| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9393939393939394, | |
| "eval_steps": 500, | |
| "global_step": 345, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 2.0642, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2e-05, | |
| "loss": 2.0704, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-05, | |
| "loss": 2.0758, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4e-05, | |
| "loss": 2.0031, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0466, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6e-05, | |
| "loss": 2.027, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7e-05, | |
| "loss": 2.0737, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8e-05, | |
| "loss": 2.1039, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9e-05, | |
| "loss": 2.0212, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001, | |
| "loss": 2.047, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.999780139628657e-05, | |
| "loss": 2.0435, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.99912057785006e-05, | |
| "loss": 2.0305, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.998021372668808e-05, | |
| "loss": 1.9602, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.996482620753565e-05, | |
| "loss": 2.0138, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.994504457428558e-05, | |
| "loss": 2.065, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.992087056661677e-05, | |
| "loss": 2.0051, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.989230631049171e-05, | |
| "loss": 2.008, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.985935431796962e-05, | |
| "loss": 1.9578, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.982201748698542e-05, | |
| "loss": 1.9591, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.978029910109491e-05, | |
| "loss": 1.9495, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.973420282918601e-05, | |
| "loss": 1.9121, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.968373272515612e-05, | |
| "loss": 1.9639, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.962889322755555e-05, | |
| "loss": 1.9431, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.956968915919725e-05, | |
| "loss": 1.9283, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.950612572673255e-05, | |
| "loss": 1.9112, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.943820852019344e-05, | |
| "loss": 1.9942, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.936594351250082e-05, | |
| "loss": 1.9536, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.928933705893924e-05, | |
| "loss": 1.9364, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.920839589659803e-05, | |
| "loss": 1.9485, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.91231271437788e-05, | |
| "loss": 1.9743, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.903353829936943e-05, | |
| "loss": 1.9492, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.893963724218455e-05, | |
| "loss": 1.9612, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.884143223027266e-05, | |
| "loss": 1.8852, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.873893190018995e-05, | |
| "loss": 1.9104, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.863214526624065e-05, | |
| "loss": 1.9762, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.852108171968436e-05, | |
| "loss": 1.9602, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.840575102791013e-05, | |
| "loss": 1.9283, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.828616333357743e-05, | |
| "loss": 1.8863, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.816232915372423e-05, | |
| "loss": 1.9776, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.8034259378842e-05, | |
| "loss": 1.8874, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.790196527191811e-05, | |
| "loss": 1.8779, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.776545846744509e-05, | |
| "loss": 1.9399, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.762475097039767e-05, | |
| "loss": 1.8758, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.747985515517683e-05, | |
| "loss": 1.903, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.733078376452171e-05, | |
| "loss": 1.9438, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.717754990838881e-05, | |
| "loss": 1.885, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.702016706279913e-05, | |
| "loss": 1.9278, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.685864906865303e-05, | |
| "loss": 1.9612, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.669301013051297e-05, | |
| "loss": 1.9352, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.652326481535435e-05, | |
| "loss": 1.9198, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.634942805128433e-05, | |
| "loss": 1.9552, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.617151512622917e-05, | |
| "loss": 1.9767, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.598954168658955e-05, | |
| "loss": 1.9933, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.580352373586467e-05, | |
| "loss": 1.9751, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.561347763324484e-05, | |
| "loss": 1.8376, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.541942009217273e-05, | |
| "loss": 1.9485, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.522136817887353e-05, | |
| "loss": 1.885, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.501933931085416e-05, | |
| "loss": 1.9241, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.481335125537138e-05, | |
| "loss": 1.8931, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.460342212786932e-05, | |
| "loss": 1.9544, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.43895703903864e-05, | |
| "loss": 1.8637, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.417181484993154e-05, | |
| "loss": 1.964, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.395017465683036e-05, | |
| "loss": 1.8605, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.372466930304091e-05, | |
| "loss": 1.9388, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.349531862043952e-05, | |
| "loss": 1.9195, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.32621427790767e-05, | |
| "loss": 1.9328, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.302516228540327e-05, | |
| "loss": 1.8901, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.278439798046697e-05, | |
| "loss": 1.9296, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.253987103807958e-05, | |
| "loss": 1.9282, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.229160296295488e-05, | |
| "loss": 1.9448, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.203961558881731e-05, | |
| "loss": 1.9211, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.178393107648193e-05, | |
| "loss": 1.8658, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.15245719119055e-05, | |
| "loss": 1.9358, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.126156090420888e-05, | |
| "loss": 1.9446, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.099492118367123e-05, | |
| "loss": 1.8403, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.072467619969572e-05, | |
| "loss": 1.9857, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.878, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.017346582226289e-05, | |
| "loss": 1.9484, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.98925489045329e-05, | |
| "loss": 1.8543, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.960812367055646e-05, | |
| "loss": 1.8613, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.93202151338687e-05, | |
| "loss": 1.887, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.902884861434065e-05, | |
| "loss": 1.8897, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.873404973595285e-05, | |
| "loss": 1.93, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.843584442454158e-05, | |
| "loss": 1.951, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.81342589055191e-05, | |
| "loss": 1.9591, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.782931970156707e-05, | |
| "loss": 1.8649, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.752105363030414e-05, | |
| "loss": 1.9752, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.720948780192746e-05, | |
| "loss": 1.9189, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.689464961682852e-05, | |
| "loss": 1.9188, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.657656676318346e-05, | |
| "loss": 1.9376, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.625526721451798e-05, | |
| "loss": 1.8662, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.593077922724733e-05, | |
| "loss": 1.8113, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.560313133819125e-05, | |
| "loss": 1.8575, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.527235236206436e-05, | |
| "loss": 1.8991, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.493847138894209e-05, | |
| "loss": 1.8475, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.46015177817023e-05, | |
| "loss": 1.8795, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.426152117344313e-05, | |
| "loss": 1.8596, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.391851146487675e-05, | |
| "loss": 1.9481, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.357251882169994e-05, | |
| "loss": 1.9213, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.322357367194109e-05, | |
| "loss": 1.9243, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.28717067032843e-05, | |
| "loss": 1.8764, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.251694886037052e-05, | |
| "loss": 1.9728, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.215933134207618e-05, | |
| "loss": 1.884, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.179888559876943e-05, | |
| "loss": 1.9589, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.143564332954425e-05, | |
| "loss": 2.017, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.106963647943274e-05, | |
| "loss": 1.9672, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.070089723659566e-05, | |
| "loss": 1.8711, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.032945802949179e-05, | |
| "loss": 1.9211, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.995535152402591e-05, | |
| "loss": 1.9019, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.957861062067614e-05, | |
| "loss": 1.8887, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.919926845160037e-05, | |
| "loss": 1.9923, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.881735837772274e-05, | |
| "loss": 1.8621, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.843291398579946e-05, | |
| "loss": 1.9584, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.804596908546529e-05, | |
| "loss": 1.943, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.765655770625997e-05, | |
| "loss": 1.8953, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.726471409463572e-05, | |
| "loss": 1.9394, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.687047271094527e-05, | |
| "loss": 1.91, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.64738682264115e-05, | |
| "loss": 1.9356, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.607493552007805e-05, | |
| "loss": 1.9421, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.56737096757421e-05, | |
| "loss": 1.8658, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.527022597886895e-05, | |
| "loss": 1.9158, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.486451991348872e-05, | |
| "loss": 1.9376, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.445662715907591e-05, | |
| "loss": 1.9365, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.40465835874115e-05, | |
| "loss": 1.8633, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.363442525942826e-05, | |
| "loss": 1.8888, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.322018842203941e-05, | |
| "loss": 1.8944, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.280390950495093e-05, | |
| "loss": 1.8644, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.238562511745768e-05, | |
| "loss": 1.9133, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.196537204522401e-05, | |
| "loss": 1.8339, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.154318724704853e-05, | |
| "loss": 1.8625, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.111910785161381e-05, | |
| "loss": 1.8877, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.06931711542212e-05, | |
| "loss": 1.9571, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.026541461351092e-05, | |
| "loss": 1.8926, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.983587584816769e-05, | |
| "loss": 1.8999, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.940459263361249e-05, | |
| "loss": 1.8957, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.897160289868042e-05, | |
| "loss": 1.9096, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.853694472228503e-05, | |
| "loss": 1.8695, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.810065633006956e-05, | |
| "loss": 1.8959, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.766277609104518e-05, | |
| "loss": 1.8242, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.722334251421665e-05, | |
| "loss": 1.8903, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.678239424519575e-05, | |
| "loss": 1.8978, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.633997006280252e-05, | |
| "loss": 1.8537, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.589610887565503e-05, | |
| "loss": 1.842, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.9226, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.500423175001705e-05, | |
| "loss": 1.8413, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.4556294246901e-05, | |
| "loss": 1.8876, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.410707660288155e-05, | |
| "loss": 1.902, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.36566183240219e-05, | |
| "loss": 1.9084, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.320495902549183e-05, | |
| "loss": 1.8314, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.275213842808383e-05, | |
| "loss": 1.9189, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.229819635471972e-05, | |
| "loss": 1.8738, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.184317272694867e-05, | |
| "loss": 1.8302, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.138710756143613e-05, | |
| "loss": 1.91, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.093004096644481e-05, | |
| "loss": 1.8405, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.0472013138307235e-05, | |
| "loss": 1.8767, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.001306435789072e-05, | |
| "loss": 1.8827, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.9553234987055006e-05, | |
| "loss": 1.938, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.909256546510257e-05, | |
| "loss": 1.8674, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.86310963052223e-05, | |
| "loss": 1.9161, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.816886809092651e-05, | |
| "loss": 1.8738, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.770592147248196e-05, | |
| "loss": 1.855, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.7242297163334804e-05, | |
| "loss": 1.895, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.6778035936530184e-05, | |
| "loss": 1.8702, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.631317862112636e-05, | |
| "loss": 1.9038, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.584776609860414e-05, | |
| "loss": 1.8444, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.538183929927152e-05, | |
| "loss": 1.9575, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.4915439198664164e-05, | |
| "loss": 1.9068, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.4448606813941805e-05, | |
| "loss": 1.8192, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.3981383200281e-05, | |
| "loss": 1.8179, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.351380944726465e-05, | |
| "loss": 1.9271, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.3045926675268344e-05, | |
| "loss": 1.8794, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.257777603184407e-05, | |
| "loss": 1.9613, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.210939868810156e-05, | |
| "loss": 1.9117, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.16408358350875e-05, | |
| "loss": 1.8796, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.117212868016303e-05, | |
| "loss": 1.881, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.07033184433798e-05, | |
| "loss": 1.9096, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.023444635385493e-05, | |
| "loss": 1.9094, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.9765553646145086e-05, | |
| "loss": 1.9166, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.9296681556620207e-05, | |
| "loss": 1.8408, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.882787131983698e-05, | |
| "loss": 1.9153, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.835916416491251e-05, | |
| "loss": 1.8751, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.789060131189845e-05, | |
| "loss": 1.9234, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.742222396815593e-05, | |
| "loss": 1.937, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.6954073324731654e-05, | |
| "loss": 1.9361, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.648619055273537e-05, | |
| "loss": 1.8353, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.601861679971901e-05, | |
| "loss": 1.9344, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.5551393186058213e-05, | |
| "loss": 1.788, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.508456080133584e-05, | |
| "loss": 1.8955, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.461816070072851e-05, | |
| "loss": 1.9064, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.415223390139588e-05, | |
| "loss": 1.858, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.368682137887365e-05, | |
| "loss": 1.8601, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.322196406346984e-05, | |
| "loss": 1.9734, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.27577028366652e-05, | |
| "loss": 1.8557, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.229407852751806e-05, | |
| "loss": 1.9431, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.183113190907349e-05, | |
| "loss": 1.9365, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.136890369477772e-05, | |
| "loss": 1.831, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.090743453489744e-05, | |
| "loss": 1.8685, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.0446765012945006e-05, | |
| "loss": 1.8634, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.998693564210929e-05, | |
| "loss": 1.8493, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.952798686169279e-05, | |
| "loss": 1.8875, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.9069959033555195e-05, | |
| "loss": 1.9417, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.861289243856388e-05, | |
| "loss": 1.8595, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.8156827273051365e-05, | |
| "loss": 1.8732, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.770180364528029e-05, | |
| "loss": 1.8268, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.7247861571916185e-05, | |
| "loss": 1.869, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.679504097450816e-05, | |
| "loss": 1.8882, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.6343381675978116e-05, | |
| "loss": 1.849, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.5892923397118474e-05, | |
| "loss": 1.88, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.5443705753099014e-05, | |
| "loss": 1.884, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.499576824998298e-05, | |
| "loss": 1.8501, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.8654, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.410389112434499e-05, | |
| "loss": 1.8698, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.366002993719747e-05, | |
| "loss": 1.8402, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.321760575480427e-05, | |
| "loss": 1.8361, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.277665748578336e-05, | |
| "loss": 1.8392, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.233722390895483e-05, | |
| "loss": 1.9025, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.1899343669930446e-05, | |
| "loss": 1.8904, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.146305527771499e-05, | |
| "loss": 1.9215, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.102839710131958e-05, | |
| "loss": 1.8426, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.0595407366387504e-05, | |
| "loss": 1.9003, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.016412415183233e-05, | |
| "loss": 1.8339, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.9734585386489093e-05, | |
| "loss": 1.8936, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.93068288457788e-05, | |
| "loss": 1.8554, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.8880892148386198e-05, | |
| "loss": 1.9069, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.8456812752951485e-05, | |
| "loss": 1.9439, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.8034627954775994e-05, | |
| "loss": 1.888, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.7614374882542317e-05, | |
| "loss": 1.9194, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.719609049504911e-05, | |
| "loss": 1.895, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.677981157796059e-05, | |
| "loss": 1.8856, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.636557474057173e-05, | |
| "loss": 1.8132, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.5953416412588503e-05, | |
| "loss": 1.7674, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.554337284092411e-05, | |
| "loss": 1.8213, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.5135480086511302e-05, | |
| "loss": 1.9571, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.472977402113107e-05, | |
| "loss": 1.792, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.4326290324257894e-05, | |
| "loss": 1.9039, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.3925064479921984e-05, | |
| "loss": 1.8781, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.3526131773588516e-05, | |
| "loss": 1.7439, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.3129527289054713e-05, | |
| "loss": 1.8181, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.2735285905364306e-05, | |
| "loss": 1.8142, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.234344229374003e-05, | |
| "loss": 1.8818, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.195403091453473e-05, | |
| "loss": 1.8229, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.156708601420053e-05, | |
| "loss": 1.79, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.1182641622277273e-05, | |
| "loss": 1.8767, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.0800731548399637e-05, | |
| "loss": 1.8782, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.042138937932388e-05, | |
| "loss": 1.8134, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.0044648475974094e-05, | |
| "loss": 1.8738, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.9670541970508223e-05, | |
| "loss": 1.8041, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.9299102763404336e-05, | |
| "loss": 1.8396, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.893036352056728e-05, | |
| "loss": 1.9001, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.8564356670455767e-05, | |
| "loss": 1.7747, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.8201114401230585e-05, | |
| "loss": 1.9078, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.7840668657923835e-05, | |
| "loss": 1.9058, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.748305113962948e-05, | |
| "loss": 1.8619, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.7128293296715703e-05, | |
| "loss": 1.9236, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.677642632805892e-05, | |
| "loss": 1.8204, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.6427481178300062e-05, | |
| "loss": 1.8528, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.6081488535123275e-05, | |
| "loss": 1.8695, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.5738478826556886e-05, | |
| "loss": 1.8079, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.539848221829769e-05, | |
| "loss": 1.8897, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.5061528611057918e-05, | |
| "loss": 1.8215, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.4727647637935649e-05, | |
| "loss": 1.8931, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.4396868661808777e-05, | |
| "loss": 1.7798, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.4069220772752683e-05, | |
| "loss": 1.8395, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.3744732785482034e-05, | |
| "loss": 1.7538, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.3423433236816563e-05, | |
| "loss": 1.753, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.3105350383171483e-05, | |
| "loss": 1.9531, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2790512198072558e-05, | |
| "loss": 1.8359, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2478946369695882e-05, | |
| "loss": 1.901, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2170680298432935e-05, | |
| "loss": 1.8663, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1865741094480909e-05, | |
| "loss": 1.8337, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1564155575458413e-05, | |
| "loss": 1.8878, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.126595026404717e-05, | |
| "loss": 1.8277, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.0971151385659356e-05, | |
| "loss": 1.8477, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.067978486613131e-05, | |
| "loss": 1.945, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.0391876329443533e-05, | |
| "loss": 1.7664, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.010745109546713e-05, | |
| "loss": 1.8162, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.826534177737107e-06, | |
| "loss": 1.8063, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.8191, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 9.275323800304286e-06, | |
| "loss": 1.8882, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.005078816328771e-06, | |
| "loss": 1.8194, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 8.738439095791123e-06, | |
| "loss": 1.8051, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 8.475428088094517e-06, | |
| "loss": 1.8454, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 8.216068923518072e-06, | |
| "loss": 1.9187, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.960384411182709e-06, | |
| "loss": 1.8383, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.708397037045129e-06, | |
| "loss": 1.9316, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.460128961920432e-06, | |
| "loss": 1.8221, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.215602019533041e-06, | |
| "loss": 1.8184, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.974837714596733e-06, | |
| "loss": 1.8572, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.737857220923305e-06, | |
| "loss": 1.8122, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.50468137956049e-06, | |
| "loss": 1.8512, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.275330696959108e-06, | |
| "loss": 1.8534, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.049825343169652e-06, | |
| "loss": 1.7876, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.828185150068472e-06, | |
| "loss": 1.9367, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.610429609613616e-06, | |
| "loss": 1.8474, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.3965778721306755e-06, | |
| "loss": 1.8483, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.186648744628636e-06, | |
| "loss": 1.9065, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.980660689145855e-06, | |
| "loss": 1.8586, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.778631821126473e-06, | |
| "loss": 1.8758, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.580579907827287e-06, | |
| "loss": 1.8067, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.386522366755169e-06, | |
| "loss": 1.8208, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.1964762641353294e-06, | |
| "loss": 1.7676, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.010458313410459e-06, | |
| "loss": 1.799, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.8284848737708325e-06, | |
| "loss": 1.7849, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.6505719487156697e-06, | |
| "loss": 1.7941, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 3.476735184645674e-06, | |
| "loss": 1.8647, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.306989869487037e-06, | |
| "loss": 1.8437, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 3.1413509313469812e-06, | |
| "loss": 1.8539, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.9798329372008825e-06, | |
| "loss": 1.8604, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.8224500916111952e-06, | |
| "loss": 1.7929, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.6692162354782944e-06, | |
| "loss": 1.8269, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.520144844823169e-06, | |
| "loss": 1.8311, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.3752490296023387e-06, | |
| "loss": 1.8252, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.2345415325549123e-06, | |
| "loss": 1.853, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.0980347280818935e-06, | |
| "loss": 1.8513, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.9657406211579966e-06, | |
| "loss": 1.918, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.8376708462757796e-06, | |
| "loss": 1.838, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.713836666422569e-06, | |
| "loss": 1.8308, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.594248972089879e-06, | |
| "loss": 1.8581, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.4789182803156332e-06, | |
| "loss": 1.8513, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.3678547337593494e-06, | |
| "loss": 1.8293, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.2610680998100478e-06, | |
| "loss": 1.8001, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.1585677697273312e-06, | |
| "loss": 1.8864, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.060362757815453e-06, | |
| "loss": 1.8015, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 9.664617006305664e-07, | |
| "loss": 1.8198, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 8.768728562211947e-07, | |
| "loss": 1.876, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.916041034019772e-07, | |
| "loss": 1.879, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.106629410607691e-07, | |
| "loss": 1.7792, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 6.340564874991905e-07, | |
| "loss": 1.9045, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 5.617914798065616e-07, | |
| "loss": 1.7875, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.938742732674529e-07, | |
| "loss": 1.7886, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.303108408027667e-07, | |
| "loss": 1.8113, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.7110677244445167e-07, | |
| "loss": 1.8897, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.162672748438844e-07, | |
| "loss": 1.8716, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.657971708139917e-07, | |
| "loss": 1.8585, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.1970089890509527e-07, | |
| "loss": 1.774, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.7798251301458513e-07, | |
| "loss": 1.9125, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1.4064568203037697e-07, | |
| "loss": 1.8269, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.076936895082925e-07, | |
| "loss": 1.815, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.912943338324596e-08, | |
| "loss": 1.8106, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 5.4955425714431353e-08, | |
| "loss": 1.7958, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.517379246436026e-08, | |
| "loss": 1.8876, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.9786273311928062e-08, | |
| "loss": 1.8588, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 8.794221499408561e-09, | |
| "loss": 1.8221, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.198603713432501e-09, | |
| "loss": 1.853, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.0, | |
| "loss": 1.7843, | |
| "step": 345 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 345, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 115, | |
| "total_flos": 7.812884393509257e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |