| { | |
| "best_metric": 5.022224426269531, | |
| "best_model_checkpoint": "./ViMedical_Diseases/checkpoint-400", | |
| "epoch": 84.76821192052981, | |
| "eval_steps": 500, | |
| "global_step": 400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.85, | |
| "eval_accuracy": 0.0024875621890547263, | |
| "eval_f1": 1.2345221781909311e-05, | |
| "eval_loss": 6.415043830871582, | |
| "eval_runtime": 0.3946, | |
| "eval_samples_per_second": 3055.887, | |
| "eval_steps_per_second": 25.339, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_accuracy": 0.0024875621890547263, | |
| "eval_f1": 1.2756729174639624e-05, | |
| "eval_loss": 6.412064075469971, | |
| "eval_runtime": 0.3736, | |
| "eval_samples_per_second": 3227.701, | |
| "eval_steps_per_second": 26.764, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "eval_accuracy": 0.0024875621890547263, | |
| "eval_f1": 1.338598487383709e-05, | |
| "eval_loss": 6.407182693481445, | |
| "eval_runtime": 0.3753, | |
| "eval_samples_per_second": 3213.746, | |
| "eval_steps_per_second": 26.648, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "eval_accuracy": 0.0041459369817578775, | |
| "eval_f1": 0.0005888539889881497, | |
| "eval_loss": 6.397731304168701, | |
| "eval_runtime": 0.3765, | |
| "eval_samples_per_second": 3202.793, | |
| "eval_steps_per_second": 26.557, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "eval_accuracy": 0.0041459369817578775, | |
| "eval_f1": 0.00024318292743308673, | |
| "eval_loss": 6.378692626953125, | |
| "eval_runtime": 0.375, | |
| "eval_samples_per_second": 3215.914, | |
| "eval_steps_per_second": 26.666, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "eval_accuracy": 0.006633499170812604, | |
| "eval_f1": 0.0006805627237639513, | |
| "eval_loss": 6.348721981048584, | |
| "eval_runtime": 0.3755, | |
| "eval_samples_per_second": 3212.024, | |
| "eval_steps_per_second": 26.634, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "eval_accuracy": 0.00912106135986733, | |
| "eval_f1": 0.0044415843789941045, | |
| "eval_loss": 6.305507183074951, | |
| "eval_runtime": 0.3784, | |
| "eval_samples_per_second": 3187.077, | |
| "eval_steps_per_second": 26.427, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "eval_accuracy": 0.015754560530679935, | |
| "eval_f1": 0.004968013448467957, | |
| "eval_loss": 6.288573265075684, | |
| "eval_runtime": 0.3772, | |
| "eval_samples_per_second": 3197.112, | |
| "eval_steps_per_second": 26.51, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "eval_accuracy": 0.01658374792703151, | |
| "eval_f1": 0.004671981131856493, | |
| "eval_loss": 6.297088623046875, | |
| "eval_runtime": 0.3776, | |
| "eval_samples_per_second": 3193.556, | |
| "eval_steps_per_second": 26.481, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "eval_accuracy": 0.013266998341625208, | |
| "eval_f1": 0.004042787456966562, | |
| "eval_loss": 6.216242790222168, | |
| "eval_runtime": 0.3779, | |
| "eval_samples_per_second": 3191.019, | |
| "eval_steps_per_second": 26.46, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "eval_accuracy": 0.020729684908789386, | |
| "eval_f1": 0.008522136976574753, | |
| "eval_loss": 6.187163352966309, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.288, | |
| "eval_steps_per_second": 26.371, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 11.87, | |
| "eval_accuracy": 0.029850746268656716, | |
| "eval_f1": 0.014040979403889549, | |
| "eval_loss": 6.177770137786865, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.276, | |
| "eval_steps_per_second": 26.37, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "eval_accuracy": 0.02902155887230514, | |
| "eval_f1": 0.011104717761289505, | |
| "eval_loss": 6.158536911010742, | |
| "eval_runtime": 0.3786, | |
| "eval_samples_per_second": 3185.18, | |
| "eval_steps_per_second": 26.411, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "eval_accuracy": 0.02570480928689884, | |
| "eval_f1": 0.01033900277156658, | |
| "eval_loss": 6.1090803146362305, | |
| "eval_runtime": 0.38, | |
| "eval_samples_per_second": 3173.403, | |
| "eval_steps_per_second": 26.313, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "eval_accuracy": 0.03316749585406302, | |
| "eval_f1": 0.015474453231986877, | |
| "eval_loss": 6.08702278137207, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.296, | |
| "eval_steps_per_second": 26.371, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "eval_accuracy": 0.04809286898839138, | |
| "eval_f1": 0.02513988397193506, | |
| "eval_loss": 6.0609517097473145, | |
| "eval_runtime": 0.3801, | |
| "eval_samples_per_second": 3172.69, | |
| "eval_steps_per_second": 26.308, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "eval_accuracy": 0.03814262023217247, | |
| "eval_f1": 0.01813347312377026, | |
| "eval_loss": 6.025730609893799, | |
| "eval_runtime": 0.3785, | |
| "eval_samples_per_second": 3185.844, | |
| "eval_steps_per_second": 26.417, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "eval_accuracy": 0.05555555555555555, | |
| "eval_f1": 0.0316611803261535, | |
| "eval_loss": 6.014459133148193, | |
| "eval_runtime": 0.3787, | |
| "eval_samples_per_second": 3184.258, | |
| "eval_steps_per_second": 26.403, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "eval_accuracy": 0.04311774461028192, | |
| "eval_f1": 0.02307944350501337, | |
| "eval_loss": 5.969913005828857, | |
| "eval_runtime": 0.3804, | |
| "eval_samples_per_second": 3169.946, | |
| "eval_steps_per_second": 26.285, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "eval_accuracy": 0.05638474295190713, | |
| "eval_f1": 0.0302836119107397, | |
| "eval_loss": 5.954006195068359, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3176.755, | |
| "eval_steps_per_second": 26.341, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 20.98, | |
| "eval_accuracy": 0.06301824212271974, | |
| "eval_f1": 0.03687442226058399, | |
| "eval_loss": 5.912012100219727, | |
| "eval_runtime": 0.3781, | |
| "eval_samples_per_second": 3189.303, | |
| "eval_steps_per_second": 26.445, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 21.83, | |
| "eval_accuracy": 0.06384742951907131, | |
| "eval_f1": 0.03535293317796501, | |
| "eval_loss": 5.8892645835876465, | |
| "eval_runtime": 0.3799, | |
| "eval_samples_per_second": 3174.661, | |
| "eval_steps_per_second": 26.324, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 22.89, | |
| "eval_accuracy": 0.07711442786069651, | |
| "eval_f1": 0.04528489909367382, | |
| "eval_loss": 5.885714530944824, | |
| "eval_runtime": 0.3814, | |
| "eval_samples_per_second": 3162.388, | |
| "eval_steps_per_second": 26.222, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 23.95, | |
| "eval_accuracy": 0.06550580431177445, | |
| "eval_f1": 0.03852630441009513, | |
| "eval_loss": 5.839115142822266, | |
| "eval_runtime": 0.3786, | |
| "eval_samples_per_second": 3185.788, | |
| "eval_steps_per_second": 26.416, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 24.79, | |
| "eval_accuracy": 0.07877280265339967, | |
| "eval_f1": 0.050102448183655314, | |
| "eval_loss": 5.812053203582764, | |
| "eval_runtime": 0.3783, | |
| "eval_samples_per_second": 3187.772, | |
| "eval_steps_per_second": 26.433, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 25.85, | |
| "eval_accuracy": 0.0845771144278607, | |
| "eval_f1": 0.05181174832064436, | |
| "eval_loss": 5.789350986480713, | |
| "eval_runtime": 0.3785, | |
| "eval_samples_per_second": 3186.218, | |
| "eval_steps_per_second": 26.42, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 26.91, | |
| "eval_accuracy": 0.09286898839137644, | |
| "eval_f1": 0.0554051006900966, | |
| "eval_loss": 5.8099284172058105, | |
| "eval_runtime": 0.3795, | |
| "eval_samples_per_second": 3177.946, | |
| "eval_steps_per_second": 26.351, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 27.97, | |
| "eval_accuracy": 0.08208955223880597, | |
| "eval_f1": 0.052706293384051014, | |
| "eval_loss": 5.745517253875732, | |
| "eval_runtime": 0.379, | |
| "eval_samples_per_second": 3182.087, | |
| "eval_steps_per_second": 26.385, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 28.82, | |
| "eval_accuracy": 0.08706467661691543, | |
| "eval_f1": 0.05317439161254717, | |
| "eval_loss": 5.725302696228027, | |
| "eval_runtime": 0.38, | |
| "eval_samples_per_second": 3173.908, | |
| "eval_steps_per_second": 26.318, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 29.88, | |
| "eval_accuracy": 0.09618573797678276, | |
| "eval_f1": 0.05978131215599709, | |
| "eval_loss": 5.701379299163818, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.126, | |
| "eval_steps_per_second": 26.369, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 30.94, | |
| "eval_accuracy": 0.09286898839137644, | |
| "eval_f1": 0.057675351156170285, | |
| "eval_loss": 5.6744208335876465, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.732, | |
| "eval_steps_per_second": 26.374, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 31.79, | |
| "grad_norm": 2.17041015625, | |
| "learning_rate": 1.255e-05, | |
| "loss": 6.0949, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.09950248756218906, | |
| "eval_f1": 0.06490404658367951, | |
| "eval_loss": 5.660266399383545, | |
| "eval_runtime": 0.379, | |
| "eval_samples_per_second": 3182.291, | |
| "eval_steps_per_second": 26.387, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 32.85, | |
| "eval_accuracy": 0.09867330016583747, | |
| "eval_f1": 0.0630415651434736, | |
| "eval_loss": 5.6351704597473145, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3176.904, | |
| "eval_steps_per_second": 26.342, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 33.91, | |
| "eval_accuracy": 0.09867330016583747, | |
| "eval_f1": 0.060441362912511246, | |
| "eval_loss": 5.600429058074951, | |
| "eval_runtime": 0.3801, | |
| "eval_samples_per_second": 3172.485, | |
| "eval_steps_per_second": 26.306, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 34.97, | |
| "eval_accuracy": 0.10281923714759536, | |
| "eval_f1": 0.0637386932721227, | |
| "eval_loss": 5.598119258880615, | |
| "eval_runtime": 0.3795, | |
| "eval_samples_per_second": 3177.533, | |
| "eval_steps_per_second": 26.348, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 35.81, | |
| "eval_accuracy": 0.1011608623548922, | |
| "eval_f1": 0.06296000608037869, | |
| "eval_loss": 5.570890426635742, | |
| "eval_runtime": 0.3788, | |
| "eval_samples_per_second": 3183.869, | |
| "eval_steps_per_second": 26.4, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 36.87, | |
| "eval_accuracy": 0.10530679933665009, | |
| "eval_f1": 0.06954752384683803, | |
| "eval_loss": 5.540558815002441, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3176.978, | |
| "eval_steps_per_second": 26.343, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 37.93, | |
| "eval_accuracy": 0.11442786069651742, | |
| "eval_f1": 0.07287036657509441, | |
| "eval_loss": 5.520463943481445, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.006, | |
| "eval_steps_per_second": 26.368, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 38.99, | |
| "eval_accuracy": 0.10862354892205639, | |
| "eval_f1": 0.07094764147857342, | |
| "eval_loss": 5.498680114746094, | |
| "eval_runtime": 0.3784, | |
| "eval_samples_per_second": 3187.079, | |
| "eval_steps_per_second": 26.427, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 39.84, | |
| "eval_accuracy": 0.1111111111111111, | |
| "eval_f1": 0.06995636683961066, | |
| "eval_loss": 5.477349758148193, | |
| "eval_runtime": 0.3792, | |
| "eval_samples_per_second": 3180.04, | |
| "eval_steps_per_second": 26.368, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 40.9, | |
| "eval_accuracy": 0.12023217247097844, | |
| "eval_f1": 0.0803838202287807, | |
| "eval_loss": 5.464529514312744, | |
| "eval_runtime": 0.3783, | |
| "eval_samples_per_second": 3188.288, | |
| "eval_steps_per_second": 26.437, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 41.96, | |
| "eval_accuracy": 0.12189054726368159, | |
| "eval_f1": 0.0797876479133484, | |
| "eval_loss": 5.439937591552734, | |
| "eval_runtime": 0.3805, | |
| "eval_samples_per_second": 3169.807, | |
| "eval_steps_per_second": 26.284, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 42.81, | |
| "eval_accuracy": 0.11774461028192372, | |
| "eval_f1": 0.07397147599836879, | |
| "eval_loss": 5.417842388153076, | |
| "eval_runtime": 0.3791, | |
| "eval_samples_per_second": 3181.28, | |
| "eval_steps_per_second": 26.379, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 43.87, | |
| "eval_accuracy": 0.12271973466003316, | |
| "eval_f1": 0.08191085838440999, | |
| "eval_loss": 5.408046245574951, | |
| "eval_runtime": 0.379, | |
| "eval_samples_per_second": 3181.692, | |
| "eval_steps_per_second": 26.382, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 44.93, | |
| "eval_accuracy": 0.1310116086235489, | |
| "eval_f1": 0.09138005527552184, | |
| "eval_loss": 5.374771595001221, | |
| "eval_runtime": 0.3787, | |
| "eval_samples_per_second": 3184.593, | |
| "eval_steps_per_second": 26.406, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 45.99, | |
| "eval_accuracy": 0.12769485903814262, | |
| "eval_f1": 0.08429454245370348, | |
| "eval_loss": 5.369958400726318, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3177.034, | |
| "eval_steps_per_second": 26.344, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 46.83, | |
| "eval_accuracy": 0.1252072968490879, | |
| "eval_f1": 0.0828030322134829, | |
| "eval_loss": 5.345793724060059, | |
| "eval_runtime": 0.3798, | |
| "eval_samples_per_second": 3175.327, | |
| "eval_steps_per_second": 26.329, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 47.89, | |
| "eval_accuracy": 0.12686567164179105, | |
| "eval_f1": 0.08616638865359265, | |
| "eval_loss": 5.33195686340332, | |
| "eval_runtime": 0.3775, | |
| "eval_samples_per_second": 3194.5, | |
| "eval_steps_per_second": 26.488, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 48.95, | |
| "eval_accuracy": 0.13598673300165837, | |
| "eval_f1": 0.09754942424775702, | |
| "eval_loss": 5.319748878479004, | |
| "eval_runtime": 0.3785, | |
| "eval_samples_per_second": 3186.573, | |
| "eval_steps_per_second": 26.423, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 49.8, | |
| "eval_accuracy": 0.13598673300165837, | |
| "eval_f1": 0.09534346865693065, | |
| "eval_loss": 5.305931091308594, | |
| "eval_runtime": 0.3795, | |
| "eval_samples_per_second": 3178.056, | |
| "eval_steps_per_second": 26.352, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 50.86, | |
| "eval_accuracy": 0.13018242122719734, | |
| "eval_f1": 0.08991729329826506, | |
| "eval_loss": 5.287778377532959, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3176.946, | |
| "eval_steps_per_second": 26.343, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 51.92, | |
| "eval_accuracy": 0.1384742951907131, | |
| "eval_f1": 0.0985219901693966, | |
| "eval_loss": 5.267482757568359, | |
| "eval_runtime": 0.378, | |
| "eval_samples_per_second": 3190.643, | |
| "eval_steps_per_second": 26.456, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 52.98, | |
| "eval_accuracy": 0.13764510779436154, | |
| "eval_f1": 0.09503565447256919, | |
| "eval_loss": 5.257174491882324, | |
| "eval_runtime": 0.3791, | |
| "eval_samples_per_second": 3181.268, | |
| "eval_steps_per_second": 26.379, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 53.83, | |
| "eval_accuracy": 0.1417910447761194, | |
| "eval_f1": 0.09976792054154651, | |
| "eval_loss": 5.241861820220947, | |
| "eval_runtime": 0.378, | |
| "eval_samples_per_second": 3190.114, | |
| "eval_steps_per_second": 26.452, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 54.89, | |
| "eval_accuracy": 0.14925373134328357, | |
| "eval_f1": 0.10946077502309857, | |
| "eval_loss": 5.227325439453125, | |
| "eval_runtime": 0.3793, | |
| "eval_samples_per_second": 3179.266, | |
| "eval_steps_per_second": 26.362, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 55.95, | |
| "eval_accuracy": 0.1451077943615257, | |
| "eval_f1": 0.10515805134406893, | |
| "eval_loss": 5.209225654602051, | |
| "eval_runtime": 0.3796, | |
| "eval_samples_per_second": 3177.24, | |
| "eval_steps_per_second": 26.345, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 56.79, | |
| "eval_accuracy": 0.14262023217247097, | |
| "eval_f1": 0.10262149479931104, | |
| "eval_loss": 5.20381498336792, | |
| "eval_runtime": 0.38, | |
| "eval_samples_per_second": 3174.088, | |
| "eval_steps_per_second": 26.319, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 57.85, | |
| "eval_accuracy": 0.14759535655058043, | |
| "eval_f1": 0.10374650435743883, | |
| "eval_loss": 5.192402362823486, | |
| "eval_runtime": 0.3799, | |
| "eval_samples_per_second": 3174.635, | |
| "eval_steps_per_second": 26.324, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 58.91, | |
| "eval_accuracy": 0.148424543946932, | |
| "eval_f1": 0.10865541288671039, | |
| "eval_loss": 5.173834800720215, | |
| "eval_runtime": 0.3803, | |
| "eval_samples_per_second": 3171.494, | |
| "eval_steps_per_second": 26.298, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 59.97, | |
| "eval_accuracy": 0.15008291873963517, | |
| "eval_f1": 0.10904088165373273, | |
| "eval_loss": 5.164543628692627, | |
| "eval_runtime": 0.3779, | |
| "eval_samples_per_second": 3190.923, | |
| "eval_steps_per_second": 26.459, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 60.82, | |
| "eval_accuracy": 0.15008291873963517, | |
| "eval_f1": 0.10954105182677536, | |
| "eval_loss": 5.1523356437683105, | |
| "eval_runtime": 0.3794, | |
| "eval_samples_per_second": 3178.663, | |
| "eval_steps_per_second": 26.357, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 61.88, | |
| "eval_accuracy": 0.15339966832504145, | |
| "eval_f1": 0.11320704863201249, | |
| "eval_loss": 5.140935897827148, | |
| "eval_runtime": 0.3968, | |
| "eval_samples_per_second": 3039.575, | |
| "eval_steps_per_second": 25.204, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 62.94, | |
| "eval_accuracy": 0.15671641791044777, | |
| "eval_f1": 0.11619378649382489, | |
| "eval_loss": 5.13328218460083, | |
| "eval_runtime": 0.3811, | |
| "eval_samples_per_second": 3164.429, | |
| "eval_steps_per_second": 26.239, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 63.58, | |
| "grad_norm": 1.4214905500411987, | |
| "learning_rate": 5.050000000000001e-06, | |
| "loss": 5.3883, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.15754560530679934, | |
| "eval_f1": 0.11639663398572815, | |
| "eval_loss": 5.12091064453125, | |
| "eval_runtime": 0.3803, | |
| "eval_samples_per_second": 3171.109, | |
| "eval_steps_per_second": 26.294, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 64.85, | |
| "eval_accuracy": 0.15754560530679934, | |
| "eval_f1": 0.11683650415743181, | |
| "eval_loss": 5.114450454711914, | |
| "eval_runtime": 0.3804, | |
| "eval_samples_per_second": 3169.982, | |
| "eval_steps_per_second": 26.285, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 65.91, | |
| "eval_accuracy": 0.1550580431177446, | |
| "eval_f1": 0.11358274444416293, | |
| "eval_loss": 5.104104995727539, | |
| "eval_runtime": 0.3928, | |
| "eval_samples_per_second": 3069.891, | |
| "eval_steps_per_second": 25.455, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 66.97, | |
| "eval_accuracy": 0.15671641791044777, | |
| "eval_f1": 0.11432989925366752, | |
| "eval_loss": 5.097550868988037, | |
| "eval_runtime": 0.3806, | |
| "eval_samples_per_second": 3168.505, | |
| "eval_steps_per_second": 26.273, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 67.81, | |
| "eval_accuracy": 0.1583747927031509, | |
| "eval_f1": 0.11862831647923934, | |
| "eval_loss": 5.090635776519775, | |
| "eval_runtime": 0.3895, | |
| "eval_samples_per_second": 3096.27, | |
| "eval_steps_per_second": 25.674, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 68.87, | |
| "eval_accuracy": 0.1625207296849088, | |
| "eval_f1": 0.12161516717070989, | |
| "eval_loss": 5.080664157867432, | |
| "eval_runtime": 0.3789, | |
| "eval_samples_per_second": 3183.124, | |
| "eval_steps_per_second": 26.394, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 69.93, | |
| "eval_accuracy": 0.16169154228855723, | |
| "eval_f1": 0.12018082388719613, | |
| "eval_loss": 5.074178218841553, | |
| "eval_runtime": 0.3782, | |
| "eval_samples_per_second": 3189.028, | |
| "eval_steps_per_second": 26.443, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 70.99, | |
| "eval_accuracy": 0.1625207296849088, | |
| "eval_f1": 0.12052740787172442, | |
| "eval_loss": 5.066323757171631, | |
| "eval_runtime": 0.3784, | |
| "eval_samples_per_second": 3186.978, | |
| "eval_steps_per_second": 26.426, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 71.84, | |
| "eval_accuracy": 0.1625207296849088, | |
| "eval_f1": 0.12160006289622655, | |
| "eval_loss": 5.062046527862549, | |
| "eval_runtime": 0.3801, | |
| "eval_samples_per_second": 3172.618, | |
| "eval_steps_per_second": 26.307, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 72.9, | |
| "eval_accuracy": 0.16417910447761194, | |
| "eval_f1": 0.12152850729201306, | |
| "eval_loss": 5.056005954742432, | |
| "eval_runtime": 0.381, | |
| "eval_samples_per_second": 3165.275, | |
| "eval_steps_per_second": 26.246, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 73.96, | |
| "eval_accuracy": 0.16666666666666666, | |
| "eval_f1": 0.12561182312709163, | |
| "eval_loss": 5.050036430358887, | |
| "eval_runtime": 0.3802, | |
| "eval_samples_per_second": 3171.695, | |
| "eval_steps_per_second": 26.299, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 74.81, | |
| "eval_accuracy": 0.16832504145936983, | |
| "eval_f1": 0.12484080598728817, | |
| "eval_loss": 5.044373035430908, | |
| "eval_runtime": 0.38, | |
| "eval_samples_per_second": 3173.886, | |
| "eval_steps_per_second": 26.317, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 75.87, | |
| "eval_accuracy": 0.17081260364842454, | |
| "eval_f1": 0.12924225508300144, | |
| "eval_loss": 5.041046619415283, | |
| "eval_runtime": 0.3807, | |
| "eval_samples_per_second": 3168.064, | |
| "eval_steps_per_second": 26.269, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 76.93, | |
| "eval_accuracy": 0.16832504145936983, | |
| "eval_f1": 0.1266665967282326, | |
| "eval_loss": 5.036989688873291, | |
| "eval_runtime": 0.3914, | |
| "eval_samples_per_second": 3081.513, | |
| "eval_steps_per_second": 25.552, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 77.99, | |
| "eval_accuracy": 0.17081260364842454, | |
| "eval_f1": 0.12804840525194425, | |
| "eval_loss": 5.0318603515625, | |
| "eval_runtime": 0.3794, | |
| "eval_samples_per_second": 3178.933, | |
| "eval_steps_per_second": 26.359, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 78.83, | |
| "eval_accuracy": 0.16998341625207297, | |
| "eval_f1": 0.1290972468615459, | |
| "eval_loss": 5.03059720993042, | |
| "eval_runtime": 0.3795, | |
| "eval_samples_per_second": 3178.278, | |
| "eval_steps_per_second": 26.354, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 79.89, | |
| "eval_accuracy": 0.17164179104477612, | |
| "eval_f1": 0.12944060169078692, | |
| "eval_loss": 5.027899265289307, | |
| "eval_runtime": 0.3787, | |
| "eval_samples_per_second": 3184.745, | |
| "eval_steps_per_second": 26.408, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 80.95, | |
| "eval_accuracy": 0.1724709784411277, | |
| "eval_f1": 0.13072735697861348, | |
| "eval_loss": 5.026247501373291, | |
| "eval_runtime": 0.3797, | |
| "eval_samples_per_second": 3176.404, | |
| "eval_steps_per_second": 26.338, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 81.8, | |
| "eval_accuracy": 0.16998341625207297, | |
| "eval_f1": 0.12880907562232394, | |
| "eval_loss": 5.024827003479004, | |
| "eval_runtime": 0.3798, | |
| "eval_samples_per_second": 3175.064, | |
| "eval_steps_per_second": 26.327, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 82.86, | |
| "eval_accuracy": 0.17081260364842454, | |
| "eval_f1": 0.12912358374020652, | |
| "eval_loss": 5.0235443115234375, | |
| "eval_runtime": 0.379, | |
| "eval_samples_per_second": 3182.237, | |
| "eval_steps_per_second": 26.387, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 83.92, | |
| "eval_accuracy": 0.17081260364842454, | |
| "eval_f1": 0.12920331329754803, | |
| "eval_loss": 5.022723197937012, | |
| "eval_runtime": 0.3809, | |
| "eval_samples_per_second": 3166.333, | |
| "eval_steps_per_second": 26.255, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 84.77, | |
| "eval_accuracy": 0.17081260364842454, | |
| "eval_f1": 0.12896882646345054, | |
| "eval_loss": 5.022224426269531, | |
| "eval_runtime": 0.3794, | |
| "eval_samples_per_second": 3178.549, | |
| "eval_steps_per_second": 26.356, | |
| "step": 400 | |
| } | |
| ], | |
| "logging_steps": 150, | |
| "max_steps": 400, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 100, | |
| "save_steps": 500, | |
| "total_flos": 2.010721223440829e+16, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |