.8615900383141765e-05, | |
| "loss": 0.0577, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.6914778856526429, | |
| "eval_loss": 2.0154926776885986, | |
| "eval_macro-f1": 0.6659900934162544, | |
| "eval_macro-precision": 0.6712869574383351, | |
| "eval_macro-recall": 0.6703643770947892, | |
| "eval_runtime": 3.6164, | |
| "eval_samples_per_second": 256.332, | |
| "eval_steps_per_second": 8.019, | |
| "eval_weighted-f1": 0.6906611756556518, | |
| "eval_weighted-precision": 0.6985916032921785, | |
| "eval_weighted-recall": 0.6914778856526429, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.6828478964401294, | |
| "eval_loss": 2.05501389503479, | |
| "eval_macro-f1": 0.6486634410090307, | |
| "eval_macro-precision": 0.6780446089713759, | |
| "eval_macro-recall": 0.6376704510538888, | |
| "eval_runtime": 3.5904, | |
| "eval_samples_per_second": 258.186, | |
| "eval_steps_per_second": 8.077, | |
| "eval_weighted-f1": 0.6755183997397308, | |
| "eval_weighted-precision": 0.6900556731555942, | |
| "eval_weighted-recall": 0.6828478964401294, | |
| "step": 4872 | |
| }, | |
| { | |
| "epoch": 21.55, | |
| "learning_rate": 2.5622605363984675e-05, | |
| "loss": 0.0489, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.6796116504854369, | |
| "eval_loss": 2.0277795791625977, | |
| "eval_macro-f1": 0.6505710558420628, | |
| "eval_macro-precision": 0.6513644147664868, | |
| "eval_macro-recall": 0.6562986108294033, | |
| "eval_runtime": 3.5971, | |
| "eval_samples_per_second": 257.71, | |
| "eval_steps_per_second": 8.062, | |
| "eval_weighted-f1": 0.6787275977253558, | |
| "eval_weighted-precision": 0.6831300629390828, | |
| "eval_weighted-recall": 0.6796116504854369, | |
| "step": 5104 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.6828478964401294, | |
| "eval_loss": 2.033602237701416, | |
| "eval_macro-f1": 0.6518094813918546, | |
| "eval_macro-precision": 0.6546940340503233, | |
| "eval_macro-recall": 0.6544133192355402, | |
| "eval_runtime": 3.6042, | |
| "eval_samples_per_second": 257.199, | |
| "eval_steps_per_second": 8.046, | |
| "eval_weighted-f1": 0.6812281754499014, | |
| "eval_weighted-precision": 0.6846668946206854, | |
| "eval_weighted-recall": 0.6828478964401294, | |
| "step": 5336 | |
| }, | |
| { | |
| "epoch": 23.71, | |
| "learning_rate": 2.2629310344827588e-05, | |
| "loss": 0.0431, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.697950377562028, | |
| "eval_loss": 2.0058302879333496, | |
| "eval_macro-f1": 0.672432519382824, | |
| "eval_macro-precision": 0.7031870372780317, | |
| "eval_macro-recall": 0.6570710914948313, | |
| "eval_runtime": 3.6053, | |
| "eval_samples_per_second": 257.123, | |
| "eval_steps_per_second": 8.044, | |
| "eval_weighted-f1": 0.6956222916424519, | |
| "eval_weighted-precision": 0.7050444644388131, | |
| "eval_weighted-recall": 0.697950377562028, | |
| "step": 5568 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.6903991370010788, | |
| "eval_loss": 2.0136592388153076, | |
| "eval_macro-f1": 0.6593408320415061, | |
| "eval_macro-precision": 0.6670076069784594, | |
| "eval_macro-recall": 0.6569151453566795, | |
| "eval_runtime": 3.6053, | |
| "eval_samples_per_second": 257.124, | |
| "eval_steps_per_second": 8.044, | |
| "eval_weighted-f1": 0.6888694520013063, | |
| "eval_weighted-precision": 0.6914605561085521, | |
| "eval_weighted-recall": 0.6903991370010788, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 25.86, | |
| "learning_rate": 1.9636015325670498e-05, | |
| "loss": 0.0372, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.6763754045307443, | |
| "eval_loss": 2.21449875831604, | |
| "eval_macro-f1": 0.6400140077032914, | |
| "eval_macro-precision": 0.649984379772116, | |
| "eval_macro-recall": 0.6499316041741732, | |
| "eval_runtime": 3.6097, | |
| "eval_samples_per_second": 256.806, | |
| "eval_steps_per_second": 8.034, | |
| "eval_weighted-f1": 0.6791315210465927, | |
| "eval_weighted-precision": 0.6961390172701661, | |
| "eval_weighted-recall": 0.6763754045307443, | |
| "step": 6032 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.6882416396979504, | |
| "eval_loss": 2.1051621437072754, | |
| "eval_macro-f1": 0.6563104764431434, | |
| "eval_macro-precision": 0.6584303947936838, | |
| "eval_macro-recall": 0.6627288799113591, | |
| "eval_runtime": 3.5999, | |
| "eval_samples_per_second": 257.511, | |
| "eval_steps_per_second": 8.056, | |
| "eval_weighted-f1": 0.6864978936047862, | |
| "eval_weighted-precision": 0.6922661220105522, | |
| "eval_weighted-recall": 0.6882416396979504, | |
| "step": 6264 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.6947141316073355, | |
| "eval_loss": 2.096506357192993, | |
| "eval_macro-f1": 0.6701111894500894, | |
| "eval_macro-precision": 0.6973353037360721, | |
| "eval_macro-recall": 0.6569800023874061, | |
| "eval_runtime": 3.6151, | |
| "eval_samples_per_second": 256.427, | |
| "eval_steps_per_second": 8.022, | |
| "eval_weighted-f1": 0.6888264314204101, | |
| "eval_weighted-precision": 0.6996849509212928, | |
| "eval_weighted-recall": 0.6947141316073355, | |
| "step": 6496 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.664272030651341e-05, | |
| "loss": 0.0357, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.6947141316073355, | |
| "eval_loss": 2.094637632369995, | |
| "eval_macro-f1": 0.6685294866541375, | |
| "eval_macro-precision": 0.6813823548362715, | |
| "eval_macro-recall": 0.6641734635067332, | |
| "eval_runtime": 3.6038, | |
| "eval_samples_per_second": 257.226, | |
| "eval_steps_per_second": 8.047, | |
| "eval_weighted-f1": 0.6943390563150669, | |
| "eval_weighted-precision": 0.7001493288391278, | |
| "eval_weighted-recall": 0.6947141316073355, | |
| "step": 6728 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.6893203883495146, | |
| "eval_loss": 2.0883684158325195, | |
| "eval_macro-f1": 0.6602023917903279, | |
| "eval_macro-precision": 0.6628650635074098, | |
| "eval_macro-recall": 0.6659307948401801, | |
| "eval_runtime": 3.6134, | |
| "eval_samples_per_second": 256.543, | |
| "eval_steps_per_second": 8.026, | |
| "eval_weighted-f1": 0.6902134268034101, | |
| "eval_weighted-precision": 0.6951233806637598, | |
| "eval_weighted-recall": 0.6893203883495146, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 30.17, | |
| "learning_rate": 1.3649425287356324e-05, | |
| "loss": 0.0286, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.703344120819849, | |
| "eval_loss": 2.126500368118286, | |
| "eval_macro-f1": 0.6756885174044472, | |
| "eval_macro-precision": 0.6887882836127188, | |
| "eval_macro-recall": 0.6701991060577401, | |
| "eval_runtime": 3.6102, | |
| "eval_samples_per_second": 256.771, | |
| "eval_steps_per_second": 8.033, | |
| "eval_weighted-f1": 0.7025418148884912, | |
| "eval_weighted-precision": 0.7070891495572698, | |
| "eval_weighted-recall": 0.703344120819849, | |
| "step": 7192 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.7022653721682848, | |
| "eval_loss": 2.110332489013672, | |
| "eval_macro-f1": 0.6838614548013121, | |
| "eval_macro-precision": 0.6918280242510296, | |
| "eval_macro-recall": 0.6819433903345183, | |
| "eval_runtime": 3.6092, | |
| "eval_samples_per_second": 256.84, | |
| "eval_steps_per_second": 8.035, | |
| "eval_weighted-f1": 0.7015731601193966, | |
| "eval_weighted-precision": 0.7068600524270214, | |
| "eval_weighted-recall": 0.7022653721682848, | |
| "step": 7424 | |
| }, | |
| { | |
| "epoch": 32.33, | |
| "learning_rate": 1.0656130268199234e-05, | |
| "loss": 0.0248, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.6925566343042071, | |
| "eval_loss": 2.076742649078369, | |
| "eval_macro-f1": 0.6677185111141611, | |
| "eval_macro-precision": 0.6682997232361648, | |
| "eval_macro-recall": 0.6726620472560384, | |
| "eval_runtime": 3.6011, | |
| "eval_samples_per_second": 257.423, | |
| "eval_steps_per_second": 8.053, | |
| "eval_weighted-f1": 0.693162535561584, | |
| "eval_weighted-precision": 0.6976019284855784, | |
| "eval_weighted-recall": 0.6925566343042071, | |
| "step": 7656 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.686084142394822, | |
| "eval_loss": 2.1223480701446533, | |
| "eval_macro-f1": 0.661874148838316, | |
| "eval_macro-precision": 0.6610753547813631, | |
| "eval_macro-recall": 0.6713446322584083, | |
| "eval_runtime": 3.6215, | |
| "eval_samples_per_second": 255.972, | |
| "eval_steps_per_second": 8.008, | |
| "eval_weighted-f1": 0.6841907777006906, | |
| "eval_weighted-precision": 0.6894138997369361, | |
| "eval_weighted-recall": 0.686084142394822, | |
| "step": 7888 | |
| }, | |
| { | |
| "epoch": 34.48, | |
| "learning_rate": 7.662835249042145e-06, | |
| "loss": 0.0235, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.6903991370010788, | |
| "eval_loss": 2.133087635040283, | |
| "eval_macro-f1": 0.662335562798904, | |
| "eval_macro-precision": 0.6736920728738612, | |
| "eval_macro-recall": 0.6597960660569285, | |
| "eval_runtime": 3.6194, | |
| "eval_samples_per_second": 256.117, | |
| "eval_steps_per_second": 8.012, | |
| "eval_weighted-f1": 0.6874362655396534, | |
| "eval_weighted-precision": 0.6919778724812079, | |
| "eval_weighted-recall": 0.6903991370010788, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.697950377562028, | |
| "eval_loss": 2.122838020324707, | |
| "eval_macro-f1": 0.670156602236844, | |
| "eval_macro-precision": 0.6690594327579752, | |
| "eval_macro-recall": 0.6751260562968397, | |
| "eval_runtime": 3.6045, | |
| "eval_samples_per_second": 257.182, | |
| "eval_steps_per_second": 8.046, | |
| "eval_weighted-f1": 0.6980890949772001, | |
| "eval_weighted-precision": 0.7007470226449659, | |
| "eval_weighted-recall": 0.697950377562028, | |
| "step": 8352 | |
| }, | |
| { | |
| "epoch": 36.64, | |
| "learning_rate": 4.669540229885057e-06, | |
| "loss": 0.0211, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.6990291262135923, | |
| "eval_loss": 2.1283230781555176, | |
| "eval_macro-f1": 0.6734553818830601, | |
| "eval_macro-precision": 0.6869153083999092, | |
| "eval_macro-recall": 0.6662304826991132, | |
| "eval_runtime": 3.6053, | |
| "eval_samples_per_second": 257.124, | |
| "eval_steps_per_second": 8.044, | |
| "eval_weighted-f1": 0.6984888091028846, | |
| "eval_weighted-precision": 0.7035914434375581, | |
| "eval_weighted-recall": 0.6990291262135923, | |
| "step": 8584 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.6957928802588996, | |
| "eval_loss": 2.121021270751953, | |
| "eval_macro-f1": 0.6693193516472609, | |
| "eval_macro-precision": 0.6852915803739035, | |
| "eval_macro-recall": 0.6603953493083169, | |
| "eval_runtime": 3.5959, | |
| "eval_samples_per_second": 257.795, | |
| "eval_steps_per_second": 8.065, | |
| "eval_weighted-f1": 0.6937517174106657, | |
| "eval_weighted-precision": 0.6980351944602388, | |
| "eval_weighted-recall": 0.6957928802588996, | |
| "step": 8816 | |
| }, | |
| { | |
| "epoch": 38.79, | |
| "learning_rate": 1.6762452107279694e-06, | |
| "loss": 0.0175, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.6957928802588996, | |
| "eval_loss": 2.125298023223877, | |
| "eval_macro-f1": 0.670175716053461, | |
| "eval_macro-precision": 0.6823954080156959, | |
| "eval_macro-recall": 0.6637507266661784, | |
| "eval_runtime": 3.6051, | |
| "eval_samples_per_second": 257.138, | |
| "eval_steps_per_second": 8.044, | |
| "eval_weighted-f1": 0.6946838990403312, | |
| "eval_weighted-precision": 0.6981818094755908, | |
| "eval_weighted-recall": 0.6957928802588996, | |
| "step": 9048 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.697950377562028, | |
| "eval_loss": 2.123837471008301, | |
| "eval_macro-f1": 0.672888533432289, | |
| "eval_macro-precision": 0.6846457775029204, | |
| "eval_macro-recall": 0.6663735016267468, | |
| "eval_runtime": 3.6082, | |
| "eval_samples_per_second": 256.916, | |
| "eval_steps_per_second": 8.037, | |
| "eval_weighted-f1": 0.6972181365180552, | |
| "eval_weighted-precision": 0.700475447211827, | |
| "eval_weighted-recall": 0.697950377562028, | |
| "step": 9280 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 9280, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 40, | |
| "save_steps": 500, | |
| "total_flos": 3.933393471111168e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |