| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9987389659520807, | |
| "global_step": 594, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 4.8018, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 4.9652, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 4.867, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-07, | |
| "loss": 4.8109, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 4.7636, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 4.7812, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-07, | |
| "loss": 4.8632, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2e-07, | |
| "loss": 4.7721, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 4.8851, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.4e-07, | |
| "loss": 4.7996, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 4.798, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.599999999999999e-07, | |
| "loss": 4.8514, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.2e-07, | |
| "loss": 4.7753, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.799999999999999e-07, | |
| "loss": 4.729, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.4e-07, | |
| "loss": 4.6173, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9e-07, | |
| "loss": 4.799, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 4.8198, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.0200000000000002e-06, | |
| "loss": 4.8884, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.08e-06, | |
| "loss": 4.6426, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.14e-06, | |
| "loss": 4.8667, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 4.8164, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.26e-06, | |
| "loss": 4.752, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.3199999999999999e-06, | |
| "loss": 4.765, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.38e-06, | |
| "loss": 4.8331, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.44e-06, | |
| "loss": 4.6025, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.5e-06, | |
| "loss": 5.0036, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.5599999999999999e-06, | |
| "loss": 4.661, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.62e-06, | |
| "loss": 4.5722, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.62e-06, | |
| "loss": 4.5755, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.68e-06, | |
| "loss": 4.7449, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.74e-06, | |
| "loss": 4.7354, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.8e-06, | |
| "loss": 4.6014, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.86e-06, | |
| "loss": 4.6082, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 4.5321, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.98e-06, | |
| "loss": 4.5979, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.0400000000000004e-06, | |
| "loss": 4.7437, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 4.5519, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.16e-06, | |
| "loss": 4.6875, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.22e-06, | |
| "loss": 4.8599, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.28e-06, | |
| "loss": 4.4819, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.34e-06, | |
| "loss": 4.5278, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 4.773, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.46e-06, | |
| "loss": 4.5056, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.52e-06, | |
| "loss": 4.6429, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.58e-06, | |
| "loss": 4.7512, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.6399999999999997e-06, | |
| "loss": 4.8136, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.7e-06, | |
| "loss": 4.7159, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.76e-06, | |
| "loss": 5.1107, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.82e-06, | |
| "loss": 5.1422, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.82e-06, | |
| "loss": 4.8905, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.88e-06, | |
| "loss": 4.3776, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 4.3734, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3e-06, | |
| "loss": 4.4207, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.06e-06, | |
| "loss": 4.4927, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.1199999999999998e-06, | |
| "loss": 4.3368, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.18e-06, | |
| "loss": 4.5341, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.24e-06, | |
| "loss": 4.3702, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.3e-06, | |
| "loss": 4.6317, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.36e-06, | |
| "loss": 4.4118, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.4200000000000003e-06, | |
| "loss": 4.178, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.48e-06, | |
| "loss": 4.4322, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.54e-06, | |
| "loss": 4.5075, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.6e-06, | |
| "loss": 4.3774, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.66e-06, | |
| "loss": 4.3254, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.72e-06, | |
| "loss": 4.3747, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 4.3197, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 4.2983, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.9e-06, | |
| "loss": 4.2579, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.96e-06, | |
| "loss": 4.3772, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.0200000000000005e-06, | |
| "loss": 4.3717, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.080000000000001e-06, | |
| "loss": 4.439, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.14e-06, | |
| "loss": 4.417, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 4.1931, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.26e-06, | |
| "loss": 4.4158, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.32e-06, | |
| "loss": 4.3812, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3799999999999996e-06, | |
| "loss": 4.3746, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.44e-06, | |
| "loss": 4.3591, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.5e-06, | |
| "loss": 4.3283, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.56e-06, | |
| "loss": 4.235, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.62e-06, | |
| "loss": 4.419, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.68e-06, | |
| "loss": 4.3521, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.74e-06, | |
| "loss": 4.3597, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 4.3871, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.86e-06, | |
| "loss": 4.3936, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.92e-06, | |
| "loss": 4.3229, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 4.2274, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.04e-06, | |
| "loss": 4.5116, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.1e-06, | |
| "loss": 4.3896, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.16e-06, | |
| "loss": 4.3493, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.22e-06, | |
| "loss": 4.4104, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.279999999999999e-06, | |
| "loss": 4.3376, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.34e-06, | |
| "loss": 4.46, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.4e-06, | |
| "loss": 4.324, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.46e-06, | |
| "loss": 4.5704, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.52e-06, | |
| "loss": 4.4629, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.58e-06, | |
| "loss": 4.5151, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.64e-06, | |
| "loss": 4.5121, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.7000000000000005e-06, | |
| "loss": 4.3518, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.76e-06, | |
| "loss": 4.5692, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.82e-06, | |
| "loss": 4.2875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 4.4111, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.940000000000001e-06, | |
| "loss": 4.2284, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6e-06, | |
| "loss": 4.2267, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 4.1831, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.12e-06, | |
| "loss": 4.2093, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.18e-06, | |
| "loss": 4.2145, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.2399999999999995e-06, | |
| "loss": 4.3612, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.3e-06, | |
| "loss": 4.3816, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.36e-06, | |
| "loss": 4.4591, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.42e-06, | |
| "loss": 4.2485, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.48e-06, | |
| "loss": 4.2481, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.54e-06, | |
| "loss": 4.3219, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.6e-06, | |
| "loss": 4.3033, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 4.3363, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.72e-06, | |
| "loss": 4.2968, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.78e-06, | |
| "loss": 4.2906, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.840000000000001e-06, | |
| "loss": 4.3766, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.900000000000001e-06, | |
| "loss": 4.1622, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.96e-06, | |
| "loss": 4.138, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.0200000000000006e-06, | |
| "loss": 4.1448, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.08e-06, | |
| "loss": 4.3726, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.14e-06, | |
| "loss": 4.4618, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.2e-06, | |
| "loss": 4.2318, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.26e-06, | |
| "loss": 4.2868, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.32e-06, | |
| "loss": 4.3029, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 4.12, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.44e-06, | |
| "loss": 4.1758, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.5e-06, | |
| "loss": 4.1404, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 4.2824, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.62e-06, | |
| "loss": 4.3133, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 4.2607, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.74e-06, | |
| "loss": 4.2107, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.8e-06, | |
| "loss": 4.303, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 4.3205, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.92e-06, | |
| "loss": 4.3713, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.98e-06, | |
| "loss": 4.2257, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.040000000000001e-06, | |
| "loss": 4.2389, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.1e-06, | |
| "loss": 4.2416, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.160000000000001e-06, | |
| "loss": 4.2426, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 4.4482, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.28e-06, | |
| "loss": 4.3518, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.340000000000001e-06, | |
| "loss": 4.4048, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 4.4533, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.459999999999999e-06, | |
| "loss": 4.2518, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.52e-06, | |
| "loss": 4.3254, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.58e-06, | |
| "loss": 4.6543, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.64e-06, | |
| "loss": 4.6074, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.7e-06, | |
| "loss": 4.5995, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.759999999999999e-06, | |
| "loss": 4.5498, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.82e-06, | |
| "loss": 4.6043, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.88e-06, | |
| "loss": 4.2354, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.939999999999999e-06, | |
| "loss": 4.1746, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9e-06, | |
| "loss": 4.1494, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.06e-06, | |
| "loss": 4.0917, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.12e-06, | |
| "loss": 4.1261, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.18e-06, | |
| "loss": 4.1239, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.24e-06, | |
| "loss": 4.2792, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.3e-06, | |
| "loss": 4.1201, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.36e-06, | |
| "loss": 4.0795, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.42e-06, | |
| "loss": 4.1985, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.48e-06, | |
| "loss": 4.2337, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.54e-06, | |
| "loss": 4.1048, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 4.1386, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.66e-06, | |
| "loss": 4.1655, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.72e-06, | |
| "loss": 4.208, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 4.3238, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.84e-06, | |
| "loss": 4.166, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.9e-06, | |
| "loss": 4.3923, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 4.2065, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.002e-05, | |
| "loss": 4.2614, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.008e-05, | |
| "loss": 4.1192, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.0140000000000001e-05, | |
| "loss": 4.1279, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.02e-05, | |
| "loss": 4.1447, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.0260000000000002e-05, | |
| "loss": 4.1434, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.032e-05, | |
| "loss": 4.2788, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.0379999999999999e-05, | |
| "loss": 4.3055, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.044e-05, | |
| "loss": 4.2359, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.05e-05, | |
| "loss": 4.1368, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.0559999999999999e-05, | |
| "loss": 4.2644, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.062e-05, | |
| "loss": 4.1369, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.068e-05, | |
| "loss": 4.2046, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.074e-05, | |
| "loss": 4.1961, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.08e-05, | |
| "loss": 4.1818, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.086e-05, | |
| "loss": 4.0751, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.092e-05, | |
| "loss": 4.4194, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.098e-05, | |
| "loss": 4.2226, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.104e-05, | |
| "loss": 4.2893, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.11e-05, | |
| "loss": 4.3173, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.116e-05, | |
| "loss": 4.4718, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.1220000000000001e-05, | |
| "loss": 4.1695, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.128e-05, | |
| "loss": 4.46, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.134e-05, | |
| "loss": 4.3791, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 4.2939, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.146e-05, | |
| "loss": 4.3483, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.152e-05, | |
| "loss": 4.3014, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.1580000000000001e-05, | |
| "loss": 4.1902, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.164e-05, | |
| "loss": 4.5967, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.1700000000000001e-05, | |
| "loss": 4.576, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.1760000000000001e-05, | |
| "loss": 4.4989, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.182e-05, | |
| "loss": 4.6207, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.1880000000000001e-05, | |
| "loss": 4.0552, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.1940000000000001e-05, | |
| "loss": 4.1887, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.2e-05, | |
| "loss": 4.2448, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.2060000000000001e-05, | |
| "loss": 4.1219, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.2120000000000001e-05, | |
| "loss": 4.1766, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.2180000000000002e-05, | |
| "loss": 4.1157, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.224e-05, | |
| "loss": 4.1767, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.2299999999999999e-05, | |
| "loss": 4.1483, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.236e-05, | |
| "loss": 4.1739, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.242e-05, | |
| "loss": 4.2733, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.2479999999999999e-05, | |
| "loss": 4.2609, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.254e-05, | |
| "loss": 4.2429, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.26e-05, | |
| "loss": 4.2089, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.2659999999999999e-05, | |
| "loss": 4.2922, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.272e-05, | |
| "loss": 4.1625, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.278e-05, | |
| "loss": 4.1871, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.284e-05, | |
| "loss": 4.076, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.29e-05, | |
| "loss": 4.1892, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.296e-05, | |
| "loss": 4.1699, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.302e-05, | |
| "loss": 4.2546, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.308e-05, | |
| "loss": 4.2833, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.314e-05, | |
| "loss": 4.2806, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.32e-05, | |
| "loss": 4.1233, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.326e-05, | |
| "loss": 4.2366, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3320000000000001e-05, | |
| "loss": 4.4089, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.338e-05, | |
| "loss": 4.1829, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.344e-05, | |
| "loss": 4.1341, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 4.185, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.356e-05, | |
| "loss": 4.3245, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.362e-05, | |
| "loss": 4.2811, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3680000000000001e-05, | |
| "loss": 4.1588, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.374e-05, | |
| "loss": 4.2389, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 4.2419, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3860000000000001e-05, | |
| "loss": 4.2317, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.392e-05, | |
| "loss": 4.3873, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3980000000000002e-05, | |
| "loss": 4.1508, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4040000000000001e-05, | |
| "loss": 4.1372, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4099999999999999e-05, | |
| "loss": 4.1833, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.416e-05, | |
| "loss": 4.1076, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.422e-05, | |
| "loss": 4.2586, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.428e-05, | |
| "loss": 4.1897, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.434e-05, | |
| "loss": 4.2108, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.44e-05, | |
| "loss": 4.2253, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.446e-05, | |
| "loss": 4.3141, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.452e-05, | |
| "loss": 4.4083, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.458e-05, | |
| "loss": 4.4121, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.464e-05, | |
| "loss": 4.4035, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.47e-05, | |
| "loss": 4.5181, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.4760000000000001e-05, | |
| "loss": 4.4993, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.482e-05, | |
| "loss": 4.5019, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.488e-05, | |
| "loss": 4.2284, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.4940000000000001e-05, | |
| "loss": 4.2068, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.5e-05, | |
| "loss": 4.1909, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.506e-05, | |
| "loss": 4.0845, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.5120000000000001e-05, | |
| "loss": 4.2476, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.518e-05, | |
| "loss": 4.1816, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.524e-05, | |
| "loss": 4.1335, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.53e-05, | |
| "loss": 4.1593, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.5360000000000002e-05, | |
| "loss": 4.2225, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.542e-05, | |
| "loss": 4.2368, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.548e-05, | |
| "loss": 4.189, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.554e-05, | |
| "loss": 4.1339, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.56e-05, | |
| "loss": 4.2449, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.5660000000000003e-05, | |
| "loss": 3.9804, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.5720000000000002e-05, | |
| "loss": 4.069, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.578e-05, | |
| "loss": 4.1806, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.584e-05, | |
| "loss": 4.1926, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.59e-05, | |
| "loss": 4.1246, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.596e-05, | |
| "loss": 4.2415, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.6020000000000002e-05, | |
| "loss": 4.2347, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6080000000000002e-05, | |
| "loss": 4.1455, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.614e-05, | |
| "loss": 4.3091, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.62e-05, | |
| "loss": 4.1648, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.626e-05, | |
| "loss": 4.1732, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6320000000000003e-05, | |
| "loss": 4.2292, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6380000000000002e-05, | |
| "loss": 4.1723, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6440000000000002e-05, | |
| "loss": 4.3017, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.65e-05, | |
| "loss": 4.0869, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.656e-05, | |
| "loss": 4.2827, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6620000000000004e-05, | |
| "loss": 4.3505, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6680000000000003e-05, | |
| "loss": 4.3437, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6740000000000002e-05, | |
| "loss": 4.224, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 4.1334, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.686e-05, | |
| "loss": 4.1048, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.6919999999999997e-05, | |
| "loss": 4.3781, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.698e-05, | |
| "loss": 4.1156, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.704e-05, | |
| "loss": 4.4761, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.71e-05, | |
| "loss": 3.9581, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.716e-05, | |
| "loss": 4.212, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.7219999999999998e-05, | |
| "loss": 4.1816, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.728e-05, | |
| "loss": 4.425, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.734e-05, | |
| "loss": 4.1161, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.74e-05, | |
| "loss": 4.2983, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.746e-05, | |
| "loss": 4.3658, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7519999999999998e-05, | |
| "loss": 4.3488, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.758e-05, | |
| "loss": 4.139, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.764e-05, | |
| "loss": 4.2258, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.77e-05, | |
| "loss": 4.399, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.776e-05, | |
| "loss": 4.4188, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.782e-05, | |
| "loss": 4.6496, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7879999999999998e-05, | |
| "loss": 4.2888, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.794e-05, | |
| "loss": 4.2271, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.8e-05, | |
| "loss": 4.3196, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.806e-05, | |
| "loss": 4.0843, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.812e-05, | |
| "loss": 4.286, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.818e-05, | |
| "loss": 4.2722, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.824e-05, | |
| "loss": 4.0551, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.83e-05, | |
| "loss": 4.2781, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.836e-05, | |
| "loss": 4.1745, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.842e-05, | |
| "loss": 4.2659, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.848e-05, | |
| "loss": 4.1224, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.854e-05, | |
| "loss": 4.1487, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.86e-05, | |
| "loss": 4.1931, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.866e-05, | |
| "loss": 4.1865, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.872e-05, | |
| "loss": 4.2107, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.878e-05, | |
| "loss": 4.2072, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.884e-05, | |
| "loss": 4.147, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 4.2357, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.896e-05, | |
| "loss": 4.0721, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.902e-05, | |
| "loss": 4.2324, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.908e-05, | |
| "loss": 4.0826, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.914e-05, | |
| "loss": 4.115, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 4.15, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.9260000000000002e-05, | |
| "loss": 4.2141, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.932e-05, | |
| "loss": 4.1611, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.938e-05, | |
| "loss": 4.1005, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.944e-05, | |
| "loss": 4.3704, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.95e-05, | |
| "loss": 4.11, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9560000000000002e-05, | |
| "loss": 4.0454, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9620000000000002e-05, | |
| "loss": 4.2552, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.968e-05, | |
| "loss": 4.1445, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.974e-05, | |
| "loss": 4.0739, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.98e-05, | |
| "loss": 4.1624, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9860000000000003e-05, | |
| "loss": 4.2192, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9920000000000002e-05, | |
| "loss": 4.2331, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9980000000000002e-05, | |
| "loss": 4.118, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.004e-05, | |
| "loss": 4.0988, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.01e-05, | |
| "loss": 4.2059, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.016e-05, | |
| "loss": 4.2145, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.0220000000000003e-05, | |
| "loss": 4.2589, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.0280000000000002e-05, | |
| "loss": 4.2607, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0340000000000002e-05, | |
| "loss": 4.3465, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.04e-05, | |
| "loss": 4.3788, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.046e-05, | |
| "loss": 4.2406, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0520000000000003e-05, | |
| "loss": 4.404, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0580000000000003e-05, | |
| "loss": 4.0988, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.064e-05, | |
| "loss": 4.2389, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.07e-05, | |
| "loss": 4.5392, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0759999999999998e-05, | |
| "loss": 4.4285, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.082e-05, | |
| "loss": 4.7097, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.088e-05, | |
| "loss": 4.1047, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.094e-05, | |
| "loss": 4.1978, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1e-05, | |
| "loss": 4.2459, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.1059999999999998e-05, | |
| "loss": 4.1, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.1119999999999998e-05, | |
| "loss": 4.1504, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.118e-05, | |
| "loss": 4.1622, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.124e-05, | |
| "loss": 4.1842, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.13e-05, | |
| "loss": 4.1016, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.136e-05, | |
| "loss": 4.1365, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.1419999999999998e-05, | |
| "loss": 4.0905, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.148e-05, | |
| "loss": 4.1797, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.154e-05, | |
| "loss": 4.1696, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.16e-05, | |
| "loss": 4.2197, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.166e-05, | |
| "loss": 4.1037, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.172e-05, | |
| "loss": 4.1754, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.178e-05, | |
| "loss": 4.1212, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.184e-05, | |
| "loss": 4.1486, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.19e-05, | |
| "loss": 4.1094, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.196e-05, | |
| "loss": 4.2153, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.202e-05, | |
| "loss": 4.121, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.208e-05, | |
| "loss": 4.0395, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.214e-05, | |
| "loss": 4.1833, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.22e-05, | |
| "loss": 4.0221, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.226e-05, | |
| "loss": 4.0563, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.232e-05, | |
| "loss": 4.217, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.238e-05, | |
| "loss": 4.1048, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.2440000000000002e-05, | |
| "loss": 4.1853, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.25e-05, | |
| "loss": 4.2369, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.256e-05, | |
| "loss": 4.0589, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.262e-05, | |
| "loss": 4.2835, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.268e-05, | |
| "loss": 4.2185, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.274e-05, | |
| "loss": 4.0766, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 3.9876, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.286e-05, | |
| "loss": 4.1926, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.292e-05, | |
| "loss": 4.368, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.298e-05, | |
| "loss": 4.1354, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.304e-05, | |
| "loss": 4.2917, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.3100000000000002e-05, | |
| "loss": 4.2033, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.3160000000000002e-05, | |
| "loss": 4.1073, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.322e-05, | |
| "loss": 4.0983, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.328e-05, | |
| "loss": 4.3189, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.334e-05, | |
| "loss": 4.239, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 4.4852, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3460000000000002e-05, | |
| "loss": 4.3024, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3520000000000002e-05, | |
| "loss": 4.2838, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.358e-05, | |
| "loss": 4.2025, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.364e-05, | |
| "loss": 4.3544, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.37e-05, | |
| "loss": 4.5671, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3760000000000003e-05, | |
| "loss": 4.3567, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3820000000000002e-05, | |
| "loss": 4.4614, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3880000000000002e-05, | |
| "loss": 4.2169, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.394e-05, | |
| "loss": 4.1263, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4e-05, | |
| "loss": 4.1705, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4060000000000003e-05, | |
| "loss": 4.1627, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4120000000000003e-05, | |
| "loss": 4.0861, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4180000000000002e-05, | |
| "loss": 4.1348, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4240000000000002e-05, | |
| "loss": 4.1424, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.43e-05, | |
| "loss": 4.0704, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.4360000000000004e-05, | |
| "loss": 4.1222, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.442e-05, | |
| "loss": 4.0907, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.448e-05, | |
| "loss": 4.0812, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.454e-05, | |
| "loss": 4.2638, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 4.1372, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.4659999999999998e-05, | |
| "loss": 4.19, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.472e-05, | |
| "loss": 4.1761, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.478e-05, | |
| "loss": 4.0935, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.484e-05, | |
| "loss": 4.1588, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.49e-05, | |
| "loss": 3.9861, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.4959999999999998e-05, | |
| "loss": 4.2586, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.502e-05, | |
| "loss": 4.1678, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.508e-05, | |
| "loss": 3.9133, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.514e-05, | |
| "loss": 4.0723, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.52e-05, | |
| "loss": 4.2253, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.526e-05, | |
| "loss": 4.1328, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.5319999999999998e-05, | |
| "loss": 4.1467, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.538e-05, | |
| "loss": 4.2043, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.544e-05, | |
| "loss": 4.1373, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.55e-05, | |
| "loss": 4.0594, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.556e-05, | |
| "loss": 4.1956, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.562e-05, | |
| "loss": 4.1423, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.568e-05, | |
| "loss": 4.2437, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.574e-05, | |
| "loss": 4.002, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.58e-05, | |
| "loss": 4.2375, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.586e-05, | |
| "loss": 4.1837, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.592e-05, | |
| "loss": 4.2282, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.5980000000000002e-05, | |
| "loss": 4.1327, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.604e-05, | |
| "loss": 4.0206, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.61e-05, | |
| "loss": 4.2462, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.616e-05, | |
| "loss": 4.1303, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.622e-05, | |
| "loss": 4.3363, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.628e-05, | |
| "loss": 4.2215, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.6340000000000002e-05, | |
| "loss": 4.37, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.64e-05, | |
| "loss": 4.2928, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.646e-05, | |
| "loss": 4.2348, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.652e-05, | |
| "loss": 4.0496, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.658e-05, | |
| "loss": 4.3644, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6640000000000002e-05, | |
| "loss": 4.3019, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6700000000000002e-05, | |
| "loss": 4.2132, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.676e-05, | |
| "loss": 4.3806, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.682e-05, | |
| "loss": 4.4193, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.688e-05, | |
| "loss": 4.2117, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.6940000000000003e-05, | |
| "loss": 4.1936, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 4.1327, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7060000000000002e-05, | |
| "loss": 4.0195, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.712e-05, | |
| "loss": 4.1855, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.718e-05, | |
| "loss": 4.1236, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.724e-05, | |
| "loss": 4.2202, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.7300000000000003e-05, | |
| "loss": 4.237, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.7360000000000002e-05, | |
| "loss": 4.1488, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.7420000000000002e-05, | |
| "loss": 4.1618, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.748e-05, | |
| "loss": 4.1037, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.754e-05, | |
| "loss": 4.093, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.7600000000000003e-05, | |
| "loss": 4.0882, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.7660000000000003e-05, | |
| "loss": 4.1926, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.7720000000000002e-05, | |
| "loss": 4.1991, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.778e-05, | |
| "loss": 4.1288, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.784e-05, | |
| "loss": 4.1492, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.79e-05, | |
| "loss": 4.1312, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.7960000000000003e-05, | |
| "loss": 4.135, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.8020000000000003e-05, | |
| "loss": 4.1085, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.8080000000000002e-05, | |
| "loss": 4.0511, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.8139999999999998e-05, | |
| "loss": 4.1314, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 4.1245, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.826e-05, | |
| "loss": 4.0092, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.832e-05, | |
| "loss": 4.0817, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.838e-05, | |
| "loss": 4.0978, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.844e-05, | |
| "loss": 4.1275, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 4.1422, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.856e-05, | |
| "loss": 4.0273, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.862e-05, | |
| "loss": 4.0166, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.868e-05, | |
| "loss": 4.2866, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.874e-05, | |
| "loss": 4.098, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.88e-05, | |
| "loss": 4.2666, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.8859999999999998e-05, | |
| "loss": 4.1968, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.892e-05, | |
| "loss": 4.19, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.898e-05, | |
| "loss": 4.1458, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.904e-05, | |
| "loss": 4.2218, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.91e-05, | |
| "loss": 4.2325, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.916e-05, | |
| "loss": 4.1593, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.922e-05, | |
| "loss": 4.0819, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.928e-05, | |
| "loss": 4.3156, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.934e-05, | |
| "loss": 4.1272, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.94e-05, | |
| "loss": 3.9766, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.946e-05, | |
| "loss": 4.2039, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.9520000000000002e-05, | |
| "loss": 4.3366, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.958e-05, | |
| "loss": 4.3495, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.964e-05, | |
| "loss": 4.2997, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.97e-05, | |
| "loss": 4.3311, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.976e-05, | |
| "loss": 4.4518, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.982e-05, | |
| "loss": 4.3531, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_loss": 4.199646472930908, | |
| "eval_runtime": 765.3338, | |
| "eval_samples_per_second": 3.452, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 1.9703887346291153, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.9880000000000002e-05, | |
| "loss": 4.2443, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.994e-05, | |
| "loss": 4.0981, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3e-05, | |
| "loss": 4.0662, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.968085106382979e-05, | |
| "loss": 4.205, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.9361702127659574e-05, | |
| "loss": 4.1501, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.9042553191489362e-05, | |
| "loss": 4.2854, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.872340425531915e-05, | |
| "loss": 4.1436, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8404255319148935e-05, | |
| "loss": 4.1215, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8085106382978723e-05, | |
| "loss": 4.0574, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.776595744680851e-05, | |
| "loss": 4.1498, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.74468085106383e-05, | |
| "loss": 4.1981, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.7127659574468088e-05, | |
| "loss": 4.117, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.6808510638297873e-05, | |
| "loss": 4.1104, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.648936170212766e-05, | |
| "loss": 4.0441, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.617021276595745e-05, | |
| "loss": 4.1672, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5851063829787234e-05, | |
| "loss": 4.1282, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5531914893617022e-05, | |
| "loss": 4.1474, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.521276595744681e-05, | |
| "loss": 4.2173, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.4893617021276595e-05, | |
| "loss": 4.0136, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.4574468085106383e-05, | |
| "loss": 4.1264, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.4255319148936168e-05, | |
| "loss": 4.2056, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3936170212765956e-05, | |
| "loss": 4.237, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3617021276595744e-05, | |
| "loss": 4.0193, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3297872340425533e-05, | |
| "loss": 3.9602, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.297872340425532e-05, | |
| "loss": 4.0401, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.265957446808511e-05, | |
| "loss": 4.0647, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.2340425531914894e-05, | |
| "loss": 4.2151, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.2021276595744682e-05, | |
| "loss": 4.0694, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.170212765957447e-05, | |
| "loss": 4.029, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.1382978723404255e-05, | |
| "loss": 4.007, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.1063829787234043e-05, | |
| "loss": 4.0029, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.074468085106383e-05, | |
| "loss": 4.235, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.0425531914893616e-05, | |
| "loss": 4.0651, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.0106382978723404e-05, | |
| "loss": 4.0912, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.978723404255319e-05, | |
| "loss": 4.0929, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.9468085106382977e-05, | |
| "loss": 4.0123, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.914893617021277e-05, | |
| "loss": 4.1711, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8829787234042554e-05, | |
| "loss": 4.1767, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8510638297872342e-05, | |
| "loss": 4.0337, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.819148936170213e-05, | |
| "loss": 4.1235, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.7872340425531915e-05, | |
| "loss": 4.0841, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.7553191489361703e-05, | |
| "loss": 4.1234, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.723404255319149e-05, | |
| "loss": 4.2567, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.6914893617021276e-05, | |
| "loss": 4.1139, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6595744680851064e-05, | |
| "loss": 4.242, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.627659574468085e-05, | |
| "loss": 4.4308, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5957446808510637e-05, | |
| "loss": 4.1724, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5638297872340426e-05, | |
| "loss": 4.4194, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.531914893617021e-05, | |
| "loss": 4.6975, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5e-05, | |
| "loss": 4.3105, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.4680851063829787e-05, | |
| "loss": 4.1572, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.4361702127659575e-05, | |
| "loss": 4.0816, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.4042553191489362e-05, | |
| "loss": 4.0894, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.372340425531915e-05, | |
| "loss": 4.0141, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3404255319148936e-05, | |
| "loss": 4.0531, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3085106382978724e-05, | |
| "loss": 4.0733, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.2765957446808511e-05, | |
| "loss": 4.1754, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.2446808510638298e-05, | |
| "loss": 4.1839, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.2127659574468084e-05, | |
| "loss": 4.0334, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1808510638297872e-05, | |
| "loss": 4.0647, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.148936170212766e-05, | |
| "loss": 4.3947, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1170212765957447e-05, | |
| "loss": 4.0124, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0851063829787235e-05, | |
| "loss": 4.0597, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0531914893617022e-05, | |
| "loss": 4.1565, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0212765957446808e-05, | |
| "loss": 4.1457, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.893617021276595e-06, | |
| "loss": 3.9888, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.574468085106385e-06, | |
| "loss": 4.0965, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.255319148936171e-06, | |
| "loss": 4.0321, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.936170212765958e-06, | |
| "loss": 3.9334, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.617021276595746e-06, | |
| "loss": 4.0827, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.297872340425532e-06, | |
| "loss": 4.0785, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.978723404255319e-06, | |
| "loss": 4.1307, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.659574468085105e-06, | |
| "loss": 4.1616, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.3404255319148934e-06, | |
| "loss": 4.0963, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.021276595744681e-06, | |
| "loss": 4.2195, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.702127659574468e-06, | |
| "loss": 4.0826, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.3829787234042555e-06, | |
| "loss": 4.0871, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.063829787234042e-06, | |
| "loss": 3.9794, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.74468085106383e-06, | |
| "loss": 3.9492, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.4255319148936176e-06, | |
| "loss": 4.2226, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.106382978723404e-06, | |
| "loss": 4.2959, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.787234042553192e-06, | |
| "loss": 4.2149, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.468085106382979e-06, | |
| "loss": 4.0045, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.148936170212766e-06, | |
| "loss": 4.1531, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.829787234042553e-06, | |
| "loss": 4.1148, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.5106382978723404e-06, | |
| "loss": 4.0061, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.1914893617021277e-06, | |
| "loss": 4.1959, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.872340425531915e-06, | |
| "loss": 4.1816, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.553191489361702e-06, | |
| "loss": 4.3031, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.2340425531914894e-06, | |
| "loss": 4.1356, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.9148936170212763e-06, | |
| "loss": 4.3695, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5957446808510639e-06, | |
| "loss": 4.262, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.276595744680851e-06, | |
| "loss": 4.4783, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.574468085106382e-07, | |
| "loss": 4.3684, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 594, | |
| "total_flos": 0.0, | |
| "train_loss": 4.269026105652754, | |
| "train_runtime": 5521.3301, | |
| "train_samples_per_second": 5.169, | |
| "train_steps_per_second": 0.108 | |
| } | |
| ], | |
| "max_steps": 594, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |