| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9215929221663578, |
| "eval_steps": 500, |
| "global_step": 90000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.01, |
| "learning_rate": 9.982933464404327e-05, |
| "loss": 1.5665, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 9.965866928808654e-05, |
| "loss": 0.8154, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 9.94880039321298e-05, |
| "loss": 0.5531, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 9.931733857617306e-05, |
| "loss": 0.4455, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 9.914667322021634e-05, |
| "loss": 0.3899, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 9.897600786425961e-05, |
| "loss": 0.3538, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 9.880534250830287e-05, |
| "loss": 0.3273, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 9.863467715234615e-05, |
| "loss": 0.3066, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 9.846401179638941e-05, |
| "loss": 0.2902, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 9.829334644043267e-05, |
| "loss": 0.276, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 9.812268108447595e-05, |
| "loss": 0.2649, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 9.795201572851921e-05, |
| "loss": 0.2548, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 9.778135037256247e-05, |
| "loss": 0.2469, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 9.761068501660574e-05, |
| "loss": 0.2383, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.7440019660649e-05, |
| "loss": 0.2324, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.726935430469227e-05, |
| "loss": 0.2258, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 9.709868894873556e-05, |
| "loss": 0.2205, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 9.692802359277882e-05, |
| "loss": 0.2154, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 9.675735823682208e-05, |
| "loss": 0.2108, |
| "step": 9500 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 9.658669288086535e-05, |
| "loss": 0.2074, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 9.641602752490861e-05, |
| "loss": 0.2038, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 9.624536216895188e-05, |
| "loss": 0.1999, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.607469681299515e-05, |
| "loss": 0.1961, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.590403145703841e-05, |
| "loss": 0.1937, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.573336610108167e-05, |
| "loss": 0.1905, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.556270074512495e-05, |
| "loss": 0.1886, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 9.539203538916821e-05, |
| "loss": 0.1845, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 9.522137003321148e-05, |
| "loss": 0.1835, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 9.505070467725476e-05, |
| "loss": 0.18, |
| "step": 14500 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 9.488003932129802e-05, |
| "loss": 0.1778, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 9.470937396534128e-05, |
| "loss": 0.1763, |
| "step": 15500 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 9.453870860938456e-05, |
| "loss": 0.1748, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 9.436804325342782e-05, |
| "loss": 0.1723, |
| "step": 16500 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 9.419737789747108e-05, |
| "loss": 0.1704, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 9.402671254151435e-05, |
| "loss": 0.1687, |
| "step": 17500 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 9.385604718555762e-05, |
| "loss": 0.1677, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 9.368538182960088e-05, |
| "loss": 0.1658, |
| "step": 18500 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 9.351471647364415e-05, |
| "loss": 0.1639, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 9.334405111768743e-05, |
| "loss": 0.1626, |
| "step": 19500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 9.317338576173069e-05, |
| "loss": 0.1619, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 9.300272040577396e-05, |
| "loss": 0.1594, |
| "step": 20500 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 9.283205504981723e-05, |
| "loss": 0.1584, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 9.266138969386049e-05, |
| "loss": 0.1577, |
| "step": 21500 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 9.249072433790376e-05, |
| "loss": 0.1564, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 9.232005898194702e-05, |
| "loss": 0.1555, |
| "step": 22500 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 9.214939362599028e-05, |
| "loss": 0.1545, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 9.197872827003356e-05, |
| "loss": 0.153, |
| "step": 23500 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 9.180806291407682e-05, |
| "loss": 0.1516, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 9.163739755812008e-05, |
| "loss": 0.1516, |
| "step": 24500 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 9.146673220216337e-05, |
| "loss": 0.1494, |
| "step": 25000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 9.129606684620663e-05, |
| "loss": 0.1488, |
| "step": 25500 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 9.112540149024989e-05, |
| "loss": 0.148, |
| "step": 26000 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 9.095473613429317e-05, |
| "loss": 0.147, |
| "step": 26500 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.078407077833643e-05, |
| "loss": 0.1458, |
| "step": 27000 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.061340542237969e-05, |
| "loss": 0.1446, |
| "step": 27500 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.044274006642297e-05, |
| "loss": 0.1443, |
| "step": 28000 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.027207471046623e-05, |
| "loss": 0.1437, |
| "step": 28500 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.010140935450949e-05, |
| "loss": 0.1425, |
| "step": 29000 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 8.993074399855276e-05, |
| "loss": 0.1415, |
| "step": 29500 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 8.976007864259602e-05, |
| "loss": 0.1416, |
| "step": 30000 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 8.95894132866393e-05, |
| "loss": 0.1404, |
| "step": 30500 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 8.941874793068257e-05, |
| "loss": 0.1409, |
| "step": 31000 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 8.924808257472584e-05, |
| "loss": 0.1387, |
| "step": 31500 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 8.90774172187691e-05, |
| "loss": 0.1382, |
| "step": 32000 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 8.890675186281237e-05, |
| "loss": 0.1382, |
| "step": 32500 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 8.873608650685563e-05, |
| "loss": 0.1377, |
| "step": 33000 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 8.85654211508989e-05, |
| "loss": 0.1373, |
| "step": 33500 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.839475579494217e-05, |
| "loss": 0.1362, |
| "step": 34000 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.822409043898543e-05, |
| "loss": 0.1358, |
| "step": 34500 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.805342508302869e-05, |
| "loss": 0.1353, |
| "step": 35000 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.788275972707197e-05, |
| "loss": 0.1343, |
| "step": 35500 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.771209437111524e-05, |
| "loss": 0.1338, |
| "step": 36000 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.75414290151585e-05, |
| "loss": 0.1331, |
| "step": 36500 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.737076365920178e-05, |
| "loss": 0.1331, |
| "step": 37000 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.720009830324504e-05, |
| "loss": 0.1329, |
| "step": 37500 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.70294329472883e-05, |
| "loss": 0.1323, |
| "step": 38000 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.685876759133158e-05, |
| "loss": 0.132, |
| "step": 38500 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.668810223537484e-05, |
| "loss": 0.1308, |
| "step": 39000 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.65174368794181e-05, |
| "loss": 0.1305, |
| "step": 39500 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.634677152346137e-05, |
| "loss": 0.1301, |
| "step": 40000 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.617610616750463e-05, |
| "loss": 0.1301, |
| "step": 40500 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 8.60054408115479e-05, |
| "loss": 0.1296, |
| "step": 41000 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 8.583477545559117e-05, |
| "loss": 0.1284, |
| "step": 41500 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 8.566411009963445e-05, |
| "loss": 0.1287, |
| "step": 42000 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 8.54934447436777e-05, |
| "loss": 0.1279, |
| "step": 42500 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 8.532277938772097e-05, |
| "loss": 0.1275, |
| "step": 43000 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 8.515211403176424e-05, |
| "loss": 0.1278, |
| "step": 43500 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 8.49814486758075e-05, |
| "loss": 0.127, |
| "step": 44000 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 8.481078331985078e-05, |
| "loss": 0.1264, |
| "step": 44500 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 8.464011796389404e-05, |
| "loss": 0.126, |
| "step": 45000 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 8.44694526079373e-05, |
| "loss": 0.1257, |
| "step": 45500 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 8.429878725198058e-05, |
| "loss": 0.1248, |
| "step": 46000 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 8.412812189602384e-05, |
| "loss": 0.1246, |
| "step": 46500 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 8.395745654006711e-05, |
| "loss": 0.1244, |
| "step": 47000 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 8.378679118411037e-05, |
| "loss": 0.1241, |
| "step": 47500 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 8.361612582815365e-05, |
| "loss": 0.124, |
| "step": 48000 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 8.344546047219691e-05, |
| "loss": 0.1228, |
| "step": 48500 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 8.327479511624017e-05, |
| "loss": 0.1234, |
| "step": 49000 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 8.310412976028345e-05, |
| "loss": 0.1232, |
| "step": 49500 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 8.293346440432671e-05, |
| "loss": 0.1225, |
| "step": 50000 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 8.276279904836998e-05, |
| "loss": 0.1225, |
| "step": 50500 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 8.259213369241324e-05, |
| "loss": 0.1221, |
| "step": 51000 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 8.24214683364565e-05, |
| "loss": 0.1213, |
| "step": 51500 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 8.225080298049978e-05, |
| "loss": 0.1214, |
| "step": 52000 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 8.208013762454306e-05, |
| "loss": 0.1208, |
| "step": 52500 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 8.190947226858632e-05, |
| "loss": 0.1202, |
| "step": 53000 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.173880691262958e-05, |
| "loss": 0.1199, |
| "step": 53500 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.156814155667285e-05, |
| "loss": 0.1203, |
| "step": 54000 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.139747620071611e-05, |
| "loss": 0.1199, |
| "step": 54500 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.122681084475938e-05, |
| "loss": 0.1196, |
| "step": 55000 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.105614548880265e-05, |
| "loss": 0.1197, |
| "step": 55500 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.088548013284591e-05, |
| "loss": 0.1187, |
| "step": 56000 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.071481477688919e-05, |
| "loss": 0.119, |
| "step": 56500 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.054414942093245e-05, |
| "loss": 0.1185, |
| "step": 57000 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 8.037348406497572e-05, |
| "loss": 0.1185, |
| "step": 57500 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 8.020281870901898e-05, |
| "loss": 0.1183, |
| "step": 58000 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 8.003215335306226e-05, |
| "loss": 0.1178, |
| "step": 58500 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.986148799710552e-05, |
| "loss": 0.1174, |
| "step": 59000 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.969082264114878e-05, |
| "loss": 0.1174, |
| "step": 59500 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.952015728519206e-05, |
| "loss": 0.1169, |
| "step": 60000 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 7.934949192923532e-05, |
| "loss": 0.1171, |
| "step": 60500 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 7.917882657327858e-05, |
| "loss": 0.1164, |
| "step": 61000 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 7.900816121732185e-05, |
| "loss": 0.1164, |
| "step": 61500 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 7.883749586136512e-05, |
| "loss": 0.1153, |
| "step": 62000 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 7.866683050540839e-05, |
| "loss": 0.1154, |
| "step": 62500 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 7.849616514945167e-05, |
| "loss": 0.116, |
| "step": 63000 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 7.832549979349493e-05, |
| "loss": 0.1151, |
| "step": 63500 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 7.815483443753819e-05, |
| "loss": 0.1153, |
| "step": 64000 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 7.798416908158146e-05, |
| "loss": 0.1149, |
| "step": 64500 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 7.781350372562472e-05, |
| "loss": 0.1141, |
| "step": 65000 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 7.764283836966799e-05, |
| "loss": 0.1142, |
| "step": 65500 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 7.747217301371126e-05, |
| "loss": 0.1145, |
| "step": 66000 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 7.730150765775452e-05, |
| "loss": 0.1143, |
| "step": 66500 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 7.713084230179778e-05, |
| "loss": 0.1144, |
| "step": 67000 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 7.696017694584106e-05, |
| "loss": 0.114, |
| "step": 67500 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 7.678951158988432e-05, |
| "loss": 0.1135, |
| "step": 68000 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 7.66188462339276e-05, |
| "loss": 0.1134, |
| "step": 68500 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 7.644818087797087e-05, |
| "loss": 0.113, |
| "step": 69000 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 7.627751552201413e-05, |
| "loss": 0.1131, |
| "step": 69500 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 7.610685016605739e-05, |
| "loss": 0.1128, |
| "step": 70000 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 7.593618481010067e-05, |
| "loss": 0.1124, |
| "step": 70500 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 7.576551945414393e-05, |
| "loss": 0.1126, |
| "step": 71000 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 7.559485409818719e-05, |
| "loss": 0.113, |
| "step": 71500 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 7.542418874223046e-05, |
| "loss": 0.1115, |
| "step": 72000 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 7.525352338627373e-05, |
| "loss": 0.1123, |
| "step": 72500 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 7.508285803031699e-05, |
| "loss": 0.1118, |
| "step": 73000 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 7.491219267436026e-05, |
| "loss": 0.1109, |
| "step": 73500 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 7.474152731840354e-05, |
| "loss": 0.1111, |
| "step": 74000 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 7.45708619624468e-05, |
| "loss": 0.1105, |
| "step": 74500 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 7.440019660649007e-05, |
| "loss": 0.1113, |
| "step": 75000 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 7.422953125053333e-05, |
| "loss": 0.1109, |
| "step": 75500 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 7.40588658945766e-05, |
| "loss": 0.1105, |
| "step": 76000 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 7.388820053861987e-05, |
| "loss": 0.1105, |
| "step": 76500 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 7.371753518266313e-05, |
| "loss": 0.1103, |
| "step": 77000 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 7.35468698267064e-05, |
| "loss": 0.1101, |
| "step": 77500 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 7.337620447074967e-05, |
| "loss": 0.1104, |
| "step": 78000 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 7.320553911479293e-05, |
| "loss": 0.1099, |
| "step": 78500 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 7.303487375883619e-05, |
| "loss": 0.1096, |
| "step": 79000 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 7.286420840287948e-05, |
| "loss": 0.1096, |
| "step": 79500 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 7.269354304692274e-05, |
| "loss": 0.1094, |
| "step": 80000 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 7.2522877690966e-05, |
| "loss": 0.1087, |
| "step": 80500 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 7.235221233500928e-05, |
| "loss": 0.1092, |
| "step": 81000 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 7.218154697905254e-05, |
| "loss": 0.1089, |
| "step": 81500 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 7.20108816230958e-05, |
| "loss": 0.1086, |
| "step": 82000 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 7.184021626713907e-05, |
| "loss": 0.1085, |
| "step": 82500 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 7.166955091118234e-05, |
| "loss": 0.1079, |
| "step": 83000 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 7.14988855552256e-05, |
| "loss": 0.1077, |
| "step": 83500 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 7.132822019926887e-05, |
| "loss": 0.1082, |
| "step": 84000 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 7.115755484331213e-05, |
| "loss": 0.1081, |
| "step": 84500 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 7.098688948735541e-05, |
| "loss": 0.1078, |
| "step": 85000 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.081622413139868e-05, |
| "loss": 0.1078, |
| "step": 85500 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.064555877544194e-05, |
| "loss": 0.108, |
| "step": 86000 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 7.04748934194852e-05, |
| "loss": 0.1067, |
| "step": 86500 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 7.030422806352848e-05, |
| "loss": 0.1072, |
| "step": 87000 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 7.013356270757174e-05, |
| "loss": 0.1073, |
| "step": 87500 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 6.9962897351615e-05, |
| "loss": 0.1067, |
| "step": 88000 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.979223199565828e-05, |
| "loss": 0.1071, |
| "step": 88500 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.962156663970154e-05, |
| "loss": 0.1066, |
| "step": 89000 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.94509012837448e-05, |
| "loss": 0.1067, |
| "step": 89500 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.928023592778808e-05, |
| "loss": 0.1062, |
| "step": 90000 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 292971, |
| "num_train_epochs": 3, |
| "save_steps": 10000, |
| "total_flos": 1.20940672057344e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|