| { | |
| "best_metric": 0.16869813203811646, | |
| "best_model_checkpoint": "filing-classification-xlmr-2/checkpoint-4272", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 4272, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0175561797752809, | |
| "grad_norm": 9.653117179870605, | |
| "learning_rate": 2.8037383177570094e-06, | |
| "loss": 3.7369, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0351123595505618, | |
| "grad_norm": 19.201862335205078, | |
| "learning_rate": 5.490654205607477e-06, | |
| "loss": 3.1929, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05266853932584269, | |
| "grad_norm": 22.668563842773438, | |
| "learning_rate": 8.411214953271028e-06, | |
| "loss": 2.7752, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0702247191011236, | |
| "grad_norm": 20.474855422973633, | |
| "learning_rate": 1.1214953271028037e-05, | |
| "loss": 2.389, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0877808988764045, | |
| "grad_norm": 15.759284973144531, | |
| "learning_rate": 1.413551401869159e-05, | |
| "loss": 1.9833, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.10533707865168539, | |
| "grad_norm": 38.20879364013672, | |
| "learning_rate": 1.705607476635514e-05, | |
| "loss": 1.5363, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12289325842696629, | |
| "grad_norm": 10.5639066696167, | |
| "learning_rate": 1.9976635514018695e-05, | |
| "loss": 1.227, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1404494382022472, | |
| "grad_norm": 47.63655471801758, | |
| "learning_rate": 2.2897196261682244e-05, | |
| "loss": 1.1232, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1580056179775281, | |
| "grad_norm": 24.818403244018555, | |
| "learning_rate": 2.570093457943925e-05, | |
| "loss": 0.9905, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.175561797752809, | |
| "grad_norm": 10.783175468444824, | |
| "learning_rate": 2.8621495327102803e-05, | |
| "loss": 1.0071, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1931179775280899, | |
| "grad_norm": 17.002634048461914, | |
| "learning_rate": 3.1542056074766355e-05, | |
| "loss": 0.842, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.21067415730337077, | |
| "grad_norm": 20.313222885131836, | |
| "learning_rate": 3.446261682242991e-05, | |
| "loss": 0.8186, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22823033707865167, | |
| "grad_norm": 24.661361694335938, | |
| "learning_rate": 3.738317757009346e-05, | |
| "loss": 0.7011, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.24578651685393257, | |
| "grad_norm": 59.52761459350586, | |
| "learning_rate": 4.030373831775701e-05, | |
| "loss": 0.8531, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2633426966292135, | |
| "grad_norm": 22.566953659057617, | |
| "learning_rate": 4.3224299065420565e-05, | |
| "loss": 0.835, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2808988764044944, | |
| "grad_norm": 15.610183715820312, | |
| "learning_rate": 4.614485981308411e-05, | |
| "loss": 0.8636, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2984550561797753, | |
| "grad_norm": 9.454130172729492, | |
| "learning_rate": 4.9065420560747664e-05, | |
| "loss": 0.6411, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3160112359550562, | |
| "grad_norm": 11.222116470336914, | |
| "learning_rate": 4.977887617065557e-05, | |
| "loss": 0.6514, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3335674157303371, | |
| "grad_norm": 26.492769241333008, | |
| "learning_rate": 4.945369406867846e-05, | |
| "loss": 0.9043, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.351123595505618, | |
| "grad_norm": 10.550625801086426, | |
| "learning_rate": 4.9128511966701355e-05, | |
| "loss": 0.5749, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3686797752808989, | |
| "grad_norm": 15.46976089477539, | |
| "learning_rate": 4.880332986472425e-05, | |
| "loss": 0.5911, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.3862359550561798, | |
| "grad_norm": 47.346580505371094, | |
| "learning_rate": 4.847814776274714e-05, | |
| "loss": 0.4303, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.40379213483146065, | |
| "grad_norm": 143.05471801757812, | |
| "learning_rate": 4.815296566077003e-05, | |
| "loss": 0.5449, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.42134831460674155, | |
| "grad_norm": 44.15991973876953, | |
| "learning_rate": 4.7827783558792924e-05, | |
| "loss": 0.6729, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.43890449438202245, | |
| "grad_norm": 28.416748046875, | |
| "learning_rate": 4.750260145681582e-05, | |
| "loss": 0.647, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.45646067415730335, | |
| "grad_norm": 23.542097091674805, | |
| "learning_rate": 4.7177419354838716e-05, | |
| "loss": 0.7437, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.47401685393258425, | |
| "grad_norm": 32.17939376831055, | |
| "learning_rate": 4.68522372528616e-05, | |
| "loss": 0.5794, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.49157303370786515, | |
| "grad_norm": 15.835025787353516, | |
| "learning_rate": 4.654006243496358e-05, | |
| "loss": 0.5945, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5091292134831461, | |
| "grad_norm": 26.716590881347656, | |
| "learning_rate": 4.6214880332986475e-05, | |
| "loss": 0.7132, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.526685393258427, | |
| "grad_norm": 168.02224731445312, | |
| "learning_rate": 4.588969823100937e-05, | |
| "loss": 0.5737, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5442415730337079, | |
| "grad_norm": 7.801482677459717, | |
| "learning_rate": 4.5603537981269515e-05, | |
| "loss": 1.8573, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5617977528089888, | |
| "grad_norm": 5.383340358734131, | |
| "learning_rate": 4.527835587929241e-05, | |
| "loss": 0.5317, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5793539325842697, | |
| "grad_norm": 6.86607551574707, | |
| "learning_rate": 4.49531737773153e-05, | |
| "loss": 0.523, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5969101123595506, | |
| "grad_norm": 43.885772705078125, | |
| "learning_rate": 4.462799167533819e-05, | |
| "loss": 0.4805, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6144662921348315, | |
| "grad_norm": 12.806220054626465, | |
| "learning_rate": 4.4302809573361085e-05, | |
| "loss": 0.6384, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6320224719101124, | |
| "grad_norm": 19.224258422851562, | |
| "learning_rate": 4.397762747138398e-05, | |
| "loss": 0.6042, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6495786516853933, | |
| "grad_norm": 24.135557174682617, | |
| "learning_rate": 4.365244536940687e-05, | |
| "loss": 0.6164, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6671348314606742, | |
| "grad_norm": 7.6507487297058105, | |
| "learning_rate": 4.332726326742976e-05, | |
| "loss": 0.5067, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6846910112359551, | |
| "grad_norm": 16.015581130981445, | |
| "learning_rate": 4.3002081165452654e-05, | |
| "loss": 0.5762, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.702247191011236, | |
| "grad_norm": 1.4693474769592285, | |
| "learning_rate": 4.267689906347555e-05, | |
| "loss": 0.458, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7198033707865169, | |
| "grad_norm": 163.2286834716797, | |
| "learning_rate": 4.235171696149844e-05, | |
| "loss": 0.7918, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7373595505617978, | |
| "grad_norm": 0.30899184942245483, | |
| "learning_rate": 4.202653485952133e-05, | |
| "loss": 0.5819, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7549157303370787, | |
| "grad_norm": 2.51499342918396, | |
| "learning_rate": 4.1701352757544224e-05, | |
| "loss": 0.7124, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.7724719101123596, | |
| "grad_norm": 27.783517837524414, | |
| "learning_rate": 4.137617065556712e-05, | |
| "loss": 0.4374, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7900280898876404, | |
| "grad_norm": 53.92084884643555, | |
| "learning_rate": 4.1050988553590015e-05, | |
| "loss": 0.3742, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8075842696629213, | |
| "grad_norm": 4.004390716552734, | |
| "learning_rate": 4.072580645161291e-05, | |
| "loss": 0.6952, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8251404494382022, | |
| "grad_norm": 21.593162536621094, | |
| "learning_rate": 4.040062434963579e-05, | |
| "loss": 0.365, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8426966292134831, | |
| "grad_norm": 4.670670509338379, | |
| "learning_rate": 4.007544224765869e-05, | |
| "loss": 0.2934, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.860252808988764, | |
| "grad_norm": 18.557373046875, | |
| "learning_rate": 3.9750260145681585e-05, | |
| "loss": 0.5487, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.8778089887640449, | |
| "grad_norm": 45.69279479980469, | |
| "learning_rate": 3.942507804370448e-05, | |
| "loss": 0.7667, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8953651685393258, | |
| "grad_norm": 115.01453399658203, | |
| "learning_rate": 3.909989594172737e-05, | |
| "loss": 0.7059, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.9129213483146067, | |
| "grad_norm": 33.03873825073242, | |
| "learning_rate": 3.877471383975026e-05, | |
| "loss": 0.8485, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9304775280898876, | |
| "grad_norm": 14.514985084533691, | |
| "learning_rate": 3.8449531737773154e-05, | |
| "loss": 0.571, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9480337078651685, | |
| "grad_norm": 15.848938941955566, | |
| "learning_rate": 3.8124349635796047e-05, | |
| "loss": 0.4613, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9655898876404494, | |
| "grad_norm": 9.903932571411133, | |
| "learning_rate": 3.779916753381894e-05, | |
| "loss": 0.5297, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.9831460674157303, | |
| "grad_norm": 0.5407254695892334, | |
| "learning_rate": 3.747398543184184e-05, | |
| "loss": 0.5063, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8672286617492097, | |
| "eval_f1_macro": 0.39173156656801417, | |
| "eval_f1_micro": 0.8672286617492097, | |
| "eval_f1_weighted": 0.8642044037651806, | |
| "eval_loss": 0.42791736125946045, | |
| "eval_precision_macro": 0.3952455698245066, | |
| "eval_precision_micro": 0.8672286617492097, | |
| "eval_precision_weighted": 0.8846072086554072, | |
| "eval_recall_macro": 0.41055025157749075, | |
| "eval_recall_micro": 0.8672286617492097, | |
| "eval_recall_weighted": 0.8672286617492097, | |
| "eval_runtime": 305.1224, | |
| "eval_samples_per_second": 9.331, | |
| "eval_steps_per_second": 0.583, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 1.0007022471910112, | |
| "grad_norm": 5.628513813018799, | |
| "learning_rate": 3.714880332986473e-05, | |
| "loss": 0.4114, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.0182584269662922, | |
| "grad_norm": 0.1851166933774948, | |
| "learning_rate": 3.6823621227887616e-05, | |
| "loss": 0.3392, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.035814606741573, | |
| "grad_norm": 38.67854690551758, | |
| "learning_rate": 3.65114464099896e-05, | |
| "loss": 0.386, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.053370786516854, | |
| "grad_norm": 23.285263061523438, | |
| "learning_rate": 3.618626430801249e-05, | |
| "loss": 0.4073, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.0709269662921348, | |
| "grad_norm": 1.340030550956726, | |
| "learning_rate": 3.5861082206035376e-05, | |
| "loss": 0.4067, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.0884831460674158, | |
| "grad_norm": 5.264111042022705, | |
| "learning_rate": 3.5535900104058275e-05, | |
| "loss": 0.4098, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.1060393258426966, | |
| "grad_norm": 23.070083618164062, | |
| "learning_rate": 3.521071800208117e-05, | |
| "loss": 0.3824, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.1235955056179776, | |
| "grad_norm": 1.6395037174224854, | |
| "learning_rate": 3.488553590010406e-05, | |
| "loss": 0.4457, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.1411516853932584, | |
| "grad_norm": 36.9826774597168, | |
| "learning_rate": 3.456035379812695e-05, | |
| "loss": 0.3469, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.1587078651685394, | |
| "grad_norm": 4.204775333404541, | |
| "learning_rate": 3.423517169614985e-05, | |
| "loss": 0.4349, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.1762640449438202, | |
| "grad_norm": 45.272926330566406, | |
| "learning_rate": 3.390998959417274e-05, | |
| "loss": 0.3978, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.1938202247191012, | |
| "grad_norm": 0.915517270565033, | |
| "learning_rate": 3.358480749219563e-05, | |
| "loss": 0.4463, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.211376404494382, | |
| "grad_norm": 1.780409336090088, | |
| "learning_rate": 3.325962539021852e-05, | |
| "loss": 0.4707, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.228932584269663, | |
| "grad_norm": 2.592899799346924, | |
| "learning_rate": 3.293444328824142e-05, | |
| "loss": 0.3541, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.2464887640449438, | |
| "grad_norm": 8.29692554473877, | |
| "learning_rate": 3.260926118626431e-05, | |
| "loss": 0.5227, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.2640449438202248, | |
| "grad_norm": 3.768015146255493, | |
| "learning_rate": 3.22840790842872e-05, | |
| "loss": 0.2983, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.2816011235955056, | |
| "grad_norm": 23.4680233001709, | |
| "learning_rate": 3.195889698231009e-05, | |
| "loss": 0.6556, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.2991573033707864, | |
| "grad_norm": 25.872844696044922, | |
| "learning_rate": 3.163371488033299e-05, | |
| "loss": 0.3711, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.3167134831460674, | |
| "grad_norm": 48.59894561767578, | |
| "learning_rate": 3.130853277835588e-05, | |
| "loss": 0.5387, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.3342696629213484, | |
| "grad_norm": 8.37211799621582, | |
| "learning_rate": 3.0983350676378775e-05, | |
| "loss": 0.3919, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.3518258426966292, | |
| "grad_norm": 4.695379734039307, | |
| "learning_rate": 3.065816857440167e-05, | |
| "loss": 0.3456, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.36938202247191, | |
| "grad_norm": 1.2714042663574219, | |
| "learning_rate": 3.0332986472424557e-05, | |
| "loss": 0.2933, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.386938202247191, | |
| "grad_norm": 17.301929473876953, | |
| "learning_rate": 3.0007804370447452e-05, | |
| "loss": 0.2552, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.404494382022472, | |
| "grad_norm": 0.5129703879356384, | |
| "learning_rate": 2.9682622268470345e-05, | |
| "loss": 0.2173, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.4220505617977528, | |
| "grad_norm": 48.075435638427734, | |
| "learning_rate": 2.9357440166493237e-05, | |
| "loss": 0.3967, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.4396067415730336, | |
| "grad_norm": 0.33172833919525146, | |
| "learning_rate": 2.9032258064516133e-05, | |
| "loss": 0.4736, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.4571629213483146, | |
| "grad_norm": 0.18778946995735168, | |
| "learning_rate": 2.8707075962539022e-05, | |
| "loss": 0.243, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.4747191011235956, | |
| "grad_norm": 8.245080947875977, | |
| "learning_rate": 2.8381893860561914e-05, | |
| "loss": 0.2598, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.4922752808988764, | |
| "grad_norm": 23.08641242980957, | |
| "learning_rate": 2.8056711758584807e-05, | |
| "loss": 0.4456, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.5098314606741572, | |
| "grad_norm": 36.80413818359375, | |
| "learning_rate": 2.7731529656607702e-05, | |
| "loss": 0.4345, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.5273876404494382, | |
| "grad_norm": 2.1705923080444336, | |
| "learning_rate": 2.7406347554630595e-05, | |
| "loss": 0.3668, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.5449438202247192, | |
| "grad_norm": 0.3322332799434662, | |
| "learning_rate": 2.708116545265349e-05, | |
| "loss": 0.2902, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.5625, | |
| "grad_norm": 0.01797402836382389, | |
| "learning_rate": 2.675598335067638e-05, | |
| "loss": 0.2279, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.5800561797752808, | |
| "grad_norm": 0.06612106412649155, | |
| "learning_rate": 2.6430801248699272e-05, | |
| "loss": 0.3507, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.5976123595505618, | |
| "grad_norm": 2.4665720462799072, | |
| "learning_rate": 2.6105619146722164e-05, | |
| "loss": 0.3281, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.6151685393258428, | |
| "grad_norm": 0.3211938440799713, | |
| "learning_rate": 2.578043704474506e-05, | |
| "loss": 0.2798, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.6327247191011236, | |
| "grad_norm": 28.933242797851562, | |
| "learning_rate": 2.5455254942767953e-05, | |
| "loss": 0.3681, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.6502808988764044, | |
| "grad_norm": 14.026947021484375, | |
| "learning_rate": 2.513007284079084e-05, | |
| "loss": 0.3509, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.6678370786516854, | |
| "grad_norm": 0.3984997570514679, | |
| "learning_rate": 2.4804890738813737e-05, | |
| "loss": 0.2291, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.6853932584269664, | |
| "grad_norm": 0.042692411690950394, | |
| "learning_rate": 2.447970863683663e-05, | |
| "loss": 0.3134, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.7029494382022472, | |
| "grad_norm": 0.6199696063995361, | |
| "learning_rate": 2.4154526534859522e-05, | |
| "loss": 0.1636, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.720505617977528, | |
| "grad_norm": 29.01231575012207, | |
| "learning_rate": 2.3829344432882414e-05, | |
| "loss": 0.4575, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.738061797752809, | |
| "grad_norm": 0.06194797903299332, | |
| "learning_rate": 2.3504162330905307e-05, | |
| "loss": 0.2346, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.75561797752809, | |
| "grad_norm": 5.528661251068115, | |
| "learning_rate": 2.3178980228928203e-05, | |
| "loss": 0.2984, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.7731741573033708, | |
| "grad_norm": 0.022516461089253426, | |
| "learning_rate": 2.285379812695109e-05, | |
| "loss": 0.1701, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.7907303370786516, | |
| "grad_norm": 0.10887257009744644, | |
| "learning_rate": 2.2528616024973987e-05, | |
| "loss": 0.1982, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.8082865168539326, | |
| "grad_norm": 5.863889694213867, | |
| "learning_rate": 2.220343392299688e-05, | |
| "loss": 0.2044, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.8258426966292136, | |
| "grad_norm": 22.965593338012695, | |
| "learning_rate": 2.1878251821019772e-05, | |
| "loss": 0.2176, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.8433988764044944, | |
| "grad_norm": 1.1603304147720337, | |
| "learning_rate": 2.1553069719042665e-05, | |
| "loss": 0.2629, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.8609550561797752, | |
| "grad_norm": 94.22467803955078, | |
| "learning_rate": 2.122788761706556e-05, | |
| "loss": 0.2317, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.8785112359550562, | |
| "grad_norm": 88.58509063720703, | |
| "learning_rate": 2.090270551508845e-05, | |
| "loss": 0.3063, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.8960674157303372, | |
| "grad_norm": 0.061591584235429764, | |
| "learning_rate": 2.0577523413111345e-05, | |
| "loss": 0.1056, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.913623595505618, | |
| "grad_norm": 0.016784893348813057, | |
| "learning_rate": 2.0252341311134237e-05, | |
| "loss": 0.3275, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.9311797752808988, | |
| "grad_norm": 0.08787299692630768, | |
| "learning_rate": 1.992715920915713e-05, | |
| "loss": 0.1606, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.9487359550561798, | |
| "grad_norm": 17.29918098449707, | |
| "learning_rate": 1.9601977107180022e-05, | |
| "loss": 0.1864, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.9662921348314608, | |
| "grad_norm": 0.06715293228626251, | |
| "learning_rate": 1.9276795005202915e-05, | |
| "loss": 0.2242, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.9838483146067416, | |
| "grad_norm": 19.53911590576172, | |
| "learning_rate": 1.8951612903225807e-05, | |
| "loss": 0.2461, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9525816649104321, | |
| "eval_f1_macro": 0.5638040591012018, | |
| "eval_f1_micro": 0.9525816649104321, | |
| "eval_f1_weighted": 0.9477276506866364, | |
| "eval_loss": 0.23304341733455658, | |
| "eval_precision_macro": 0.5866607806536552, | |
| "eval_precision_micro": 0.9525816649104321, | |
| "eval_precision_weighted": 0.9451448284466338, | |
| "eval_recall_macro": 0.5565307246666726, | |
| "eval_recall_micro": 0.9525816649104321, | |
| "eval_recall_weighted": 0.9525816649104321, | |
| "eval_runtime": 307.8773, | |
| "eval_samples_per_second": 9.247, | |
| "eval_steps_per_second": 0.578, | |
| "step": 2848 | |
| }, | |
| { | |
| "epoch": 2.0014044943820224, | |
| "grad_norm": 0.06152953952550888, | |
| "learning_rate": 1.86264308012487e-05, | |
| "loss": 0.1596, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.018960674157303, | |
| "grad_norm": 35.94683074951172, | |
| "learning_rate": 1.8301248699271592e-05, | |
| "loss": 0.1538, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.0365168539325844, | |
| "grad_norm": 8.920159339904785, | |
| "learning_rate": 1.7976066597294488e-05, | |
| "loss": 0.26, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.054073033707865, | |
| "grad_norm": 12.313779830932617, | |
| "learning_rate": 1.765088449531738e-05, | |
| "loss": 0.2101, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.071629213483146, | |
| "grad_norm": 0.3984639346599579, | |
| "learning_rate": 1.7325702393340272e-05, | |
| "loss": 0.0441, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.089185393258427, | |
| "grad_norm": 19.14760398864746, | |
| "learning_rate": 1.7000520291363165e-05, | |
| "loss": 0.2119, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.106741573033708, | |
| "grad_norm": 28.54094123840332, | |
| "learning_rate": 1.6675338189386057e-05, | |
| "loss": 0.3158, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.124297752808989, | |
| "grad_norm": 0.06162691116333008, | |
| "learning_rate": 1.635015608740895e-05, | |
| "loss": 0.2421, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.1418539325842696, | |
| "grad_norm": 4.776229381561279, | |
| "learning_rate": 1.6024973985431842e-05, | |
| "loss": 0.1528, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.1594101123595504, | |
| "grad_norm": 15.048027992248535, | |
| "learning_rate": 1.5699791883454734e-05, | |
| "loss": 0.0803, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.1769662921348316, | |
| "grad_norm": 0.01680239848792553, | |
| "learning_rate": 1.5374609781477627e-05, | |
| "loss": 0.0592, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.1945224719101124, | |
| "grad_norm": 5.08790397644043, | |
| "learning_rate": 1.5049427679500522e-05, | |
| "loss": 0.1673, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.212078651685393, | |
| "grad_norm": 0.014381933026015759, | |
| "learning_rate": 1.4724245577523413e-05, | |
| "loss": 0.15, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.229634831460674, | |
| "grad_norm": 0.036015789955854416, | |
| "learning_rate": 1.4399063475546307e-05, | |
| "loss": 0.0784, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.247191011235955, | |
| "grad_norm": 0.25334903597831726, | |
| "learning_rate": 1.4073881373569201e-05, | |
| "loss": 0.2201, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.264747191011236, | |
| "grad_norm": 0.04687151685357094, | |
| "learning_rate": 1.3748699271592092e-05, | |
| "loss": 0.1387, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.282303370786517, | |
| "grad_norm": 0.015004009008407593, | |
| "learning_rate": 1.3423517169614986e-05, | |
| "loss": 0.2383, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.2998595505617976, | |
| "grad_norm": 0.027077874168753624, | |
| "learning_rate": 1.3098335067637878e-05, | |
| "loss": 0.0806, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.317415730337079, | |
| "grad_norm": 3.859339714050293, | |
| "learning_rate": 1.277315296566077e-05, | |
| "loss": 0.2287, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.3349719101123596, | |
| "grad_norm": 0.00701533816754818, | |
| "learning_rate": 1.2447970863683663e-05, | |
| "loss": 0.1842, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.3525280898876404, | |
| "grad_norm": 0.05476459860801697, | |
| "learning_rate": 1.2122788761706557e-05, | |
| "loss": 0.1219, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.370084269662921, | |
| "grad_norm": 10.582536697387695, | |
| "learning_rate": 1.179760665972945e-05, | |
| "loss": 0.1652, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.3876404494382024, | |
| "grad_norm": 6.6982598304748535, | |
| "learning_rate": 1.1472424557752342e-05, | |
| "loss": 0.0834, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.405196629213483, | |
| "grad_norm": 1.1341116428375244, | |
| "learning_rate": 1.1147242455775234e-05, | |
| "loss": 0.2628, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.422752808988764, | |
| "grad_norm": 0.013976281508803368, | |
| "learning_rate": 1.0822060353798127e-05, | |
| "loss": 0.1003, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.440308988764045, | |
| "grad_norm": 54.462608337402344, | |
| "learning_rate": 1.0496878251821019e-05, | |
| "loss": 0.1237, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.457865168539326, | |
| "grad_norm": 6.74662971496582, | |
| "learning_rate": 1.0171696149843913e-05, | |
| "loss": 0.1257, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.475421348314607, | |
| "grad_norm": 0.050433021038770676, | |
| "learning_rate": 9.846514047866806e-06, | |
| "loss": 0.1494, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.4929775280898876, | |
| "grad_norm": 0.013178324326872826, | |
| "learning_rate": 9.521331945889698e-06, | |
| "loss": 0.1797, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.5105337078651684, | |
| "grad_norm": 0.04092850908637047, | |
| "learning_rate": 9.19614984391259e-06, | |
| "loss": 0.2765, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.5280898876404496, | |
| "grad_norm": 21.01885414123535, | |
| "learning_rate": 8.870967741935484e-06, | |
| "loss": 0.2107, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.5456460674157304, | |
| "grad_norm": 7.4579997062683105, | |
| "learning_rate": 8.545785639958377e-06, | |
| "loss": 0.2029, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.563202247191011, | |
| "grad_norm": 0.04982079565525055, | |
| "learning_rate": 8.22060353798127e-06, | |
| "loss": 0.1373, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.580758426966292, | |
| "grad_norm": 0.033542681485414505, | |
| "learning_rate": 7.895421436004163e-06, | |
| "loss": 0.2837, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.5983146067415728, | |
| "grad_norm": 0.010033758357167244, | |
| "learning_rate": 7.570239334027056e-06, | |
| "loss": 0.0557, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.615870786516854, | |
| "grad_norm": 13.881911277770996, | |
| "learning_rate": 7.245057232049948e-06, | |
| "loss": 0.1513, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.633426966292135, | |
| "grad_norm": 0.040514521300792694, | |
| "learning_rate": 6.91987513007284e-06, | |
| "loss": 0.2899, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.6509831460674156, | |
| "grad_norm": 0.017260883003473282, | |
| "learning_rate": 6.5946930280957345e-06, | |
| "loss": 0.0614, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.668539325842697, | |
| "grad_norm": 2.554452419281006, | |
| "learning_rate": 6.269510926118627e-06, | |
| "loss": 0.0714, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.6860955056179776, | |
| "grad_norm": 4.296275615692139, | |
| "learning_rate": 5.944328824141519e-06, | |
| "loss": 0.1717, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.7036516853932584, | |
| "grad_norm": 10.5920991897583, | |
| "learning_rate": 5.6191467221644125e-06, | |
| "loss": 0.1186, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.721207865168539, | |
| "grad_norm": 0.007871459238231182, | |
| "learning_rate": 5.293964620187305e-06, | |
| "loss": 0.164, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.73876404494382, | |
| "grad_norm": 0.1039140522480011, | |
| "learning_rate": 4.968782518210198e-06, | |
| "loss": 0.0116, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.756320224719101, | |
| "grad_norm": 0.010159538127481937, | |
| "learning_rate": 4.643600416233091e-06, | |
| "loss": 0.0667, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.773876404494382, | |
| "grad_norm": 0.015021737664937973, | |
| "learning_rate": 4.318418314255984e-06, | |
| "loss": 0.1262, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.791432584269663, | |
| "grad_norm": 0.009528973139822483, | |
| "learning_rate": 3.993236212278877e-06, | |
| "loss": 0.1403, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.808988764044944, | |
| "grad_norm": 0.32688748836517334, | |
| "learning_rate": 3.668054110301769e-06, | |
| "loss": 0.1795, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.826544943820225, | |
| "grad_norm": 0.021688124164938927, | |
| "learning_rate": 3.342872008324662e-06, | |
| "loss": 0.1127, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.8441011235955056, | |
| "grad_norm": 0.030653346329927444, | |
| "learning_rate": 3.017689906347555e-06, | |
| "loss": 0.1341, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.8616573033707864, | |
| "grad_norm": 0.0213681161403656, | |
| "learning_rate": 2.6925078043704477e-06, | |
| "loss": 0.1922, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.879213483146067, | |
| "grad_norm": 0.039987191557884216, | |
| "learning_rate": 2.3673257023933405e-06, | |
| "loss": 0.0696, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.8967696629213484, | |
| "grad_norm": 2.405183792114258, | |
| "learning_rate": 2.0421436004162333e-06, | |
| "loss": 0.2016, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.914325842696629, | |
| "grad_norm": 14.973130226135254, | |
| "learning_rate": 1.716961498439126e-06, | |
| "loss": 0.2583, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.93188202247191, | |
| "grad_norm": 0.013798320665955544, | |
| "learning_rate": 1.3917793964620187e-06, | |
| "loss": 0.092, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.949438202247191, | |
| "grad_norm": 0.055772531777620316, | |
| "learning_rate": 1.0665972944849116e-06, | |
| "loss": 0.1097, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.966994382022472, | |
| "grad_norm": 9.692510604858398, | |
| "learning_rate": 7.414151925078044e-07, | |
| "loss": 0.117, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.984550561797753, | |
| "grad_norm": 0.012657186016440392, | |
| "learning_rate": 4.1623309053069727e-07, | |
| "loss": 0.0464, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9617140850017563, | |
| "eval_f1_macro": 0.6470233113292341, | |
| "eval_f1_micro": 0.9617140850017563, | |
| "eval_f1_weighted": 0.9597252404005653, | |
| "eval_loss": 0.16869813203811646, | |
| "eval_precision_macro": 0.6657138827178418, | |
| "eval_precision_micro": 0.9617140850017563, | |
| "eval_precision_weighted": 0.9600327052750102, | |
| "eval_recall_macro": 0.6540179851686874, | |
| "eval_recall_micro": 0.9617140850017563, | |
| "eval_recall_weighted": 0.9617140850017563, | |
| "eval_runtime": 301.0331, | |
| "eval_samples_per_second": 9.457, | |
| "eval_steps_per_second": 0.591, | |
| "step": 4272 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 4272, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.1836675993421824e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |