| { | |
| "best_metric": 0.752547554347826, | |
| "best_model_checkpoint": "whisper-base-ft-common-language-id/checkpoint-6940", | |
| "epoch": 10.0, | |
| "global_step": 6940, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.602305475504323e-07, | |
| "loss": 3.8186, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.204610951008646e-07, | |
| "loss": 3.819, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.0806916426512968e-06, | |
| "loss": 3.8193, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.4409221902017292e-06, | |
| "loss": 3.8144, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8011527377521614e-06, | |
| "loss": 3.8163, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.1613832853025936e-06, | |
| "loss": 3.8082, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.521613832853026e-06, | |
| "loss": 3.8033, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.8818443804034585e-06, | |
| "loss": 3.7872, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.242074927953891e-06, | |
| "loss": 3.7689, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.602305475504323e-06, | |
| "loss": 3.7204, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.962536023054755e-06, | |
| "loss": 3.6704, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.322766570605187e-06, | |
| "loss": 3.5733, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.68299711815562e-06, | |
| "loss": 3.4915, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.043227665706052e-06, | |
| "loss": 3.4345, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.403458213256485e-06, | |
| "loss": 3.3252, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.763688760806917e-06, | |
| "loss": 3.2767, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.123919308357349e-06, | |
| "loss": 3.2485, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.484149855907782e-06, | |
| "loss": 3.1341, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.844380403458214e-06, | |
| "loss": 3.0653, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.190201729106629e-06, | |
| "loss": 3.0425, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.550432276657062e-06, | |
| "loss": 2.9207, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.910662824207493e-06, | |
| "loss": 2.8703, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.270893371757926e-06, | |
| "loss": 2.836, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.631123919308358e-06, | |
| "loss": 2.7754, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.991354466858791e-06, | |
| "loss": 2.6743, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.351585014409222e-06, | |
| "loss": 2.625, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.711815561959655e-06, | |
| "loss": 2.5291, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.48063858695652173, | |
| "eval_loss": 2.4787049293518066, | |
| "eval_runtime": 144.8732, | |
| "eval_samples_per_second": 40.642, | |
| "eval_steps_per_second": 2.54, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.991994876721104e-06, | |
| "loss": 2.5502, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.951969260326609e-06, | |
| "loss": 2.4174, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.911943643932117e-06, | |
| "loss": 2.3037, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.871918027537624e-06, | |
| "loss": 2.257, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.831892411143133e-06, | |
| "loss": 2.2217, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.79186679474864e-06, | |
| "loss": 2.1369, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.751841178354148e-06, | |
| "loss": 2.0743, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.711815561959655e-06, | |
| "loss": 2.09, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.671789945565162e-06, | |
| "loss": 2.0694, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.631764329170669e-06, | |
| "loss": 2.0017, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.591738712776178e-06, | |
| "loss": 1.9875, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.551713096381685e-06, | |
| "loss": 1.9366, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.511687479987193e-06, | |
| "loss": 1.8543, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.4716618635927e-06, | |
| "loss": 1.8858, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.431636247198209e-06, | |
| "loss": 1.8539, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.391610630803715e-06, | |
| "loss": 1.8339, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.351585014409222e-06, | |
| "loss": 1.8022, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.31155939801473e-06, | |
| "loss": 1.7081, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.271533781620238e-06, | |
| "loss": 1.7609, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.233109189881525e-06, | |
| "loss": 1.7326, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.193083573487034e-06, | |
| "loss": 1.7605, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.15305795709254e-06, | |
| "loss": 1.7325, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.113032340698047e-06, | |
| "loss": 1.6518, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.073006724303554e-06, | |
| "loss": 1.5706, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.032981107909063e-06, | |
| "loss": 1.6398, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.99295549151457e-06, | |
| "loss": 1.58, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 8.952929875120078e-06, | |
| "loss": 1.4585, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.912904258725585e-06, | |
| "loss": 1.5801, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6260190217391305, | |
| "eval_loss": 1.6258126497268677, | |
| "eval_runtime": 142.9849, | |
| "eval_samples_per_second": 41.179, | |
| "eval_steps_per_second": 2.574, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.872878642331094e-06, | |
| "loss": 1.4591, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.832853025936599e-06, | |
| "loss": 1.3388, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.792827409542108e-06, | |
| "loss": 1.2865, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.752801793147614e-06, | |
| "loss": 1.3659, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.712776176753123e-06, | |
| "loss": 1.3075, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.67275056035863e-06, | |
| "loss": 1.2352, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.632724943964139e-06, | |
| "loss": 1.299, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.592699327569645e-06, | |
| "loss": 1.2533, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.552673711175152e-06, | |
| "loss": 1.2684, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 8.51264809478066e-06, | |
| "loss": 1.2821, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.472622478386168e-06, | |
| "loss": 1.165, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 8.432596861991675e-06, | |
| "loss": 1.2305, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 8.392571245597183e-06, | |
| "loss": 1.1679, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.35254562920269e-06, | |
| "loss": 1.135, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.312520012808199e-06, | |
| "loss": 1.134, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 8.272494396413706e-06, | |
| "loss": 1.111, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 8.232468780019213e-06, | |
| "loss": 1.0913, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 8.192443163624721e-06, | |
| "loss": 1.0577, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 8.152417547230228e-06, | |
| "loss": 1.0247, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 8.112391930835735e-06, | |
| "loss": 1.0808, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 8.072366314441244e-06, | |
| "loss": 1.0829, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 8.03234069804675e-06, | |
| "loss": 1.0763, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.992315081652257e-06, | |
| "loss": 1.1263, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.952289465257766e-06, | |
| "loss": 1.13, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.912263848863273e-06, | |
| "loss": 1.0864, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.872238232468781e-06, | |
| "loss": 0.9774, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.832212616074288e-06, | |
| "loss": 1.0433, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.792186999679795e-06, | |
| "loss": 1.0144, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6815557065217391, | |
| "eval_loss": 1.288601279258728, | |
| "eval_runtime": 143.9643, | |
| "eval_samples_per_second": 40.899, | |
| "eval_steps_per_second": 2.556, | |
| "step": 2082 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.752161383285304e-06, | |
| "loss": 0.8903, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.71213576689081e-06, | |
| "loss": 0.8757, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.672110150496318e-06, | |
| "loss": 0.8285, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 7.632084534101826e-06, | |
| "loss": 0.7831, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 7.592058917707333e-06, | |
| "loss": 0.8158, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 7.552033301312841e-06, | |
| "loss": 0.8271, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.512007684918349e-06, | |
| "loss": 0.8424, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 7.4719820685238555e-06, | |
| "loss": 0.7962, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 7.431956452129363e-06, | |
| "loss": 0.7954, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 7.391930835734871e-06, | |
| "loss": 0.7962, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 7.351905219340379e-06, | |
| "loss": 0.8528, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 7.311879602945886e-06, | |
| "loss": 0.7725, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 7.271853986551393e-06, | |
| "loss": 0.8015, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 7.231828370156901e-06, | |
| "loss": 0.6676, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.191802753762408e-06, | |
| "loss": 0.7208, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 7.151777137367916e-06, | |
| "loss": 0.8135, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 7.1117515209734235e-06, | |
| "loss": 0.8313, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 7.071725904578931e-06, | |
| "loss": 0.7266, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 7.031700288184439e-06, | |
| "loss": 0.7825, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 6.991674671789947e-06, | |
| "loss": 0.7119, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.951649055395454e-06, | |
| "loss": 0.8349, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 6.9116234390009606e-06, | |
| "loss": 0.6898, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 6.871597822606468e-06, | |
| "loss": 0.7068, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 6.831572206211976e-06, | |
| "loss": 0.7628, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.791546589817484e-06, | |
| "loss": 0.6544, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 6.7515209734229915e-06, | |
| "loss": 0.7577, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 6.711495357028499e-06, | |
| "loss": 0.7262, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.671469740634007e-06, | |
| "loss": 0.7442, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7236752717391305, | |
| "eval_loss": 1.0782954692840576, | |
| "eval_runtime": 142.9904, | |
| "eval_samples_per_second": 41.178, | |
| "eval_steps_per_second": 2.574, | |
| "step": 2776 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.631444124239513e-06, | |
| "loss": 0.5559, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 6.591418507845021e-06, | |
| "loss": 0.5587, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 6.551392891450529e-06, | |
| "loss": 0.5495, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.511367275056036e-06, | |
| "loss": 0.519, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 6.471341658661544e-06, | |
| "loss": 0.5675, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 6.431316042267052e-06, | |
| "loss": 0.509, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 6.3912904258725596e-06, | |
| "loss": 0.5276, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 6.351264809478066e-06, | |
| "loss": 0.5094, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 6.311239193083573e-06, | |
| "loss": 0.5092, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.271213576689081e-06, | |
| "loss": 0.5025, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 6.231187960294589e-06, | |
| "loss": 0.5393, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 6.191162343900097e-06, | |
| "loss": 0.5807, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 6.151136727505604e-06, | |
| "loss": 0.5582, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.5373, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 6.071085494716619e-06, | |
| "loss": 0.5123, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 6.031059878322127e-06, | |
| "loss": 0.5087, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 5.992635286583414e-06, | |
| "loss": 0.5189, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 5.9526096701889215e-06, | |
| "loss": 0.5056, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 5.912584053794429e-06, | |
| "loss": 0.5393, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 5.872558437399937e-06, | |
| "loss": 0.5533, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 5.832532821005445e-06, | |
| "loss": 0.535, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 5.794108229266732e-06, | |
| "loss": 0.618, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 5.754082612872239e-06, | |
| "loss": 0.4511, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.7140569964777456e-06, | |
| "loss": 0.49, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.674031380083253e-06, | |
| "loss": 0.5176, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.634005763688761e-06, | |
| "loss": 0.4858, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 5.593980147294269e-06, | |
| "loss": 0.4802, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7265625, | |
| "eval_loss": 1.0582159757614136, | |
| "eval_runtime": 141.9251, | |
| "eval_samples_per_second": 41.487, | |
| "eval_steps_per_second": 2.593, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.5539545308997765e-06, | |
| "loss": 0.5603, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 5.513928914505284e-06, | |
| "loss": 0.3679, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 5.473903298110791e-06, | |
| "loss": 0.3718, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 5.433877681716299e-06, | |
| "loss": 0.3868, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 5.393852065321807e-06, | |
| "loss": 0.4013, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 5.3538264489273136e-06, | |
| "loss": 0.4519, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 5.313800832532821e-06, | |
| "loss": 0.3731, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 5.273775216138329e-06, | |
| "loss": 0.39, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 5.233749599743837e-06, | |
| "loss": 0.3997, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 5.193723983349344e-06, | |
| "loss": 0.3921, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 5.1536983669548514e-06, | |
| "loss": 0.3662, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 5.113672750560359e-06, | |
| "loss": 0.3706, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 5.073647134165867e-06, | |
| "loss": 0.372, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 5.033621517771375e-06, | |
| "loss": 0.3909, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.993595901376882e-06, | |
| "loss": 0.3774, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 4.953570284982389e-06, | |
| "loss": 0.3634, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 4.913544668587897e-06, | |
| "loss": 0.3733, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 4.873519052193404e-06, | |
| "loss": 0.2952, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.833493435798912e-06, | |
| "loss": 0.3792, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 4.7934678194044195e-06, | |
| "loss": 0.404, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 4.753442203009927e-06, | |
| "loss": 0.3741, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 4.713416586615434e-06, | |
| "loss": 0.3327, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 4.673390970220942e-06, | |
| "loss": 0.3472, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 4.63336535382645e-06, | |
| "loss": 0.3365, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 4.5933397374319565e-06, | |
| "loss": 0.3276, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.553314121037464e-06, | |
| "loss": 0.3571, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 4.513288504642972e-06, | |
| "loss": 0.3283, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 4.47326288824848e-06, | |
| "loss": 0.3378, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7416779891304348, | |
| "eval_loss": 1.017274022102356, | |
| "eval_runtime": 145.4891, | |
| "eval_samples_per_second": 40.47, | |
| "eval_steps_per_second": 2.529, | |
| "step": 4164 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.433237271853987e-06, | |
| "loss": 0.2904, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 4.393211655459494e-06, | |
| "loss": 0.2767, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 4.353186039065002e-06, | |
| "loss": 0.3324, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 4.313160422670509e-06, | |
| "loss": 0.25, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 4.273134806276017e-06, | |
| "loss": 0.2794, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 4.2331091898815245e-06, | |
| "loss": 0.2126, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 4.193083573487031e-06, | |
| "loss": 0.2909, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 4.153057957092539e-06, | |
| "loss": 0.3043, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 4.113032340698047e-06, | |
| "loss": 0.248, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 4.073006724303555e-06, | |
| "loss": 0.2869, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 4.0329811079090615e-06, | |
| "loss": 0.2531, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 3.992955491514569e-06, | |
| "loss": 0.2674, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 3.952929875120077e-06, | |
| "loss": 0.2855, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 3.912904258725585e-06, | |
| "loss": 0.2846, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 3.8728786423310925e-06, | |
| "loss": 0.2807, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 3.832853025936599e-06, | |
| "loss": 0.2874, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 3.7928274095421076e-06, | |
| "loss": 0.2484, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 3.7528017931476145e-06, | |
| "loss": 0.2182, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.7127761767531222e-06, | |
| "loss": 0.2364, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 3.67275056035863e-06, | |
| "loss": 0.2318, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 3.6327249439641373e-06, | |
| "loss": 0.2633, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 3.5926993275696446e-06, | |
| "loss": 0.2379, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 3.5526737111751524e-06, | |
| "loss": 0.2937, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 3.51264809478066e-06, | |
| "loss": 0.2859, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 3.4726224783861674e-06, | |
| "loss": 0.2498, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 3.432596861991675e-06, | |
| "loss": 0.2853, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 3.3925712455971825e-06, | |
| "loss": 0.2845, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 3.35254562920269e-06, | |
| "loss": 0.1941, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.7445652173913043, | |
| "eval_loss": 1.0053629875183105, | |
| "eval_runtime": 144.6986, | |
| "eval_samples_per_second": 40.691, | |
| "eval_steps_per_second": 2.543, | |
| "step": 4858 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.3125200128081976e-06, | |
| "loss": 0.2266, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 3.2724943964137053e-06, | |
| "loss": 0.2138, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 3.2324687800192126e-06, | |
| "loss": 0.2221, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 3.19244316362472e-06, | |
| "loss": 0.2113, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 3.1524175472302277e-06, | |
| "loss": 0.1793, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 3.1123919308357354e-06, | |
| "loss": 0.2251, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 3.0723663144412423e-06, | |
| "loss": 0.1947, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 3.03234069804675e-06, | |
| "loss": 0.1985, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 2.992315081652258e-06, | |
| "loss": 0.2318, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 2.952289465257765e-06, | |
| "loss": 0.1684, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 2.9122638488632725e-06, | |
| "loss": 0.1688, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 2.87223823246878e-06, | |
| "loss": 0.2049, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 2.832212616074288e-06, | |
| "loss": 0.1908, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 2.7921869996797953e-06, | |
| "loss": 0.2174, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 2.7521613832853026e-06, | |
| "loss": 0.1624, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 2.7121357668908103e-06, | |
| "loss": 0.178, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 2.6721101504963177e-06, | |
| "loss": 0.1888, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 2.6320845341018254e-06, | |
| "loss": 0.173, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 2.592058917707333e-06, | |
| "loss": 0.1769, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 2.5520333013128405e-06, | |
| "loss": 0.1988, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 2.512007684918348e-06, | |
| "loss": 0.2147, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 2.4719820685238555e-06, | |
| "loss": 0.2033, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 2.431956452129363e-06, | |
| "loss": 0.1437, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 2.3919308357348706e-06, | |
| "loss": 0.1775, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 2.351905219340378e-06, | |
| "loss": 0.2166, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 2.3118796029458857e-06, | |
| "loss": 0.1861, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 2.271853986551393e-06, | |
| "loss": 0.1677, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.2318283701569007e-06, | |
| "loss": 0.1424, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.7508491847826086, | |
| "eval_loss": 1.021315574645996, | |
| "eval_runtime": 144.0799, | |
| "eval_samples_per_second": 40.866, | |
| "eval_steps_per_second": 2.554, | |
| "step": 5552 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 2.191802753762408e-06, | |
| "loss": 0.1602, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 2.1517771373679154e-06, | |
| "loss": 0.1476, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 2.111751520973423e-06, | |
| "loss": 0.1191, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 2.0717259045789305e-06, | |
| "loss": 0.1326, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 2.031700288184438e-06, | |
| "loss": 0.1524, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 1.9916746717899455e-06, | |
| "loss": 0.1016, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 1.9516490553954533e-06, | |
| "loss": 0.1648, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 1.911623439000961e-06, | |
| "loss": 0.1673, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 1.8715978226064681e-06, | |
| "loss": 0.1429, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 1.8315722062119759e-06, | |
| "loss": 0.1407, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "learning_rate": 1.7915465898174834e-06, | |
| "loss": 0.1412, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 1.751520973422991e-06, | |
| "loss": 0.1647, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.7114953570284985e-06, | |
| "loss": 0.1708, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 1.6714697406340058e-06, | |
| "loss": 0.0913, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 1.6314441242395135e-06, | |
| "loss": 0.1356, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 1.5914185078450208e-06, | |
| "loss": 0.1485, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 1.5513928914505286e-06, | |
| "loss": 0.1975, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 1.511367275056036e-06, | |
| "loss": 0.142, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 1.4713416586615434e-06, | |
| "loss": 0.1636, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 1.431316042267051e-06, | |
| "loss": 0.1419, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 1.3912904258725585e-06, | |
| "loss": 0.0922, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 1.351264809478066e-06, | |
| "loss": 0.1593, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 1.3112391930835736e-06, | |
| "loss": 0.1294, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 1.2712135766890813e-06, | |
| "loss": 0.1151, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 1.2311879602945886e-06, | |
| "loss": 0.1378, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 1.1911623439000962e-06, | |
| "loss": 0.1484, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 1.1511367275056037e-06, | |
| "loss": 0.1242, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.7494904891304348, | |
| "eval_loss": 1.0566785335540771, | |
| "eval_runtime": 142.7018, | |
| "eval_samples_per_second": 41.261, | |
| "eval_steps_per_second": 2.579, | |
| "step": 6246 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.112712135766891e-06, | |
| "loss": 0.1634, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 1.0726865193723985e-06, | |
| "loss": 0.131, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 1.032660902977906e-06, | |
| "loss": 0.1299, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 9.926352865834136e-07, | |
| "loss": 0.151, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 9.526096701889209e-07, | |
| "loss": 0.1339, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 9.125840537944285e-07, | |
| "loss": 0.1128, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 8.72558437399936e-07, | |
| "loss": 0.0972, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 8.325328210054436e-07, | |
| "loss": 0.0865, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 7.925072046109511e-07, | |
| "loss": 0.1107, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 7.524815882164586e-07, | |
| "loss": 0.0914, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 7.124559718219662e-07, | |
| "loss": 0.1116, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 6.724303554274736e-07, | |
| "loss": 0.1026, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 6.324047390329811e-07, | |
| "loss": 0.1231, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 5.923791226384887e-07, | |
| "loss": 0.1046, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 5.523535062439962e-07, | |
| "loss": 0.1142, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 5.123278898495037e-07, | |
| "loss": 0.086, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 9.58, | |
| "learning_rate": 4.723022734550112e-07, | |
| "loss": 0.0921, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 4.322766570605188e-07, | |
| "loss": 0.1157, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 3.9225104066602624e-07, | |
| "loss": 0.1245, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 3.522254242715338e-07, | |
| "loss": 0.113, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 3.121998078770413e-07, | |
| "loss": 0.0821, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 2.7217419148254884e-07, | |
| "loss": 0.1227, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 2.3214857508805637e-07, | |
| "loss": 0.1208, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 1.921229586935639e-07, | |
| "loss": 0.1323, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 1.520973422990714e-07, | |
| "loss": 0.1217, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 1.1207172590457893e-07, | |
| "loss": 0.1162, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 7.204610951008646e-08, | |
| "loss": 0.0977, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 3.2020493115593984e-08, | |
| "loss": 0.1527, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.752547554347826, | |
| "eval_loss": 1.0725239515304565, | |
| "eval_runtime": 143.4336, | |
| "eval_samples_per_second": 41.05, | |
| "eval_steps_per_second": 2.566, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 6940, | |
| "total_flos": 6.380841226895999e+18, | |
| "train_loss": 0.8764074790546461, | |
| "train_runtime": 7685.9678, | |
| "train_samples_per_second": 28.876, | |
| "train_steps_per_second": 0.903 | |
| } | |
| ], | |
| "max_steps": 6940, | |
| "num_train_epochs": 10, | |
| "total_flos": 6.380841226895999e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |