| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 58602, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.982935735981707e-05, | |
| "loss": 3.8869, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9658714719634144e-05, | |
| "loss": 3.5549, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9488072079451213e-05, | |
| "loss": 3.4728, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9317429439268286e-05, | |
| "loss": 3.4133, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9146786799085356e-05, | |
| "loss": 3.3688, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.897614415890243e-05, | |
| "loss": 3.3409, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.88055015187195e-05, | |
| "loss": 3.3136, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.863485887853657e-05, | |
| "loss": 3.2963, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.846421623835364e-05, | |
| "loss": 3.2804, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8293573598170713e-05, | |
| "loss": 3.262, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.8122930957987783e-05, | |
| "loss": 3.25, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7952288317804852e-05, | |
| "loss": 3.2357, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.7781645677621925e-05, | |
| "loss": 3.2248, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7611003037438995e-05, | |
| "loss": 3.2203, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.7440360397256068e-05, | |
| "loss": 3.2102, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.726971775707314e-05, | |
| "loss": 3.204, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.709907511689021e-05, | |
| "loss": 3.1905, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6928432476707283e-05, | |
| "loss": 3.1944, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6757789836524352e-05, | |
| "loss": 3.1832, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.658714719634142e-05, | |
| "loss": 3.1815, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.6416504556158495e-05, | |
| "loss": 3.1743, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6245861915975564e-05, | |
| "loss": 3.1757, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6075219275792637e-05, | |
| "loss": 3.1616, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.5904576635609706e-05, | |
| "loss": 3.1563, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.573393399542678e-05, | |
| "loss": 3.1534, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.556329135524385e-05, | |
| "loss": 3.1499, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.539264871506092e-05, | |
| "loss": 3.1483, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.5222006074877991e-05, | |
| "loss": 3.1494, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.5051363434695064e-05, | |
| "loss": 3.137, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4880720794512133e-05, | |
| "loss": 3.1392, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4710078154329206e-05, | |
| "loss": 3.1353, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.4539435514146276e-05, | |
| "loss": 3.1302, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4368792873963347e-05, | |
| "loss": 3.1261, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4198150233780418e-05, | |
| "loss": 3.1311, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.402750759359749e-05, | |
| "loss": 3.1183, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.3856864953414559e-05, | |
| "loss": 3.12, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.3686222313231632e-05, | |
| "loss": 3.1146, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3515579673048703e-05, | |
| "loss": 3.1177, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.3344937032865774e-05, | |
| "loss": 3.1208, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3174294392682845e-05, | |
| "loss": 3.0989, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.3003651752499915e-05, | |
| "loss": 3.1027, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.2833009112316987e-05, | |
| "loss": 3.1046, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.2662366472134057e-05, | |
| "loss": 3.1001, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.2491723831951128e-05, | |
| "loss": 3.0987, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2321081191768201e-05, | |
| "loss": 3.0972, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.215043855158527e-05, | |
| "loss": 3.0912, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1979795911402343e-05, | |
| "loss": 3.0915, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1809153271219413e-05, | |
| "loss": 3.0942, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.1638510631036484e-05, | |
| "loss": 3.0837, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.1467867990853557e-05, | |
| "loss": 3.0893, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.1297225350670626e-05, | |
| "loss": 3.0833, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.1126582710487698e-05, | |
| "loss": 3.0838, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.0955940070304769e-05, | |
| "loss": 3.0791, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.078529743012184e-05, | |
| "loss": 3.0814, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.0614654789938911e-05, | |
| "loss": 3.0803, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.0444012149755982e-05, | |
| "loss": 3.071, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0273369509573052e-05, | |
| "loss": 3.0789, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0102726869390125e-05, | |
| "loss": 3.0737, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.932084229207196e-06, | |
| "loss": 3.0745, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.761441589024267e-06, | |
| "loss": 3.0739, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.590798948841338e-06, | |
| "loss": 3.0738, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.420156308658408e-06, | |
| "loss": 3.0741, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.249513668475479e-06, | |
| "loss": 3.0734, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.078871028292552e-06, | |
| "loss": 3.0709, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.908228388109623e-06, | |
| "loss": 3.0677, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.737585747926692e-06, | |
| "loss": 3.0731, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.566943107743763e-06, | |
| "loss": 3.0683, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.396300467560835e-06, | |
| "loss": 3.0707, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.225657827377906e-06, | |
| "loss": 3.0672, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.055015187194977e-06, | |
| "loss": 3.0627, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.884372547012048e-06, | |
| "loss": 3.0608, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.71372990682912e-06, | |
| "loss": 3.0594, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.54308726664619e-06, | |
| "loss": 3.0569, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.372444626463261e-06, | |
| "loss": 3.0652, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.201801986280332e-06, | |
| "loss": 3.0634, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 7.031159346097404e-06, | |
| "loss": 3.0606, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.860516705914474e-06, | |
| "loss": 3.0605, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.6898740657315454e-06, | |
| "loss": 3.0526, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.519231425548617e-06, | |
| "loss": 3.0495, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.348588785365688e-06, | |
| "loss": 3.0444, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.177946145182758e-06, | |
| "loss": 3.0488, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.007303504999829e-06, | |
| "loss": 3.0506, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.836660864816901e-06, | |
| "loss": 3.0516, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.6660182246339725e-06, | |
| "loss": 3.0532, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.495375584451043e-06, | |
| "loss": 3.0501, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.324732944268114e-06, | |
| "loss": 3.0473, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.154090304085185e-06, | |
| "loss": 3.0499, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.983447663902256e-06, | |
| "loss": 3.0452, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.8128050237193275e-06, | |
| "loss": 3.0436, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.642162383536399e-06, | |
| "loss": 3.0458, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.47151974335347e-06, | |
| "loss": 3.0481, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.30087710317054e-06, | |
| "loss": 3.0464, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.130234462987612e-06, | |
| "loss": 3.0419, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.9595918228046825e-06, | |
| "loss": 3.0432, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.788949182621754e-06, | |
| "loss": 3.0441, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.618306542438825e-06, | |
| "loss": 3.0409, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.447663902255896e-06, | |
| "loss": 3.0384, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.2770212620729668e-06, | |
| "loss": 3.0385, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.1063786218900384e-06, | |
| "loss": 3.0414, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.935735981707109e-06, | |
| "loss": 3.0386, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.7650933415241803e-06, | |
| "loss": 3.0413, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.594450701341251e-06, | |
| "loss": 3.0401, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.4238080611583226e-06, | |
| "loss": 3.0379, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.253165420975394e-06, | |
| "loss": 3.0379, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.0825227807924646e-06, | |
| "loss": 3.044, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.9118801406095357e-06, | |
| "loss": 3.0332, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.7412375004266067e-06, | |
| "loss": 3.0332, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.5705948602436777e-06, | |
| "loss": 3.0364, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.3999522200607488e-06, | |
| "loss": 3.0381, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.22930957987782e-06, | |
| "loss": 3.0381, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.0586669396948912e-06, | |
| "loss": 3.0401, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.880242995119621e-07, | |
| "loss": 3.035, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.173816593290332e-07, | |
| "loss": 3.04, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.467390191461043e-07, | |
| "loss": 3.0429, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.7609637896317533e-07, | |
| "loss": 3.0402, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.0545373878024642e-07, | |
| "loss": 3.0356, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.48110985973175e-08, | |
| "loss": 3.0419, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 58602, | |
| "total_flos": 0, | |
| "train_runtime": 11941.6971, | |
| "train_samples_per_second": 4.907 | |
| } | |
| ], | |
| "max_steps": 58602, | |
| "num_train_epochs": 3, | |
| "total_flos": 0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |