{ "best_metric": 27.49828990734407, "best_model_checkpoint": "/home/rcj2772/data_sets/Seneca/output_dir/checkpoint-7000", "epoch": 25.00231719876416, "global_step": 7768, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.2000000000000006e-07, "loss": 4.3385, "step": 25 }, { "epoch": 0.01, "learning_rate": 9.200000000000001e-07, "loss": 3.4704, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.42e-06, "loss": 2.4271, "step": 75 }, { "epoch": 0.01, "learning_rate": 1.9200000000000003e-06, "loss": 1.6065, "step": 100 }, { "epoch": 0.02, "learning_rate": 2.42e-06, "loss": 1.1884, "step": 125 }, { "epoch": 0.02, "learning_rate": 2.92e-06, "loss": 0.9446, "step": 150 }, { "epoch": 0.02, "learning_rate": 3.4200000000000007e-06, "loss": 0.7629, "step": 175 }, { "epoch": 0.03, "learning_rate": 3.920000000000001e-06, "loss": 0.7097, "step": 200 }, { "epoch": 0.03, "learning_rate": 4.42e-06, "loss": 0.6508, "step": 225 }, { "epoch": 0.03, "learning_rate": 4.92e-06, "loss": 0.5891, "step": 250 }, { "epoch": 0.04, "learning_rate": 5.420000000000001e-06, "loss": 0.5815, "step": 275 }, { "epoch": 0.04, "learning_rate": 5.92e-06, "loss": 0.5616, "step": 300 }, { "epoch": 1.0, "learning_rate": 6.42e-06, "loss": 0.455, "step": 325 }, { "epoch": 1.01, "learning_rate": 6.92e-06, "loss": 0.4164, "step": 350 }, { "epoch": 1.01, "learning_rate": 7.420000000000001e-06, "loss": 0.4369, "step": 375 }, { "epoch": 1.01, "learning_rate": 7.92e-06, "loss": 0.4078, "step": 400 }, { "epoch": 1.01, "learning_rate": 8.42e-06, "loss": 0.4149, "step": 425 }, { "epoch": 1.02, "learning_rate": 8.920000000000001e-06, "loss": 0.4046, "step": 450 }, { "epoch": 1.02, "learning_rate": 9.42e-06, "loss": 0.3988, "step": 475 }, { "epoch": 1.02, "learning_rate": 9.920000000000002e-06, "loss": 0.3853, "step": 500 }, { "epoch": 1.03, "learning_rate": 9.971106219042378e-06, "loss": 0.3684, "step": 525 }, { "epoch": 1.03, "learning_rate": 9.936708860759493e-06, "loss": 0.375, "step": 550 }, { "epoch": 1.03, "learning_rate": 9.90231150247661e-06, "loss": 0.3839, "step": 575 }, { "epoch": 1.04, "learning_rate": 9.867914144193726e-06, "loss": 0.3818, "step": 600 }, { "epoch": 2.0, "learning_rate": 9.833516785910844e-06, "loss": 0.3041, "step": 625 }, { "epoch": 2.0, "learning_rate": 9.799119427627959e-06, "loss": 0.2169, "step": 650 }, { "epoch": 2.01, "learning_rate": 9.764722069345075e-06, "loss": 0.2362, "step": 675 }, { "epoch": 2.01, "learning_rate": 9.730324711062192e-06, "loss": 0.2139, "step": 700 }, { "epoch": 2.01, "learning_rate": 9.695927352779307e-06, "loss": 0.2241, "step": 725 }, { "epoch": 2.02, "learning_rate": 9.661529994496423e-06, "loss": 0.2259, "step": 750 }, { "epoch": 2.02, "learning_rate": 9.62713263621354e-06, "loss": 0.2239, "step": 775 }, { "epoch": 2.02, "learning_rate": 9.592735277930656e-06, "loss": 0.2189, "step": 800 }, { "epoch": 2.03, "learning_rate": 9.558337919647771e-06, "loss": 0.2327, "step": 825 }, { "epoch": 2.03, "learning_rate": 9.523940561364889e-06, "loss": 0.2268, "step": 850 }, { "epoch": 2.03, "learning_rate": 9.489543203082004e-06, "loss": 0.259, "step": 875 }, { "epoch": 2.04, "learning_rate": 9.45514584479912e-06, "loss": 0.2077, "step": 900 }, { "epoch": 2.04, "learning_rate": 9.420748486516237e-06, "loss": 0.2208, "step": 925 }, { "epoch": 3.0, "learning_rate": 9.386351128233353e-06, "loss": 0.1276, "step": 950 }, { "epoch": 3.01, "learning_rate": 9.351953769950468e-06, "loss": 0.1161, "step": 975 }, { "epoch": 3.01, "learning_rate": 9.317556411667584e-06, "loss": 0.1151, "step": 1000 }, { "epoch": 3.01, "eval_loss": 0.4013126492500305, "eval_runtime": 2274.8036, "eval_samples_per_second": 0.567, "eval_steps_per_second": 0.284, "eval_wer": 31.83259747528139, "step": 1000 }, { "epoch": 3.01, "learning_rate": 9.283159053384701e-06, "loss": 0.1103, "step": 1025 }, { "epoch": 3.02, "learning_rate": 9.248761695101817e-06, "loss": 0.1075, "step": 1050 }, { "epoch": 3.02, "learning_rate": 9.214364336818932e-06, "loss": 0.1121, "step": 1075 }, { "epoch": 3.02, "learning_rate": 9.17996697853605e-06, "loss": 0.1189, "step": 1100 }, { "epoch": 3.03, "learning_rate": 9.145569620253165e-06, "loss": 0.1147, "step": 1125 }, { "epoch": 3.03, "learning_rate": 9.11117226197028e-06, "loss": 0.126, "step": 1150 }, { "epoch": 3.03, "learning_rate": 9.076774903687398e-06, "loss": 0.1177, "step": 1175 }, { "epoch": 3.03, "learning_rate": 9.042377545404513e-06, "loss": 0.1294, "step": 1200 }, { "epoch": 3.04, "learning_rate": 9.00798018712163e-06, "loss": 0.1222, "step": 1225 }, { "epoch": 4.0, "learning_rate": 8.973582828838746e-06, "loss": 0.0985, "step": 1250 }, { "epoch": 4.0, "learning_rate": 8.939185470555862e-06, "loss": 0.0479, "step": 1275 }, { "epoch": 4.01, "learning_rate": 8.90478811227298e-06, "loss": 0.059, "step": 1300 }, { "epoch": 4.01, "learning_rate": 8.870390753990095e-06, "loss": 0.0533, "step": 1325 }, { "epoch": 4.01, "learning_rate": 8.83599339570721e-06, "loss": 0.0564, "step": 1350 }, { "epoch": 4.02, "learning_rate": 8.801596037424328e-06, "loss": 0.0574, "step": 1375 }, { "epoch": 4.02, "learning_rate": 8.767198679141443e-06, "loss": 0.0623, "step": 1400 }, { "epoch": 4.02, "learning_rate": 8.732801320858559e-06, "loss": 0.0628, "step": 1425 }, { "epoch": 4.03, "learning_rate": 8.698403962575674e-06, "loss": 0.0579, "step": 1450 }, { "epoch": 4.03, "learning_rate": 8.66400660429279e-06, "loss": 0.0579, "step": 1475 }, { "epoch": 4.03, "learning_rate": 8.629609246009907e-06, "loss": 0.0613, "step": 1500 }, { "epoch": 4.04, "learning_rate": 8.595211887727023e-06, "loss": 0.0675, "step": 1525 }, { "epoch": 4.04, "learning_rate": 8.56081452944414e-06, "loss": 0.0606, "step": 1550 }, { "epoch": 5.0, "learning_rate": 8.526417171161256e-06, "loss": 0.0299, "step": 1575 }, { "epoch": 5.01, "learning_rate": 8.492019812878371e-06, "loss": 0.0246, "step": 1600 }, { "epoch": 5.01, "learning_rate": 8.457622454595488e-06, "loss": 0.0287, "step": 1625 }, { "epoch": 5.01, "learning_rate": 8.423225096312604e-06, "loss": 0.0307, "step": 1650 }, { "epoch": 5.02, "learning_rate": 8.38882773802972e-06, "loss": 0.0326, "step": 1675 }, { "epoch": 5.02, "learning_rate": 8.354430379746837e-06, "loss": 0.0312, "step": 1700 }, { "epoch": 5.02, "learning_rate": 8.320033021463952e-06, "loss": 0.0326, "step": 1725 }, { "epoch": 5.03, "learning_rate": 8.285635663181068e-06, "loss": 0.0393, "step": 1750 }, { "epoch": 5.03, "learning_rate": 8.251238304898185e-06, "loss": 0.0381, "step": 1775 }, { "epoch": 5.03, "learning_rate": 8.2168409466153e-06, "loss": 0.0297, "step": 1800 }, { "epoch": 5.04, "learning_rate": 8.182443588332416e-06, "loss": 0.0327, "step": 1825 }, { "epoch": 5.04, "learning_rate": 8.148046230049534e-06, "loss": 0.0324, "step": 1850 }, { "epoch": 6.0, "learning_rate": 8.11364887176665e-06, "loss": 0.0233, "step": 1875 }, { "epoch": 6.01, "learning_rate": 8.079251513483765e-06, "loss": 0.0194, "step": 1900 }, { "epoch": 6.01, "learning_rate": 8.04485415520088e-06, "loss": 0.0206, "step": 1925 }, { "epoch": 6.01, "learning_rate": 8.010456796917998e-06, "loss": 0.0165, "step": 1950 }, { "epoch": 6.01, "learning_rate": 7.976059438635113e-06, "loss": 0.0217, "step": 1975 }, { "epoch": 6.02, "learning_rate": 7.941662080352229e-06, "loss": 0.0185, "step": 2000 }, { "epoch": 6.02, "eval_loss": 0.47960373759269714, "eval_runtime": 2219.3816, "eval_samples_per_second": 0.581, "eval_steps_per_second": 0.291, "eval_wer": 30.16603445059387, "step": 2000 }, { "epoch": 6.02, "learning_rate": 7.907264722069346e-06, "loss": 0.0185, "step": 2025 }, { "epoch": 6.02, "learning_rate": 7.872867363786462e-06, "loss": 0.0158, "step": 2050 }, { "epoch": 6.03, "learning_rate": 7.838470005503577e-06, "loss": 0.0196, "step": 2075 }, { "epoch": 6.03, "learning_rate": 7.804072647220694e-06, "loss": 0.0224, "step": 2100 }, { "epoch": 6.03, "learning_rate": 7.76967528893781e-06, "loss": 0.0173, "step": 2125 }, { "epoch": 6.04, "learning_rate": 7.735277930654927e-06, "loss": 0.0208, "step": 2150 }, { "epoch": 7.0, "learning_rate": 7.700880572372043e-06, "loss": 0.018, "step": 2175 }, { "epoch": 7.0, "learning_rate": 7.666483214089158e-06, "loss": 0.0116, "step": 2200 }, { "epoch": 7.01, "learning_rate": 7.632085855806276e-06, "loss": 0.0141, "step": 2225 }, { "epoch": 7.01, "learning_rate": 7.597688497523391e-06, "loss": 0.013, "step": 2250 }, { "epoch": 7.01, "learning_rate": 7.563291139240508e-06, "loss": 0.0151, "step": 2275 }, { "epoch": 7.02, "learning_rate": 7.528893780957623e-06, "loss": 0.0129, "step": 2300 }, { "epoch": 7.02, "learning_rate": 7.49449642267474e-06, "loss": 0.0156, "step": 2325 }, { "epoch": 7.02, "learning_rate": 7.460099064391856e-06, "loss": 0.0122, "step": 2350 }, { "epoch": 7.03, "learning_rate": 7.425701706108971e-06, "loss": 0.0141, "step": 2375 }, { "epoch": 7.03, "learning_rate": 7.391304347826087e-06, "loss": 0.0164, "step": 2400 }, { "epoch": 7.03, "learning_rate": 7.3569069895432036e-06, "loss": 0.0134, "step": 2425 }, { "epoch": 7.04, "learning_rate": 7.322509631260319e-06, "loss": 0.0135, "step": 2450 }, { "epoch": 7.04, "learning_rate": 7.2881122729774356e-06, "loss": 0.0135, "step": 2475 }, { "epoch": 8.0, "learning_rate": 7.253714914694552e-06, "loss": 0.0082, "step": 2500 }, { "epoch": 8.01, "learning_rate": 7.219317556411668e-06, "loss": 0.0083, "step": 2525 }, { "epoch": 8.01, "learning_rate": 7.184920198128784e-06, "loss": 0.0096, "step": 2550 }, { "epoch": 8.01, "learning_rate": 7.1505228398459e-06, "loss": 0.0091, "step": 2575 }, { "epoch": 8.02, "learning_rate": 7.116125481563017e-06, "loss": 0.0136, "step": 2600 }, { "epoch": 8.02, "learning_rate": 7.081728123280132e-06, "loss": 0.0088, "step": 2625 }, { "epoch": 8.02, "learning_rate": 7.047330764997249e-06, "loss": 0.0121, "step": 2650 }, { "epoch": 8.03, "learning_rate": 7.012933406714365e-06, "loss": 0.0084, "step": 2675 }, { "epoch": 8.03, "learning_rate": 6.978536048431482e-06, "loss": 0.0072, "step": 2700 }, { "epoch": 8.03, "learning_rate": 6.944138690148597e-06, "loss": 0.0117, "step": 2725 }, { "epoch": 8.03, "learning_rate": 6.909741331865714e-06, "loss": 0.0062, "step": 2750 }, { "epoch": 8.04, "learning_rate": 6.87534397358283e-06, "loss": 0.0102, "step": 2775 }, { "epoch": 9.0, "learning_rate": 6.840946615299946e-06, "loss": 0.0088, "step": 2800 }, { "epoch": 9.0, "learning_rate": 6.806549257017062e-06, "loss": 0.0071, "step": 2825 }, { "epoch": 9.01, "learning_rate": 6.772151898734178e-06, "loss": 0.0063, "step": 2850 }, { "epoch": 9.01, "learning_rate": 6.737754540451293e-06, "loss": 0.0076, "step": 2875 }, { "epoch": 9.01, "learning_rate": 6.70335718216841e-06, "loss": 0.0094, "step": 2900 }, { "epoch": 9.02, "learning_rate": 6.668959823885526e-06, "loss": 0.0053, "step": 2925 }, { "epoch": 9.02, "learning_rate": 6.6345624656026416e-06, "loss": 0.0078, "step": 2950 }, { "epoch": 9.02, "learning_rate": 6.600165107319758e-06, "loss": 0.0078, "step": 2975 }, { "epoch": 9.03, "learning_rate": 6.565767749036874e-06, "loss": 0.0062, "step": 3000 }, { "epoch": 9.03, "eval_loss": 0.5142830014228821, "eval_runtime": 2235.4063, "eval_samples_per_second": 0.577, "eval_steps_per_second": 0.289, "eval_wer": 30.041664075617184, "step": 3000 }, { "epoch": 9.03, "learning_rate": 6.531370390753991e-06, "loss": 0.0044, "step": 3025 }, { "epoch": 9.03, "learning_rate": 6.496973032471106e-06, "loss": 0.0084, "step": 3050 }, { "epoch": 9.04, "learning_rate": 6.462575674188223e-06, "loss": 0.0068, "step": 3075 }, { "epoch": 9.04, "learning_rate": 6.428178315905339e-06, "loss": 0.008, "step": 3100 }, { "epoch": 10.0, "learning_rate": 6.393780957622455e-06, "loss": 0.0047, "step": 3125 }, { "epoch": 10.01, "learning_rate": 6.359383599339571e-06, "loss": 0.0049, "step": 3150 }, { "epoch": 10.01, "learning_rate": 6.324986241056688e-06, "loss": 0.0048, "step": 3175 }, { "epoch": 10.01, "learning_rate": 6.290588882773804e-06, "loss": 0.0036, "step": 3200 }, { "epoch": 10.02, "learning_rate": 6.25619152449092e-06, "loss": 0.0045, "step": 3225 }, { "epoch": 10.02, "learning_rate": 6.221794166208036e-06, "loss": 0.006, "step": 3250 }, { "epoch": 10.02, "learning_rate": 6.1873968079251525e-06, "loss": 0.0045, "step": 3275 }, { "epoch": 10.03, "learning_rate": 6.152999449642269e-06, "loss": 0.0046, "step": 3300 }, { "epoch": 10.03, "learning_rate": 6.1186020913593845e-06, "loss": 0.0052, "step": 3325 }, { "epoch": 10.03, "learning_rate": 6.0842047330765e-06, "loss": 0.0036, "step": 3350 }, { "epoch": 10.04, "learning_rate": 6.049807374793616e-06, "loss": 0.0064, "step": 3375 }, { "epoch": 10.04, "learning_rate": 6.015410016510732e-06, "loss": 0.0062, "step": 3400 }, { "epoch": 11.0, "learning_rate": 5.9810126582278484e-06, "loss": 0.0058, "step": 3425 }, { "epoch": 11.01, "learning_rate": 5.946615299944965e-06, "loss": 0.0042, "step": 3450 }, { "epoch": 11.01, "learning_rate": 5.9122179416620804e-06, "loss": 0.0038, "step": 3475 }, { "epoch": 11.01, "learning_rate": 5.877820583379197e-06, "loss": 0.0036, "step": 3500 }, { "epoch": 11.01, "learning_rate": 5.843423225096313e-06, "loss": 0.0067, "step": 3525 }, { "epoch": 11.02, "learning_rate": 5.809025866813429e-06, "loss": 0.0063, "step": 3550 }, { "epoch": 11.02, "learning_rate": 5.774628508530545e-06, "loss": 0.0045, "step": 3575 }, { "epoch": 11.02, "learning_rate": 5.740231150247662e-06, "loss": 0.0032, "step": 3600 }, { "epoch": 11.03, "learning_rate": 5.705833791964778e-06, "loss": 0.0046, "step": 3625 }, { "epoch": 11.03, "learning_rate": 5.671436433681894e-06, "loss": 0.005, "step": 3650 }, { "epoch": 11.03, "learning_rate": 5.63703907539901e-06, "loss": 0.004, "step": 3675 }, { "epoch": 11.04, "learning_rate": 5.6026417171161265e-06, "loss": 0.0037, "step": 3700 }, { "epoch": 12.0, "learning_rate": 5.568244358833242e-06, "loss": 0.0024, "step": 3725 }, { "epoch": 12.0, "learning_rate": 5.5338470005503585e-06, "loss": 0.002, "step": 3750 }, { "epoch": 12.01, "learning_rate": 5.499449642267475e-06, "loss": 0.0014, "step": 3775 }, { "epoch": 12.01, "learning_rate": 5.465052283984591e-06, "loss": 0.0039, "step": 3800 }, { "epoch": 12.01, "learning_rate": 5.430654925701706e-06, "loss": 0.0025, "step": 3825 }, { "epoch": 12.02, "learning_rate": 5.3962575674188225e-06, "loss": 0.0026, "step": 3850 }, { "epoch": 12.02, "learning_rate": 5.361860209135938e-06, "loss": 0.0013, "step": 3875 }, { "epoch": 12.02, "learning_rate": 5.3274628508530544e-06, "loss": 0.0027, "step": 3900 }, { "epoch": 12.03, "learning_rate": 5.293065492570171e-06, "loss": 0.0022, "step": 3925 }, { "epoch": 12.03, "learning_rate": 5.258668134287287e-06, "loss": 0.0024, "step": 3950 }, { "epoch": 12.03, "learning_rate": 5.224270776004403e-06, "loss": 0.0034, "step": 3975 }, { "epoch": 12.04, "learning_rate": 5.189873417721519e-06, "loss": 0.0022, "step": 4000 }, { "epoch": 12.04, "eval_loss": 0.5468949675559998, "eval_runtime": 2225.9485, "eval_samples_per_second": 0.579, "eval_steps_per_second": 0.29, "eval_wer": 28.630060319631863, "step": 4000 }, { "epoch": 12.04, "learning_rate": 5.155476059438636e-06, "loss": 0.0028, "step": 4025 }, { "epoch": 13.0, "learning_rate": 5.121078701155751e-06, "loss": 0.0019, "step": 4050 }, { "epoch": 13.01, "learning_rate": 5.086681342872868e-06, "loss": 0.0022, "step": 4075 }, { "epoch": 13.01, "learning_rate": 5.052283984589984e-06, "loss": 0.0015, "step": 4100 }, { "epoch": 13.01, "learning_rate": 5.0178866263071005e-06, "loss": 0.0015, "step": 4125 }, { "epoch": 13.02, "learning_rate": 4.983489268024216e-06, "loss": 0.0027, "step": 4150 }, { "epoch": 13.02, "learning_rate": 4.949091909741332e-06, "loss": 0.0024, "step": 4175 }, { "epoch": 13.02, "learning_rate": 4.914694551458448e-06, "loss": 0.0016, "step": 4200 }, { "epoch": 13.03, "learning_rate": 4.8802971931755645e-06, "loss": 0.0027, "step": 4225 }, { "epoch": 13.03, "learning_rate": 4.845899834892681e-06, "loss": 0.0023, "step": 4250 }, { "epoch": 13.03, "learning_rate": 4.8115024766097965e-06, "loss": 0.0023, "step": 4275 }, { "epoch": 13.03, "learning_rate": 4.777105118326913e-06, "loss": 0.0026, "step": 4300 }, { "epoch": 13.04, "learning_rate": 4.742707760044029e-06, "loss": 0.002, "step": 4325 }, { "epoch": 14.0, "learning_rate": 4.708310401761145e-06, "loss": 0.0009, "step": 4350 }, { "epoch": 14.0, "learning_rate": 4.673913043478261e-06, "loss": 0.0034, "step": 4375 }, { "epoch": 14.01, "learning_rate": 4.639515685195378e-06, "loss": 0.0012, "step": 4400 }, { "epoch": 14.01, "learning_rate": 4.605118326912493e-06, "loss": 0.002, "step": 4425 }, { "epoch": 14.01, "learning_rate": 4.57072096862961e-06, "loss": 0.0007, "step": 4450 }, { "epoch": 14.02, "learning_rate": 4.536323610346725e-06, "loss": 0.0015, "step": 4475 }, { "epoch": 14.02, "learning_rate": 4.501926252063842e-06, "loss": 0.0008, "step": 4500 }, { "epoch": 14.02, "learning_rate": 4.467528893780958e-06, "loss": 0.0019, "step": 4525 }, { "epoch": 14.03, "learning_rate": 4.4331315354980745e-06, "loss": 0.0013, "step": 4550 }, { "epoch": 14.03, "learning_rate": 4.39873417721519e-06, "loss": 0.0017, "step": 4575 }, { "epoch": 14.03, "learning_rate": 4.3643368189323065e-06, "loss": 0.0031, "step": 4600 }, { "epoch": 14.04, "learning_rate": 4.329939460649423e-06, "loss": 0.0014, "step": 4625 }, { "epoch": 14.04, "learning_rate": 4.2955421023665385e-06, "loss": 0.001, "step": 4650 }, { "epoch": 15.0, "learning_rate": 4.261144744083655e-06, "loss": 0.0006, "step": 4675 }, { "epoch": 15.01, "learning_rate": 4.2267473858007705e-06, "loss": 0.0004, "step": 4700 }, { "epoch": 15.01, "learning_rate": 4.192350027517887e-06, "loss": 0.0014, "step": 4725 }, { "epoch": 15.01, "learning_rate": 4.157952669235003e-06, "loss": 0.0009, "step": 4750 }, { "epoch": 15.02, "learning_rate": 4.123555310952119e-06, "loss": 0.0006, "step": 4775 }, { "epoch": 15.02, "learning_rate": 4.089157952669235e-06, "loss": 0.001, "step": 4800 }, { "epoch": 15.02, "learning_rate": 4.054760594386352e-06, "loss": 0.0005, "step": 4825 }, { "epoch": 15.03, "learning_rate": 4.020363236103468e-06, "loss": 0.0005, "step": 4850 }, { "epoch": 15.03, "learning_rate": 3.985965877820584e-06, "loss": 0.0014, "step": 4875 }, { "epoch": 15.03, "learning_rate": 3.951568519537699e-06, "loss": 0.0004, "step": 4900 }, { "epoch": 15.04, "learning_rate": 3.917171161254816e-06, "loss": 0.0005, "step": 4925 }, { "epoch": 15.04, "learning_rate": 3.882773802971932e-06, "loss": 0.0007, "step": 4950 }, { "epoch": 16.0, "learning_rate": 3.8483764446890486e-06, "loss": 0.0005, "step": 4975 }, { "epoch": 16.01, "learning_rate": 3.813979086406164e-06, "loss": 0.0004, "step": 5000 }, { "epoch": 16.01, "eval_loss": 0.569535493850708, "eval_runtime": 2212.0775, "eval_samples_per_second": 0.583, "eval_steps_per_second": 0.292, "eval_wer": 27.952241776008957, "step": 5000 }, { "epoch": 16.01, "learning_rate": 3.7795817281232806e-06, "loss": 0.0006, "step": 5025 }, { "epoch": 16.01, "learning_rate": 3.7451843698403965e-06, "loss": 0.0012, "step": 5050 }, { "epoch": 16.01, "learning_rate": 3.710787011557513e-06, "loss": 0.0005, "step": 5075 }, { "epoch": 16.02, "learning_rate": 3.676389653274629e-06, "loss": 0.0004, "step": 5100 }, { "epoch": 16.02, "learning_rate": 3.6419922949917445e-06, "loss": 0.0007, "step": 5125 }, { "epoch": 16.02, "learning_rate": 3.607594936708861e-06, "loss": 0.0007, "step": 5150 }, { "epoch": 16.03, "learning_rate": 3.573197578425977e-06, "loss": 0.0004, "step": 5175 }, { "epoch": 16.03, "learning_rate": 3.5388002201430934e-06, "loss": 0.0003, "step": 5200 }, { "epoch": 16.03, "learning_rate": 3.5044028618602094e-06, "loss": 0.0004, "step": 5225 }, { "epoch": 16.04, "learning_rate": 3.4700055035773258e-06, "loss": 0.0004, "step": 5250 }, { "epoch": 17.0, "learning_rate": 3.4356081452944418e-06, "loss": 0.0002, "step": 5275 }, { "epoch": 17.0, "learning_rate": 3.4012107870115578e-06, "loss": 0.0002, "step": 5300 }, { "epoch": 17.01, "learning_rate": 3.366813428728674e-06, "loss": 0.0002, "step": 5325 }, { "epoch": 17.01, "learning_rate": 3.33241607044579e-06, "loss": 0.0001, "step": 5350 }, { "epoch": 17.01, "learning_rate": 3.2980187121629057e-06, "loss": 0.0001, "step": 5375 }, { "epoch": 17.02, "learning_rate": 3.263621353880022e-06, "loss": 0.0002, "step": 5400 }, { "epoch": 17.02, "learning_rate": 3.229223995597138e-06, "loss": 0.0001, "step": 5425 }, { "epoch": 17.02, "learning_rate": 3.1948266373142546e-06, "loss": 0.0004, "step": 5450 }, { "epoch": 17.03, "learning_rate": 3.1604292790313706e-06, "loss": 0.0002, "step": 5475 }, { "epoch": 17.03, "learning_rate": 3.126031920748487e-06, "loss": 0.0002, "step": 5500 }, { "epoch": 17.03, "learning_rate": 3.091634562465603e-06, "loss": 0.0002, "step": 5525 }, { "epoch": 17.04, "learning_rate": 3.057237204182719e-06, "loss": 0.0017, "step": 5550 }, { "epoch": 17.04, "learning_rate": 3.0228398458998354e-06, "loss": 0.0002, "step": 5575 }, { "epoch": 18.0, "learning_rate": 2.988442487616951e-06, "loss": 0.0004, "step": 5600 }, { "epoch": 18.01, "learning_rate": 2.9540451293340674e-06, "loss": 0.0001, "step": 5625 }, { "epoch": 18.01, "learning_rate": 2.9196477710511834e-06, "loss": 0.0002, "step": 5650 }, { "epoch": 18.01, "learning_rate": 2.8852504127682994e-06, "loss": 0.0001, "step": 5675 }, { "epoch": 18.02, "learning_rate": 2.850853054485416e-06, "loss": 0.0001, "step": 5700 }, { "epoch": 18.02, "learning_rate": 2.8164556962025318e-06, "loss": 0.0002, "step": 5725 }, { "epoch": 18.02, "learning_rate": 2.782058337919648e-06, "loss": 0.0003, "step": 5750 }, { "epoch": 18.03, "learning_rate": 2.747660979636764e-06, "loss": 0.0001, "step": 5775 }, { "epoch": 18.03, "learning_rate": 2.7132636213538806e-06, "loss": 0.0001, "step": 5800 }, { "epoch": 18.03, "learning_rate": 2.6788662630709966e-06, "loss": 0.0001, "step": 5825 }, { "epoch": 18.03, "learning_rate": 2.644468904788112e-06, "loss": 0.0001, "step": 5850 }, { "epoch": 18.04, "learning_rate": 2.6100715465052286e-06, "loss": 0.0001, "step": 5875 }, { "epoch": 19.0, "learning_rate": 2.5756741882223446e-06, "loss": 0.0001, "step": 5900 }, { "epoch": 19.0, "learning_rate": 2.5412768299394606e-06, "loss": 0.0005, "step": 5925 }, { "epoch": 19.01, "learning_rate": 2.506879471656577e-06, "loss": 0.0002, "step": 5950 }, { "epoch": 19.01, "learning_rate": 2.472482113373693e-06, "loss": 0.0001, "step": 5975 }, { "epoch": 19.01, "learning_rate": 2.4380847550908094e-06, "loss": 0.0001, "step": 6000 }, { "epoch": 19.01, "eval_loss": 0.5891119837760925, "eval_runtime": 2229.5631, "eval_samples_per_second": 0.578, "eval_steps_per_second": 0.289, "eval_wer": 27.52938250108824, "step": 6000 }, { "epoch": 19.02, "learning_rate": 2.4036873968079254e-06, "loss": 0.0001, "step": 6025 }, { "epoch": 19.02, "learning_rate": 2.3692900385250414e-06, "loss": 0.0001, "step": 6050 }, { "epoch": 19.02, "learning_rate": 2.3348926802421574e-06, "loss": 0.0001, "step": 6075 }, { "epoch": 19.03, "learning_rate": 2.300495321959274e-06, "loss": 0.0003, "step": 6100 }, { "epoch": 19.03, "learning_rate": 2.26609796367639e-06, "loss": 0.0001, "step": 6125 }, { "epoch": 19.03, "learning_rate": 2.231700605393506e-06, "loss": 0.0001, "step": 6150 }, { "epoch": 19.04, "learning_rate": 2.1973032471106222e-06, "loss": 0.0001, "step": 6175 }, { "epoch": 19.04, "learning_rate": 2.1629058888277382e-06, "loss": 0.0001, "step": 6200 }, { "epoch": 20.0, "learning_rate": 2.1285085305448542e-06, "loss": 0.0001, "step": 6225 }, { "epoch": 20.01, "learning_rate": 2.0941111722619706e-06, "loss": 0.0004, "step": 6250 }, { "epoch": 20.01, "learning_rate": 2.0597138139790866e-06, "loss": 0.0001, "step": 6275 }, { "epoch": 20.01, "learning_rate": 2.0253164556962026e-06, "loss": 0.0001, "step": 6300 }, { "epoch": 20.02, "learning_rate": 1.990919097413319e-06, "loss": 0.0001, "step": 6325 }, { "epoch": 20.02, "learning_rate": 1.956521739130435e-06, "loss": 0.0001, "step": 6350 }, { "epoch": 20.02, "learning_rate": 1.922124380847551e-06, "loss": 0.0001, "step": 6375 }, { "epoch": 20.03, "learning_rate": 1.887727022564667e-06, "loss": 0.0001, "step": 6400 }, { "epoch": 20.03, "learning_rate": 1.8533296642817832e-06, "loss": 0.0001, "step": 6425 }, { "epoch": 20.03, "learning_rate": 1.8189323059988994e-06, "loss": 0.0001, "step": 6450 }, { "epoch": 20.04, "learning_rate": 1.7845349477160156e-06, "loss": 0.0001, "step": 6475 }, { "epoch": 20.04, "learning_rate": 1.7501375894331316e-06, "loss": 0.0003, "step": 6500 }, { "epoch": 21.0, "learning_rate": 1.7157402311502476e-06, "loss": 0.0001, "step": 6525 }, { "epoch": 21.01, "learning_rate": 1.6813428728673638e-06, "loss": 0.0001, "step": 6550 }, { "epoch": 21.01, "learning_rate": 1.64694551458448e-06, "loss": 0.0001, "step": 6575 }, { "epoch": 21.01, "learning_rate": 1.6125481563015962e-06, "loss": 0.0001, "step": 6600 }, { "epoch": 21.01, "learning_rate": 1.5781507980187122e-06, "loss": 0.0001, "step": 6625 }, { "epoch": 21.02, "learning_rate": 1.5437534397358285e-06, "loss": 0.0001, "step": 6650 }, { "epoch": 21.02, "learning_rate": 1.5093560814529444e-06, "loss": 0.0001, "step": 6675 }, { "epoch": 21.02, "learning_rate": 1.4749587231700607e-06, "loss": 0.0005, "step": 6700 }, { "epoch": 21.03, "learning_rate": 1.4405613648871769e-06, "loss": 0.0001, "step": 6725 }, { "epoch": 21.03, "learning_rate": 1.4061640066042929e-06, "loss": 0.0003, "step": 6750 }, { "epoch": 21.03, "learning_rate": 1.371766648321409e-06, "loss": 0.0001, "step": 6775 }, { "epoch": 21.04, "learning_rate": 1.3373692900385253e-06, "loss": 0.0001, "step": 6800 }, { "epoch": 22.0, "learning_rate": 1.3029719317556413e-06, "loss": 0.0001, "step": 6825 }, { "epoch": 22.0, "learning_rate": 1.2685745734727575e-06, "loss": 0.0001, "step": 6850 }, { "epoch": 22.01, "learning_rate": 1.2341772151898737e-06, "loss": 0.0001, "step": 6875 }, { "epoch": 22.01, "learning_rate": 1.1997798569069897e-06, "loss": 0.0001, "step": 6900 }, { "epoch": 22.01, "learning_rate": 1.1653824986241059e-06, "loss": 0.0001, "step": 6925 }, { "epoch": 22.02, "learning_rate": 1.1309851403412219e-06, "loss": 0.0002, "step": 6950 }, { "epoch": 22.02, "learning_rate": 1.096587782058338e-06, "loss": 0.0001, "step": 6975 }, { "epoch": 22.02, "learning_rate": 1.0621904237754543e-06, "loss": 0.0002, "step": 7000 }, { "epoch": 22.02, "eval_loss": 0.5983160138130188, "eval_runtime": 2191.4681, "eval_samples_per_second": 0.588, "eval_steps_per_second": 0.294, "eval_wer": 27.49828990734407, "step": 7000 }, { "epoch": 22.03, "learning_rate": 1.0277930654925703e-06, "loss": 0.0001, "step": 7025 }, { "epoch": 22.03, "learning_rate": 9.933957072096865e-07, "loss": 0.0001, "step": 7050 }, { "epoch": 22.03, "learning_rate": 9.589983489268025e-07, "loss": 0.0001, "step": 7075 }, { "epoch": 22.04, "learning_rate": 9.246009906439187e-07, "loss": 0.0001, "step": 7100 }, { "epoch": 22.04, "learning_rate": 8.902036323610347e-07, "loss": 0.0001, "step": 7125 }, { "epoch": 23.0, "learning_rate": 8.558062740781509e-07, "loss": 0.0001, "step": 7150 }, { "epoch": 23.01, "learning_rate": 8.214089157952671e-07, "loss": 0.0001, "step": 7175 }, { "epoch": 23.01, "learning_rate": 7.870115575123831e-07, "loss": 0.0001, "step": 7200 }, { "epoch": 23.01, "learning_rate": 7.526141992294993e-07, "loss": 0.0001, "step": 7225 }, { "epoch": 23.02, "learning_rate": 7.182168409466153e-07, "loss": 0.0001, "step": 7250 }, { "epoch": 23.02, "learning_rate": 6.838194826637315e-07, "loss": 0.0001, "step": 7275 }, { "epoch": 23.02, "learning_rate": 6.494221243808475e-07, "loss": 0.0001, "step": 7300 }, { "epoch": 23.03, "learning_rate": 6.150247660979637e-07, "loss": 0.0001, "step": 7325 }, { "epoch": 23.03, "learning_rate": 5.806274078150798e-07, "loss": 0.0001, "step": 7350 }, { "epoch": 23.03, "learning_rate": 5.462300495321959e-07, "loss": 0.0001, "step": 7375 }, { "epoch": 23.03, "learning_rate": 5.118326912493121e-07, "loss": 0.0001, "step": 7400 }, { "epoch": 23.04, "learning_rate": 4.774353329664282e-07, "loss": 0.0001, "step": 7425 }, { "epoch": 24.0, "learning_rate": 4.4303797468354435e-07, "loss": 0.0001, "step": 7450 }, { "epoch": 24.0, "learning_rate": 4.0864061640066045e-07, "loss": 0.0001, "step": 7475 }, { "epoch": 24.01, "learning_rate": 3.7424325811777655e-07, "loss": 0.0001, "step": 7500 }, { "epoch": 24.01, "learning_rate": 3.3984589983489265e-07, "loss": 0.0001, "step": 7525 }, { "epoch": 24.01, "learning_rate": 3.054485415520088e-07, "loss": 0.0001, "step": 7550 }, { "epoch": 24.02, "learning_rate": 2.7105118326912496e-07, "loss": 0.0001, "step": 7575 }, { "epoch": 24.02, "learning_rate": 2.3665382498624108e-07, "loss": 0.0001, "step": 7600 }, { "epoch": 24.02, "learning_rate": 2.0225646670335718e-07, "loss": 0.0001, "step": 7625 }, { "epoch": 24.03, "learning_rate": 1.6785910842047334e-07, "loss": 0.0001, "step": 7650 }, { "epoch": 24.03, "learning_rate": 1.3346175013758944e-07, "loss": 0.0001, "step": 7675 }, { "epoch": 24.03, "learning_rate": 9.906439185470557e-08, "loss": 0.0001, "step": 7700 }, { "epoch": 24.04, "learning_rate": 6.466703357182169e-08, "loss": 0.0001, "step": 7725 }, { "epoch": 24.04, "learning_rate": 3.026967528893781e-08, "loss": 0.0001, "step": 7750 }, { "epoch": 25.0, "step": 7768, "total_flos": 1.2664482317815356e+20, "train_loss": 0.0937204750527475, "train_runtime": 49250.3081, "train_samples_per_second": 2.524, "train_steps_per_second": 0.158 } ], "max_steps": 7768, "num_train_epochs": 9223372036854775807, "total_flos": 1.2664482317815356e+20, "trial_name": null, "trial_params": null }