diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,14117 +1,7365 @@ { - "best_metric": 1.8946523666381836, - "best_model_checkpoint": "lora/google/gemma-1.1-7b-it/unaligned/checkpoint-1800", - "epoch": 2.5591810620601407, - "eval_steps": 200, - "global_step": 2000, + "best_metric": 1.7247449159622192, + "best_model_checkpoint": "lora_lr/google/gemma-1.1-7b-it/unaligned/checkpoint-1000", + "epoch": 1.310300703774792, + "eval_steps": 50, + "global_step": 1024, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, - "grad_norm": 0.9921875, - "learning_rate": 2.0000000000000002e-07, + "grad_norm": 1.0390625, + "learning_rate": 2.0000000000000003e-06, "loss": 4.7488, "step": 1 }, { "epoch": 0.0, - "grad_norm": 0.92578125, - "learning_rate": 4.0000000000000003e-07, + "grad_norm": 0.97265625, + "learning_rate": 4.000000000000001e-06, "loss": 4.5149, "step": 2 }, { "epoch": 0.0, - "grad_norm": 0.890625, - "learning_rate": 6.000000000000001e-07, - "loss": 4.3983, + "grad_norm": 0.9296875, + "learning_rate": 6e-06, + "loss": 4.3836, "step": 3 }, { "epoch": 0.01, - "grad_norm": 0.984375, - "learning_rate": 8.000000000000001e-07, - "loss": 4.6728, + "grad_norm": 1.0234375, + "learning_rate": 8.000000000000001e-06, + "loss": 4.6777, "step": 4 }, { "epoch": 0.01, - "grad_norm": 0.98046875, - "learning_rate": 1.0000000000000002e-06, - "loss": 4.6766, + "grad_norm": 1.03125, + "learning_rate": 1e-05, + "loss": 4.6688, "step": 5 }, { "epoch": 0.01, - "grad_norm": 0.91015625, - "learning_rate": 1.2000000000000002e-06, - "loss": 4.5609, + "grad_norm": 0.94921875, + "learning_rate": 1.2e-05, + "loss": 4.5565, "step": 6 }, { "epoch": 0.01, - "grad_norm": 0.953125, - "learning_rate": 1.4000000000000001e-06, - "loss": 4.575, + "grad_norm": 1.0078125, + "learning_rate": 1.4000000000000001e-05, + "loss": 4.5658, "step": 7 }, { "epoch": 0.01, - "grad_norm": 0.96875, - "learning_rate": 1.6000000000000001e-06, - "loss": 4.6766, + "grad_norm": 1.015625, + "learning_rate": 1.6000000000000003e-05, + "loss": 4.6618, "step": 8 }, { "epoch": 0.01, - "grad_norm": 0.9921875, - "learning_rate": 1.8000000000000001e-06, - "loss": 4.6223, + "grad_norm": 1.0546875, + "learning_rate": 1.8e-05, + "loss": 4.6095, "step": 9 }, { "epoch": 0.01, - "grad_norm": 0.99609375, - "learning_rate": 2.0000000000000003e-06, - "loss": 4.7933, + "grad_norm": 1.0703125, + "learning_rate": 2e-05, + "loss": 4.7608, "step": 10 }, { "epoch": 0.01, - "grad_norm": 0.9609375, - "learning_rate": 2.2e-06, - "loss": 4.5758, + "grad_norm": 1.0546875, + "learning_rate": 2.2000000000000003e-05, + "loss": 4.5421, "step": 11 }, { "epoch": 0.02, - "grad_norm": 0.8671875, - "learning_rate": 2.4000000000000003e-06, - "loss": 4.3448, + "grad_norm": 0.9609375, + "learning_rate": 2.4e-05, + "loss": 4.2895, "step": 12 }, { "epoch": 0.02, - "grad_norm": 0.98828125, - "learning_rate": 2.6e-06, - "loss": 4.872, + "grad_norm": 1.125, + "learning_rate": 2.6000000000000002e-05, + "loss": 4.8112, "step": 13 }, { "epoch": 0.02, - "grad_norm": 0.97265625, - "learning_rate": 2.8000000000000003e-06, - "loss": 4.7587, + "grad_norm": 1.1328125, + "learning_rate": 2.8000000000000003e-05, + "loss": 4.6709, "step": 14 }, { "epoch": 0.02, - "grad_norm": 0.953125, - "learning_rate": 3e-06, - "loss": 4.6186, + "grad_norm": 1.1484375, + "learning_rate": 3e-05, + "loss": 4.5234, "step": 15 }, { "epoch": 0.02, - "grad_norm": 0.9140625, - "learning_rate": 3.2000000000000003e-06, - "loss": 4.4867, + "grad_norm": 1.1328125, + "learning_rate": 3.2000000000000005e-05, + "loss": 4.3751, "step": 16 }, { "epoch": 0.02, - "grad_norm": 0.98828125, - "learning_rate": 3.4000000000000005e-06, - "loss": 4.6872, + "grad_norm": 1.3046875, + "learning_rate": 3.4000000000000007e-05, + "loss": 4.5247, "step": 17 }, { "epoch": 0.02, - "grad_norm": 0.90234375, - "learning_rate": 3.6000000000000003e-06, - "loss": 4.5666, + "grad_norm": 1.265625, + "learning_rate": 3.6e-05, + "loss": 4.3906, "step": 18 }, { "epoch": 0.02, - "grad_norm": 0.953125, - "learning_rate": 3.8000000000000005e-06, - "loss": 4.545, + "grad_norm": 1.40625, + "learning_rate": 3.8e-05, + "loss": 4.3193, "step": 19 }, { "epoch": 0.03, - "grad_norm": 0.9140625, - "learning_rate": 4.000000000000001e-06, - "loss": 4.5483, + "grad_norm": 1.4296875, + "learning_rate": 4e-05, + "loss": 4.2987, "step": 20 }, { "epoch": 0.03, - "grad_norm": 0.88671875, - "learning_rate": 4.2000000000000004e-06, - "loss": 4.5187, + "grad_norm": 1.4375, + "learning_rate": 4.2e-05, + "loss": 4.2167, "step": 21 }, { "epoch": 0.03, - "grad_norm": 0.9375, - "learning_rate": 4.4e-06, - "loss": 4.6781, + "grad_norm": 1.6171875, + "learning_rate": 4.4000000000000006e-05, + "loss": 4.3088, "step": 22 }, { "epoch": 0.03, - "grad_norm": 0.94140625, - "learning_rate": 4.600000000000001e-06, - "loss": 4.7005, + "grad_norm": 1.6640625, + "learning_rate": 4.600000000000001e-05, + "loss": 4.2658, "step": 23 }, { "epoch": 0.03, - "grad_norm": 0.93359375, - "learning_rate": 4.800000000000001e-06, - "loss": 4.6542, + "grad_norm": 1.734375, + "learning_rate": 4.8e-05, + "loss": 4.1625, "step": 24 }, { "epoch": 0.03, - "grad_norm": 0.9375, - "learning_rate": 5e-06, - "loss": 4.609, + "grad_norm": 1.8203125, + "learning_rate": 5e-05, + "loss": 4.0392, "step": 25 }, { "epoch": 0.03, - "grad_norm": 0.98046875, - "learning_rate": 5.2e-06, - "loss": 4.6407, + "grad_norm": 1.8828125, + "learning_rate": 5.2000000000000004e-05, + "loss": 3.9772, "step": 26 }, { "epoch": 0.03, - "grad_norm": 0.9609375, - "learning_rate": 5.400000000000001e-06, - "loss": 4.6363, + "grad_norm": 1.8828125, + "learning_rate": 5.4000000000000005e-05, + "loss": 3.9029, "step": 27 }, { "epoch": 0.04, - "grad_norm": 0.96484375, - "learning_rate": 5.600000000000001e-06, - "loss": 4.6094, + "grad_norm": 1.875, + "learning_rate": 5.6000000000000006e-05, + "loss": 3.8107, "step": 28 }, { "epoch": 0.04, - "grad_norm": 0.98828125, - "learning_rate": 5.8e-06, - "loss": 4.7547, + "grad_norm": 1.96875, + "learning_rate": 5.8e-05, + "loss": 3.8199, "step": 29 }, { "epoch": 0.04, - "grad_norm": 0.953125, - "learning_rate": 6e-06, - "loss": 4.4928, + "grad_norm": 1.953125, + "learning_rate": 6e-05, + "loss": 3.5035, "step": 30 }, { "epoch": 0.04, - "grad_norm": 0.9921875, - "learning_rate": 6.200000000000001e-06, - "loss": 4.7672, + "grad_norm": 2.03125, + "learning_rate": 6.2e-05, + "loss": 3.6236, "step": 31 }, { "epoch": 0.04, - "grad_norm": 0.98046875, - "learning_rate": 6.4000000000000006e-06, - "loss": 4.5136, + "grad_norm": 1.796875, + "learning_rate": 6.400000000000001e-05, + "loss": 3.3359, "step": 32 }, { "epoch": 0.04, - "grad_norm": 0.9765625, - "learning_rate": 6.600000000000001e-06, - "loss": 4.5678, + "grad_norm": 1.7578125, + "learning_rate": 6.6e-05, + "loss": 3.3019, "step": 33 }, { "epoch": 0.04, - "grad_norm": 1.0, - "learning_rate": 6.800000000000001e-06, - "loss": 4.5823, + "grad_norm": 1.7109375, + "learning_rate": 6.800000000000001e-05, + "loss": 3.2003, "step": 34 }, { "epoch": 0.04, - "grad_norm": 1.015625, - "learning_rate": 7e-06, - "loss": 4.6403, + "grad_norm": 1.6640625, + "learning_rate": 7e-05, + "loss": 3.1585, "step": 35 }, { "epoch": 0.05, - "grad_norm": 1.0234375, - "learning_rate": 7.2000000000000005e-06, - "loss": 4.5009, + "grad_norm": 1.53125, + "learning_rate": 7.2e-05, + "loss": 2.9377, "step": 36 }, { "epoch": 0.05, - "grad_norm": 1.03125, - "learning_rate": 7.4e-06, - "loss": 4.8083, + "grad_norm": 1.5625, + "learning_rate": 7.4e-05, + "loss": 3.13, "step": 37 }, { "epoch": 0.05, - "grad_norm": 1.0, - "learning_rate": 7.600000000000001e-06, - "loss": 4.5979, + "grad_norm": 1.453125, + "learning_rate": 7.6e-05, + "loss": 2.9445, "step": 38 }, { "epoch": 0.05, - "grad_norm": 1.0546875, - "learning_rate": 7.800000000000002e-06, - "loss": 4.5374, + "grad_norm": 1.21875, + "learning_rate": 7.800000000000001e-05, + "loss": 2.7348, "step": 39 }, { "epoch": 0.05, - "grad_norm": 1.0234375, - "learning_rate": 8.000000000000001e-06, - "loss": 4.5749, + "grad_norm": 1.1171875, + "learning_rate": 8e-05, + "loss": 2.6842, "step": 40 }, { "epoch": 0.05, - "grad_norm": 1.0234375, - "learning_rate": 8.2e-06, - "loss": 4.5442, + "grad_norm": 0.90625, + "learning_rate": 8.2e-05, + "loss": 2.6921, "step": 41 }, { "epoch": 0.05, - "grad_norm": 1.03125, - "learning_rate": 8.400000000000001e-06, - "loss": 4.6575, + "grad_norm": 0.9453125, + "learning_rate": 8.4e-05, + "loss": 2.6504, "step": 42 }, { "epoch": 0.06, - "grad_norm": 1.0859375, - "learning_rate": 8.6e-06, - "loss": 4.5536, + "grad_norm": 1.125, + "learning_rate": 8.6e-05, + "loss": 2.4995, "step": 43 }, { "epoch": 0.06, - "grad_norm": 1.0546875, - "learning_rate": 8.8e-06, - "loss": 4.4276, + "grad_norm": 1.34375, + "learning_rate": 8.800000000000001e-05, + "loss": 2.4545, "step": 44 }, { "epoch": 0.06, - "grad_norm": 1.0234375, - "learning_rate": 9e-06, - "loss": 4.5334, + "grad_norm": 1.28125, + "learning_rate": 9e-05, + "loss": 2.4311, "step": 45 }, { "epoch": 0.06, - "grad_norm": 1.0390625, - "learning_rate": 9.200000000000002e-06, - "loss": 4.4418, + "grad_norm": 0.640625, + "learning_rate": 9.200000000000001e-05, + "loss": 2.4045, "step": 46 }, { "epoch": 0.06, - "grad_norm": 1.078125, - "learning_rate": 9.4e-06, - "loss": 4.5444, + "grad_norm": 0.54296875, + "learning_rate": 9.4e-05, + "loss": 2.413, "step": 47 }, { "epoch": 0.06, - "grad_norm": 1.0859375, - "learning_rate": 9.600000000000001e-06, - "loss": 4.4572, + "grad_norm": 0.578125, + "learning_rate": 9.6e-05, + "loss": 2.3691, "step": 48 }, { "epoch": 0.06, - "grad_norm": 1.09375, - "learning_rate": 9.800000000000001e-06, - "loss": 4.4011, + "grad_norm": 0.65234375, + "learning_rate": 9.8e-05, + "loss": 2.2936, "step": 49 }, { "epoch": 0.06, - "grad_norm": 1.078125, - "learning_rate": 1e-05, - "loss": 4.4517, + "grad_norm": 0.62109375, + "learning_rate": 0.0001, + "loss": 2.3459, + "step": 50 + }, + { + "epoch": 0.06, + "eval_loss": 2.298555374145508, + "eval_runtime": 125.5458, + "eval_samples_per_second": 39.826, + "eval_steps_per_second": 1.251, "step": 50 }, { "epoch": 0.07, - "grad_norm": 1.2109375, - "learning_rate": 1.02e-05, - "loss": 4.6872, + "grad_norm": 0.5859375, + "learning_rate": 0.00010200000000000001, + "loss": 2.458, "step": 51 }, { "epoch": 0.07, - "grad_norm": 1.1484375, - "learning_rate": 1.04e-05, - "loss": 4.4797, + "grad_norm": 0.61328125, + "learning_rate": 0.00010400000000000001, + "loss": 2.2395, "step": 52 }, { "epoch": 0.07, - "grad_norm": 1.1015625, - "learning_rate": 1.0600000000000002e-05, - "loss": 4.3362, + "grad_norm": 0.55859375, + "learning_rate": 0.00010600000000000002, + "loss": 2.2084, "step": 53 }, { "epoch": 0.07, - "grad_norm": 1.2890625, - "learning_rate": 1.0800000000000002e-05, - "loss": 4.6347, + "grad_norm": 0.83203125, + "learning_rate": 0.00010800000000000001, + "loss": 2.2806, "step": 54 }, { "epoch": 0.07, - "grad_norm": 1.2734375, - "learning_rate": 1.1000000000000001e-05, - "loss": 4.3814, + "grad_norm": 1.15625, + "learning_rate": 0.00011000000000000002, + "loss": 2.1546, "step": 55 }, { "epoch": 0.07, - "grad_norm": 1.2890625, - "learning_rate": 1.1200000000000001e-05, - "loss": 4.6578, + "grad_norm": 0.91015625, + "learning_rate": 0.00011200000000000001, + "loss": 2.2027, "step": 56 }, { "epoch": 0.07, - "grad_norm": 1.296875, - "learning_rate": 1.14e-05, - "loss": 4.506, + "grad_norm": 0.6640625, + "learning_rate": 0.00011399999999999999, + "loss": 2.1795, "step": 57 }, { "epoch": 0.07, - "grad_norm": 1.3125, - "learning_rate": 1.16e-05, - "loss": 4.5851, + "grad_norm": 0.5, + "learning_rate": 0.000116, + "loss": 2.1918, "step": 58 }, { "epoch": 0.08, - "grad_norm": 1.25, - "learning_rate": 1.18e-05, - "loss": 4.3112, + "grad_norm": 0.39453125, + "learning_rate": 0.000118, + "loss": 2.143, "step": 59 }, { "epoch": 0.08, - "grad_norm": 1.2890625, - "learning_rate": 1.2e-05, - "loss": 4.3152, + "grad_norm": 0.4375, + "learning_rate": 0.00012, + "loss": 2.1451, "step": 60 }, { "epoch": 0.08, - "grad_norm": 1.34375, - "learning_rate": 1.22e-05, - "loss": 4.3976, + "grad_norm": 0.388671875, + "learning_rate": 0.000122, + "loss": 2.1542, "step": 61 }, { "epoch": 0.08, - "grad_norm": 1.4140625, - "learning_rate": 1.2400000000000002e-05, - "loss": 4.3991, + "grad_norm": 0.3984375, + "learning_rate": 0.000124, + "loss": 2.1125, "step": 62 }, { "epoch": 0.08, - "grad_norm": 1.453125, - "learning_rate": 1.2600000000000001e-05, - "loss": 4.4819, + "grad_norm": 0.6953125, + "learning_rate": 0.000126, + "loss": 2.1123, "step": 63 }, { "epoch": 0.08, - "grad_norm": 1.390625, - "learning_rate": 1.2800000000000001e-05, - "loss": 4.3024, + "grad_norm": 0.451171875, + "learning_rate": 0.00012800000000000002, + "loss": 2.0354, "step": 64 }, { "epoch": 0.08, - "grad_norm": 1.421875, - "learning_rate": 1.3000000000000001e-05, - "loss": 4.3959, + "grad_norm": 0.451171875, + "learning_rate": 0.00013000000000000002, + "loss": 2.1228, "step": 65 }, { "epoch": 0.08, - "grad_norm": 1.453125, - "learning_rate": 1.3200000000000002e-05, - "loss": 4.3763, + "grad_norm": 0.77734375, + "learning_rate": 0.000132, + "loss": 2.1666, "step": 66 }, { "epoch": 0.09, - "grad_norm": 1.40625, - "learning_rate": 1.3400000000000002e-05, - "loss": 4.2396, + "grad_norm": 0.455078125, + "learning_rate": 0.000134, + "loss": 2.0429, "step": 67 }, { "epoch": 0.09, - "grad_norm": 1.5390625, - "learning_rate": 1.3600000000000002e-05, - "loss": 4.3083, + "grad_norm": 0.384765625, + "learning_rate": 0.00013600000000000003, + "loss": 2.0367, "step": 68 }, { "epoch": 0.09, - "grad_norm": 1.4453125, - "learning_rate": 1.38e-05, - "loss": 4.0614, + "grad_norm": 0.365234375, + "learning_rate": 0.000138, + "loss": 2.0139, "step": 69 }, { "epoch": 0.09, - "grad_norm": 1.53125, - "learning_rate": 1.4e-05, - "loss": 4.1741, + "grad_norm": 0.3671875, + "learning_rate": 0.00014, + "loss": 2.0284, "step": 70 }, { "epoch": 0.09, - "grad_norm": 1.625, - "learning_rate": 1.4200000000000001e-05, - "loss": 4.3138, + "grad_norm": 0.404296875, + "learning_rate": 0.000142, + "loss": 2.0303, "step": 71 }, { "epoch": 0.09, - "grad_norm": 1.6171875, - "learning_rate": 1.4400000000000001e-05, - "loss": 4.298, + "grad_norm": 0.3515625, + "learning_rate": 0.000144, + "loss": 2.0645, "step": 72 }, { "epoch": 0.09, - "grad_norm": 1.6484375, - "learning_rate": 1.46e-05, - "loss": 4.2097, + "grad_norm": 0.365234375, + "learning_rate": 0.000146, + "loss": 2.001, "step": 73 }, { "epoch": 0.09, - "grad_norm": 1.53125, - "learning_rate": 1.48e-05, - "loss": 4.1488, + "grad_norm": 0.3046875, + "learning_rate": 0.000148, + "loss": 2.0946, "step": 74 }, { "epoch": 0.1, - "grad_norm": 1.6015625, - "learning_rate": 1.5000000000000002e-05, - "loss": 4.1016, + "grad_norm": 0.314453125, + "learning_rate": 0.00015000000000000001, + "loss": 1.9969, "step": 75 }, { "epoch": 0.1, - "grad_norm": 1.6640625, - "learning_rate": 1.5200000000000002e-05, - "loss": 4.112, + "grad_norm": 0.419921875, + "learning_rate": 0.000152, + "loss": 1.9911, "step": 76 }, { "epoch": 0.1, - "grad_norm": 1.7578125, - "learning_rate": 1.54e-05, - "loss": 4.1549, + "grad_norm": 0.328125, + "learning_rate": 0.000154, + "loss": 2.0011, "step": 77 }, { "epoch": 0.1, - "grad_norm": 1.640625, - "learning_rate": 1.5600000000000003e-05, - "loss": 3.9731, + "grad_norm": 0.234375, + "learning_rate": 0.00015600000000000002, + "loss": 1.9893, "step": 78 }, { "epoch": 0.1, - "grad_norm": 1.6640625, - "learning_rate": 1.58e-05, - "loss": 4.0544, + "grad_norm": 0.380859375, + "learning_rate": 0.00015800000000000002, + "loss": 1.9336, "step": 79 }, { "epoch": 0.1, - "grad_norm": 1.7890625, - "learning_rate": 1.6000000000000003e-05, - "loss": 4.2139, + "grad_norm": 0.2890625, + "learning_rate": 0.00016, + "loss": 1.9876, "step": 80 }, { "epoch": 0.1, - "grad_norm": 1.7578125, - "learning_rate": 1.62e-05, - "loss": 4.1495, + "grad_norm": 0.255859375, + "learning_rate": 0.000162, + "loss": 1.9679, "step": 81 }, { "epoch": 0.1, - "grad_norm": 1.890625, - "learning_rate": 1.64e-05, - "loss": 4.0582, + "grad_norm": 0.23828125, + "learning_rate": 0.000164, + "loss": 1.9157, "step": 82 }, { "epoch": 0.11, - "grad_norm": 1.8515625, - "learning_rate": 1.66e-05, - "loss": 4.094, + "grad_norm": 0.373046875, + "learning_rate": 0.000166, + "loss": 1.9939, "step": 83 }, { "epoch": 0.11, - "grad_norm": 1.8125, - "learning_rate": 1.6800000000000002e-05, - "loss": 4.0135, + "grad_norm": 0.255859375, + "learning_rate": 0.000168, + "loss": 1.9457, "step": 84 }, { "epoch": 0.11, - "grad_norm": 1.8515625, - "learning_rate": 1.7e-05, - "loss": 3.9677, + "grad_norm": 0.26171875, + "learning_rate": 0.00017, + "loss": 1.9924, "step": 85 }, { "epoch": 0.11, - "grad_norm": 1.765625, - "learning_rate": 1.72e-05, - "loss": 3.8012, + "grad_norm": 0.2275390625, + "learning_rate": 0.000172, + "loss": 1.8708, "step": 86 }, { "epoch": 0.11, - "grad_norm": 1.859375, - "learning_rate": 1.7400000000000003e-05, - "loss": 3.929, + "grad_norm": 0.26953125, + "learning_rate": 0.000174, + "loss": 1.946, "step": 87 }, { "epoch": 0.11, - "grad_norm": 1.875, - "learning_rate": 1.76e-05, - "loss": 3.7884, + "grad_norm": 0.255859375, + "learning_rate": 0.00017600000000000002, + "loss": 1.9743, "step": 88 }, { "epoch": 0.11, - "grad_norm": 2.0625, - "learning_rate": 1.7800000000000002e-05, - "loss": 4.0489, + "grad_norm": 0.2490234375, + "learning_rate": 0.00017800000000000002, + "loss": 1.9257, "step": 89 }, { "epoch": 0.12, - "grad_norm": 1.953125, - "learning_rate": 1.8e-05, - "loss": 3.9287, + "grad_norm": 0.26953125, + "learning_rate": 0.00018, + "loss": 1.9041, "step": 90 }, { "epoch": 0.12, - "grad_norm": 1.9140625, - "learning_rate": 1.8200000000000002e-05, - "loss": 3.7971, + "grad_norm": 0.265625, + "learning_rate": 0.000182, + "loss": 1.9502, "step": 91 }, { "epoch": 0.12, - "grad_norm": 2.0625, - "learning_rate": 1.8400000000000003e-05, - "loss": 3.8392, + "grad_norm": 0.240234375, + "learning_rate": 0.00018400000000000003, + "loss": 1.8514, "step": 92 }, { "epoch": 0.12, - "grad_norm": 1.7890625, - "learning_rate": 1.86e-05, - "loss": 3.7233, + "grad_norm": 0.255859375, + "learning_rate": 0.00018600000000000002, + "loss": 2.0176, "step": 93 }, { "epoch": 0.12, - "grad_norm": 1.9296875, - "learning_rate": 1.88e-05, - "loss": 3.7073, + "grad_norm": 0.298828125, + "learning_rate": 0.000188, + "loss": 1.9056, "step": 94 }, { "epoch": 0.12, - "grad_norm": 2.03125, - "learning_rate": 1.9e-05, - "loss": 3.8126, + "grad_norm": 0.26171875, + "learning_rate": 0.00019, + "loss": 1.9149, "step": 95 }, { "epoch": 0.12, - "grad_norm": 1.8671875, - "learning_rate": 1.9200000000000003e-05, - "loss": 3.6088, + "grad_norm": 0.212890625, + "learning_rate": 0.000192, + "loss": 1.9587, "step": 96 }, { "epoch": 0.12, - "grad_norm": 2.046875, - "learning_rate": 1.94e-05, - "loss": 3.7216, + "grad_norm": 0.26953125, + "learning_rate": 0.000194, + "loss": 1.9103, "step": 97 }, { "epoch": 0.13, - "grad_norm": 1.9375, - "learning_rate": 1.9600000000000002e-05, - "loss": 3.5325, + "grad_norm": 0.28515625, + "learning_rate": 0.000196, + "loss": 1.8814, "step": 98 }, { "epoch": 0.13, - "grad_norm": 2.0, - "learning_rate": 1.98e-05, - "loss": 3.5355, + "grad_norm": 0.25, + "learning_rate": 0.00019800000000000002, + "loss": 1.842, "step": 99 }, { "epoch": 0.13, - "grad_norm": 2.109375, - "learning_rate": 2e-05, - "loss": 3.4773, + "grad_norm": 0.205078125, + "learning_rate": 0.0002, + "loss": 1.8251, + "step": 100 + }, + { + "epoch": 0.13, + "eval_loss": 1.8968226909637451, + "eval_runtime": 125.222, + "eval_samples_per_second": 39.929, + "eval_steps_per_second": 1.254, "step": 100 }, { "epoch": 0.13, - "grad_norm": 2.078125, - "learning_rate": 1.9989473684210526e-05, - "loss": 3.5609, + "grad_norm": 0.275390625, + "learning_rate": 0.00019978354978354978, + "loss": 1.84, "step": 101 }, { "epoch": 0.13, - "grad_norm": 2.03125, - "learning_rate": 1.9978947368421054e-05, - "loss": 3.5099, + "grad_norm": 0.2236328125, + "learning_rate": 0.00019956709956709957, + "loss": 1.9066, "step": 102 }, { "epoch": 0.13, - "grad_norm": 1.9609375, - "learning_rate": 1.9968421052631582e-05, - "loss": 3.5835, + "grad_norm": 0.244140625, + "learning_rate": 0.00019935064935064936, + "loss": 1.9553, "step": 103 }, { "epoch": 0.13, - "grad_norm": 1.859375, - "learning_rate": 1.9957894736842107e-05, - "loss": 3.3607, + "grad_norm": 0.2451171875, + "learning_rate": 0.00019913419913419916, + "loss": 1.922, "step": 104 }, { "epoch": 0.13, - "grad_norm": 1.9765625, - "learning_rate": 1.994736842105263e-05, - "loss": 3.3529, + "grad_norm": 0.255859375, + "learning_rate": 0.00019891774891774892, + "loss": 1.8927, "step": 105 }, { "epoch": 0.14, - "grad_norm": 1.8828125, - "learning_rate": 1.993684210526316e-05, - "loss": 3.3442, + "grad_norm": 0.21484375, + "learning_rate": 0.00019870129870129872, + "loss": 1.9244, "step": 106 }, { "epoch": 0.14, - "grad_norm": 1.890625, - "learning_rate": 1.9926315789473688e-05, - "loss": 3.2842, + "grad_norm": 0.265625, + "learning_rate": 0.0001984848484848485, + "loss": 1.8875, "step": 107 }, { "epoch": 0.14, - "grad_norm": 1.96875, - "learning_rate": 1.9915789473684212e-05, - "loss": 3.2969, + "grad_norm": 0.388671875, + "learning_rate": 0.00019826839826839827, + "loss": 1.837, "step": 108 }, { "epoch": 0.14, - "grad_norm": 1.828125, - "learning_rate": 1.990526315789474e-05, - "loss": 3.2052, + "grad_norm": 0.228515625, + "learning_rate": 0.00019805194805194807, + "loss": 1.8613, "step": 109 }, { "epoch": 0.14, - "grad_norm": 1.9140625, - "learning_rate": 1.9894736842105265e-05, - "loss": 3.1979, + "grad_norm": 0.2216796875, + "learning_rate": 0.00019783549783549783, + "loss": 1.9369, "step": 110 }, { "epoch": 0.14, - "grad_norm": 1.9453125, - "learning_rate": 1.988421052631579e-05, - "loss": 3.3072, + "grad_norm": 0.234375, + "learning_rate": 0.00019761904761904763, + "loss": 1.8546, "step": 111 }, { "epoch": 0.14, - "grad_norm": 1.8984375, - "learning_rate": 1.9873684210526318e-05, - "loss": 3.3039, + "grad_norm": 0.3125, + "learning_rate": 0.00019740259740259742, + "loss": 1.8774, "step": 112 }, { "epoch": 0.14, - "grad_norm": 1.8046875, - "learning_rate": 1.9863157894736846e-05, - "loss": 3.1589, + "grad_norm": 0.28125, + "learning_rate": 0.0001971861471861472, + "loss": 1.8799, "step": 113 }, { "epoch": 0.15, - "grad_norm": 1.9375, - "learning_rate": 1.985263157894737e-05, - "loss": 3.1928, + "grad_norm": 0.224609375, + "learning_rate": 0.00019696969696969698, + "loss": 1.8776, "step": 114 }, { "epoch": 0.15, - "grad_norm": 1.7890625, - "learning_rate": 1.9842105263157895e-05, - "loss": 3.1733, + "grad_norm": 0.2412109375, + "learning_rate": 0.00019675324675324675, + "loss": 1.8823, "step": 115 }, { "epoch": 0.15, - "grad_norm": 1.7734375, - "learning_rate": 1.9831578947368423e-05, - "loss": 3.0587, + "grad_norm": 0.234375, + "learning_rate": 0.00019653679653679654, + "loss": 1.8877, "step": 116 }, { "epoch": 0.15, - "grad_norm": 1.7578125, - "learning_rate": 1.982105263157895e-05, - "loss": 3.1787, + "grad_norm": 0.259765625, + "learning_rate": 0.00019632034632034633, + "loss": 1.9692, "step": 117 }, { "epoch": 0.15, - "grad_norm": 1.625, - "learning_rate": 1.9810526315789476e-05, - "loss": 2.9748, + "grad_norm": 0.208984375, + "learning_rate": 0.00019610389610389613, + "loss": 1.8985, "step": 118 }, { "epoch": 0.15, - "grad_norm": 1.7734375, - "learning_rate": 1.98e-05, - "loss": 3.0123, + "grad_norm": 0.248046875, + "learning_rate": 0.0001958874458874459, + "loss": 1.8564, "step": 119 }, { "epoch": 0.15, - "grad_norm": 1.625, - "learning_rate": 1.9789473684210528e-05, - "loss": 2.9925, + "grad_norm": 0.2109375, + "learning_rate": 0.00019567099567099566, + "loss": 1.9001, "step": 120 }, { "epoch": 0.15, - "grad_norm": 1.46875, - "learning_rate": 1.9778947368421056e-05, - "loss": 2.9014, + "grad_norm": 0.2080078125, + "learning_rate": 0.00019545454545454548, + "loss": 1.879, "step": 121 }, { "epoch": 0.16, - "grad_norm": 1.8046875, - "learning_rate": 1.976842105263158e-05, - "loss": 3.045, + "grad_norm": 0.326171875, + "learning_rate": 0.00019523809523809525, + "loss": 1.8385, "step": 122 }, { "epoch": 0.16, - "grad_norm": 1.46875, - "learning_rate": 1.9757894736842105e-05, - "loss": 2.8676, + "grad_norm": 0.24609375, + "learning_rate": 0.00019502164502164504, + "loss": 1.8581, "step": 123 }, { "epoch": 0.16, - "grad_norm": 1.4921875, - "learning_rate": 1.9747368421052633e-05, - "loss": 2.9012, + "grad_norm": 0.2294921875, + "learning_rate": 0.0001948051948051948, + "loss": 1.8517, "step": 124 }, { "epoch": 0.16, - "grad_norm": 1.5390625, - "learning_rate": 1.9736842105263158e-05, - "loss": 2.8709, + "grad_norm": 0.21875, + "learning_rate": 0.0001945887445887446, + "loss": 1.8695, "step": 125 }, { "epoch": 0.16, - "grad_norm": 1.5859375, - "learning_rate": 1.9726315789473686e-05, - "loss": 2.8891, + "grad_norm": 0.3125, + "learning_rate": 0.0001943722943722944, + "loss": 1.8675, "step": 126 }, { "epoch": 0.16, - "grad_norm": 1.46875, - "learning_rate": 1.9715789473684214e-05, - "loss": 2.8579, + "grad_norm": 0.220703125, + "learning_rate": 0.00019415584415584416, + "loss": 1.9089, "step": 127 }, { "epoch": 0.16, - "grad_norm": 1.3671875, - "learning_rate": 1.970526315789474e-05, - "loss": 2.7449, + "grad_norm": 0.2314453125, + "learning_rate": 0.00019393939393939395, + "loss": 1.831, "step": 128 }, { "epoch": 0.17, - "grad_norm": 1.2734375, - "learning_rate": 1.9694736842105263e-05, - "loss": 2.7723, + "grad_norm": 0.2734375, + "learning_rate": 0.00019372294372294372, + "loss": 1.932, "step": 129 }, { "epoch": 0.17, - "grad_norm": 1.203125, - "learning_rate": 1.968421052631579e-05, - "loss": 2.7487, + "grad_norm": 0.2216796875, + "learning_rate": 0.00019350649350649354, + "loss": 1.917, "step": 130 }, { "epoch": 0.17, - "grad_norm": 1.5078125, - "learning_rate": 1.967368421052632e-05, - "loss": 2.7837, + "grad_norm": 0.28125, + "learning_rate": 0.0001932900432900433, + "loss": 1.8179, "step": 131 }, { "epoch": 0.17, - "grad_norm": 1.3515625, - "learning_rate": 1.9663157894736844e-05, - "loss": 2.7682, + "grad_norm": 0.294921875, + "learning_rate": 0.00019307359307359307, + "loss": 1.8764, "step": 132 }, { "epoch": 0.17, - "grad_norm": 1.203125, - "learning_rate": 1.965263157894737e-05, - "loss": 2.7534, + "grad_norm": 0.26171875, + "learning_rate": 0.00019285714285714286, + "loss": 1.8926, "step": 133 }, { "epoch": 0.17, - "grad_norm": 1.1484375, - "learning_rate": 1.9642105263157897e-05, - "loss": 2.7202, + "grad_norm": 0.24609375, + "learning_rate": 0.00019264069264069266, + "loss": 1.9443, "step": 134 }, { "epoch": 0.17, - "grad_norm": 1.1171875, - "learning_rate": 1.9631578947368425e-05, - "loss": 2.6697, + "grad_norm": 0.251953125, + "learning_rate": 0.00019242424242424245, + "loss": 1.8702, "step": 135 }, { "epoch": 0.17, - "grad_norm": 1.1015625, - "learning_rate": 1.962105263157895e-05, - "loss": 2.6528, + "grad_norm": 0.32421875, + "learning_rate": 0.00019220779220779222, + "loss": 1.8728, "step": 136 }, { "epoch": 0.18, - "grad_norm": 1.140625, - "learning_rate": 1.9610526315789474e-05, - "loss": 2.6111, + "grad_norm": 0.287109375, + "learning_rate": 0.000191991341991342, + "loss": 1.7953, "step": 137 }, { "epoch": 0.18, - "grad_norm": 0.99609375, - "learning_rate": 1.9600000000000002e-05, - "loss": 2.5176, + "grad_norm": 0.283203125, + "learning_rate": 0.00019177489177489178, + "loss": 1.7883, "step": 138 }, { "epoch": 0.18, - "grad_norm": 1.1328125, - "learning_rate": 1.9589473684210527e-05, - "loss": 2.6876, + "grad_norm": 0.302734375, + "learning_rate": 0.00019155844155844157, + "loss": 1.8741, "step": 139 }, { "epoch": 0.18, - "grad_norm": 0.89453125, - "learning_rate": 1.9578947368421055e-05, - "loss": 2.5215, + "grad_norm": 0.25390625, + "learning_rate": 0.00019134199134199136, + "loss": 1.8094, "step": 140 }, { "epoch": 0.18, - "grad_norm": 1.0078125, - "learning_rate": 1.956842105263158e-05, - "loss": 2.6103, + "grad_norm": 0.302734375, + "learning_rate": 0.00019112554112554113, + "loss": 1.8629, "step": 141 }, { "epoch": 0.18, - "grad_norm": 0.82421875, - "learning_rate": 1.9557894736842107e-05, - "loss": 2.545, + "grad_norm": 0.2333984375, + "learning_rate": 0.00019090909090909092, + "loss": 1.855, "step": 142 }, { "epoch": 0.18, - "grad_norm": 0.83984375, - "learning_rate": 1.9547368421052632e-05, - "loss": 2.5271, + "grad_norm": 0.236328125, + "learning_rate": 0.0001906926406926407, + "loss": 1.848, "step": 143 }, { "epoch": 0.18, - "grad_norm": 0.79296875, - "learning_rate": 1.953684210526316e-05, - "loss": 2.5178, + "grad_norm": 0.265625, + "learning_rate": 0.00019047619047619048, + "loss": 1.8853, "step": 144 }, { "epoch": 0.19, - "grad_norm": 0.87890625, - "learning_rate": 1.9526315789473688e-05, - "loss": 2.5122, + "grad_norm": 0.47265625, + "learning_rate": 0.00019025974025974027, + "loss": 1.8435, "step": 145 }, { "epoch": 0.19, - "grad_norm": 0.703125, - "learning_rate": 1.9515789473684213e-05, - "loss": 2.4519, + "grad_norm": 0.263671875, + "learning_rate": 0.00019004329004329004, + "loss": 1.8619, "step": 146 }, { "epoch": 0.19, - "grad_norm": 0.7109375, - "learning_rate": 1.9505263157894737e-05, - "loss": 2.4343, + "grad_norm": 0.37890625, + "learning_rate": 0.00018982683982683983, + "loss": 1.8298, "step": 147 }, { "epoch": 0.19, - "grad_norm": 0.69140625, - "learning_rate": 1.9494736842105265e-05, - "loss": 2.3365, + "grad_norm": 0.236328125, + "learning_rate": 0.00018961038961038963, + "loss": 1.7478, "step": 148 }, { "epoch": 0.19, - "grad_norm": 0.6953125, - "learning_rate": 1.9484210526315793e-05, - "loss": 2.508, + "grad_norm": 0.275390625, + "learning_rate": 0.00018939393939393942, + "loss": 1.906, "step": 149 }, { "epoch": 0.19, - "grad_norm": 0.69921875, - "learning_rate": 1.9473684210526318e-05, - "loss": 2.4876, + "grad_norm": 0.3046875, + "learning_rate": 0.0001891774891774892, + "loss": 1.9013, + "step": 150 + }, + { + "epoch": 0.19, + "eval_loss": 1.8299968242645264, + "eval_runtime": 125.2613, + "eval_samples_per_second": 39.917, + "eval_steps_per_second": 1.253, "step": 150 }, { "epoch": 0.19, - "grad_norm": 0.70703125, - "learning_rate": 1.9463157894736843e-05, - "loss": 2.4209, + "grad_norm": 0.29296875, + "learning_rate": 0.00018896103896103895, + "loss": 1.7772, "step": 151 }, { "epoch": 0.19, - "grad_norm": 0.671875, - "learning_rate": 1.945263157894737e-05, - "loss": 2.5121, + "grad_norm": 0.2490234375, + "learning_rate": 0.00018874458874458875, + "loss": 1.9428, "step": 152 }, { "epoch": 0.2, - "grad_norm": 0.6484375, - "learning_rate": 1.9442105263157895e-05, - "loss": 2.3667, + "grad_norm": 0.2314453125, + "learning_rate": 0.00018852813852813854, + "loss": 1.8121, "step": 153 }, { "epoch": 0.2, - "grad_norm": 0.640625, - "learning_rate": 1.9431578947368423e-05, - "loss": 2.3891, + "grad_norm": 0.29296875, + "learning_rate": 0.00018831168831168833, + "loss": 1.7921, "step": 154 }, { "epoch": 0.2, - "grad_norm": 0.55859375, - "learning_rate": 1.9421052631578948e-05, - "loss": 2.386, + "grad_norm": 0.357421875, + "learning_rate": 0.0001880952380952381, + "loss": 1.846, "step": 155 }, { "epoch": 0.2, - "grad_norm": 0.55859375, - "learning_rate": 1.9410526315789476e-05, - "loss": 2.298, + "grad_norm": 0.2236328125, + "learning_rate": 0.0001878787878787879, + "loss": 1.7703, "step": 156 }, { "epoch": 0.2, - "grad_norm": 0.48828125, - "learning_rate": 1.94e-05, - "loss": 2.2953, + "grad_norm": 0.2119140625, + "learning_rate": 0.00018766233766233769, + "loss": 1.8048, "step": 157 }, { "epoch": 0.2, - "grad_norm": 0.625, - "learning_rate": 1.9389473684210525e-05, - "loss": 2.3843, + "grad_norm": 0.33203125, + "learning_rate": 0.00018744588744588745, + "loss": 1.8353, "step": 158 }, { "epoch": 0.2, - "grad_norm": 0.52734375, - "learning_rate": 1.9378947368421053e-05, - "loss": 2.3736, + "grad_norm": 0.26953125, + "learning_rate": 0.00018722943722943725, + "loss": 1.8538, "step": 159 }, { "epoch": 0.2, - "grad_norm": 0.57421875, - "learning_rate": 1.936842105263158e-05, - "loss": 2.3611, + "grad_norm": 0.259765625, + "learning_rate": 0.000187012987012987, + "loss": 1.828, "step": 160 }, { "epoch": 0.21, - "grad_norm": 0.4765625, - "learning_rate": 1.9357894736842106e-05, - "loss": 2.3288, + "grad_norm": 0.2412109375, + "learning_rate": 0.0001867965367965368, + "loss": 1.8555, "step": 161 }, { "epoch": 0.21, - "grad_norm": 0.55078125, - "learning_rate": 1.9347368421052634e-05, - "loss": 2.287, + "grad_norm": 0.3125, + "learning_rate": 0.0001865800865800866, + "loss": 1.8, "step": 162 }, { "epoch": 0.21, - "grad_norm": 0.51953125, - "learning_rate": 1.9336842105263162e-05, - "loss": 2.3166, + "grad_norm": 0.236328125, + "learning_rate": 0.00018636363636363636, + "loss": 1.8241, "step": 163 }, { "epoch": 0.21, - "grad_norm": 0.55859375, - "learning_rate": 1.9326315789473687e-05, - "loss": 2.3146, + "grad_norm": 0.25, + "learning_rate": 0.00018614718614718616, + "loss": 1.7903, "step": 164 }, { "epoch": 0.21, - "grad_norm": 0.47265625, - "learning_rate": 1.931578947368421e-05, - "loss": 2.2308, + "grad_norm": 0.2431640625, + "learning_rate": 0.00018593073593073592, + "loss": 1.7279, "step": 165 }, { "epoch": 0.21, - "grad_norm": 0.55078125, - "learning_rate": 1.930526315789474e-05, - "loss": 2.2967, + "grad_norm": 0.33984375, + "learning_rate": 0.00018571428571428572, + "loss": 1.8231, "step": 166 }, { "epoch": 0.21, - "grad_norm": 0.61328125, - "learning_rate": 1.9294736842105264e-05, - "loss": 2.3688, + "grad_norm": 0.2451171875, + "learning_rate": 0.0001854978354978355, + "loss": 1.8534, "step": 167 }, { "epoch": 0.21, - "grad_norm": 0.49609375, - "learning_rate": 1.9284210526315792e-05, - "loss": 2.3645, + "grad_norm": 0.2470703125, + "learning_rate": 0.0001852813852813853, + "loss": 1.9175, "step": 168 }, { "epoch": 0.22, - "grad_norm": 0.59765625, - "learning_rate": 1.9273684210526317e-05, - "loss": 2.3172, + "grad_norm": 0.341796875, + "learning_rate": 0.00018506493506493507, + "loss": 1.7986, "step": 169 }, { "epoch": 0.22, - "grad_norm": 0.466796875, - "learning_rate": 1.9263157894736845e-05, - "loss": 2.3765, + "grad_norm": 0.294921875, + "learning_rate": 0.00018484848484848484, + "loss": 1.9097, "step": 170 }, { "epoch": 0.22, - "grad_norm": 0.498046875, - "learning_rate": 1.925263157894737e-05, - "loss": 2.2943, + "grad_norm": 0.3046875, + "learning_rate": 0.00018463203463203466, + "loss": 1.8232, "step": 171 }, { "epoch": 0.22, - "grad_norm": 0.423828125, - "learning_rate": 1.9242105263157894e-05, - "loss": 2.195, + "grad_norm": 0.38671875, + "learning_rate": 0.00018441558441558442, + "loss": 1.7505, "step": 172 }, { "epoch": 0.22, - "grad_norm": 0.47265625, - "learning_rate": 1.9231578947368422e-05, - "loss": 2.2773, + "grad_norm": 0.283203125, + "learning_rate": 0.00018419913419913422, + "loss": 1.8454, "step": 173 }, { "epoch": 0.22, - "grad_norm": 0.48046875, - "learning_rate": 1.922105263157895e-05, - "loss": 2.2888, + "grad_norm": 0.306640625, + "learning_rate": 0.00018398268398268398, + "loss": 1.8028, "step": 174 }, { "epoch": 0.22, - "grad_norm": 0.515625, - "learning_rate": 1.9210526315789474e-05, - "loss": 2.3508, + "grad_norm": 0.28125, + "learning_rate": 0.00018376623376623378, + "loss": 1.9089, "step": 175 }, { "epoch": 0.23, - "grad_norm": 0.47265625, - "learning_rate": 1.9200000000000003e-05, - "loss": 2.2978, + "grad_norm": 0.26953125, + "learning_rate": 0.00018354978354978357, + "loss": 1.8681, "step": 176 }, { "epoch": 0.23, - "grad_norm": 0.443359375, - "learning_rate": 1.918947368421053e-05, - "loss": 2.2199, + "grad_norm": 0.45703125, + "learning_rate": 0.00018333333333333334, + "loss": 1.8012, "step": 177 }, { "epoch": 0.23, - "grad_norm": 0.53125, - "learning_rate": 1.9178947368421055e-05, - "loss": 2.2035, + "grad_norm": 0.376953125, + "learning_rate": 0.00018311688311688313, + "loss": 1.7399, "step": 178 }, { "epoch": 0.23, - "grad_norm": 0.439453125, - "learning_rate": 1.916842105263158e-05, - "loss": 2.2574, + "grad_norm": 0.3125, + "learning_rate": 0.0001829004329004329, + "loss": 1.8339, "step": 179 }, { "epoch": 0.23, - "grad_norm": 0.462890625, - "learning_rate": 1.9157894736842108e-05, - "loss": 2.2449, + "grad_norm": 0.470703125, + "learning_rate": 0.00018268398268398272, + "loss": 1.8236, "step": 180 }, { "epoch": 0.23, - "grad_norm": 0.4375, - "learning_rate": 1.9147368421052632e-05, - "loss": 2.2151, + "grad_norm": 0.466796875, + "learning_rate": 0.00018246753246753248, + "loss": 1.8088, "step": 181 }, { "epoch": 0.23, - "grad_norm": 0.421875, - "learning_rate": 1.913684210526316e-05, - "loss": 2.1975, + "grad_norm": 0.2890625, + "learning_rate": 0.00018225108225108225, + "loss": 1.77, "step": 182 }, { "epoch": 0.23, - "grad_norm": 0.474609375, - "learning_rate": 1.9126315789473685e-05, - "loss": 2.2349, + "grad_norm": 0.453125, + "learning_rate": 0.00018203463203463204, + "loss": 1.7954, "step": 183 }, { "epoch": 0.24, - "grad_norm": 0.458984375, - "learning_rate": 1.9115789473684213e-05, - "loss": 2.223, + "grad_norm": 0.53125, + "learning_rate": 0.00018181818181818183, + "loss": 1.7682, "step": 184 }, { "epoch": 0.24, - "grad_norm": 0.431640625, - "learning_rate": 1.9105263157894738e-05, - "loss": 2.3053, + "grad_norm": 0.248046875, + "learning_rate": 0.00018160173160173163, + "loss": 1.9032, "step": 185 }, { "epoch": 0.24, - "grad_norm": 0.45703125, - "learning_rate": 1.9094736842105262e-05, - "loss": 2.2333, + "grad_norm": 0.2431640625, + "learning_rate": 0.0001813852813852814, + "loss": 1.8145, "step": 186 }, { "epoch": 0.24, - "grad_norm": 0.478515625, - "learning_rate": 1.908421052631579e-05, - "loss": 2.2358, + "grad_norm": 0.482421875, + "learning_rate": 0.0001811688311688312, + "loss": 1.8433, "step": 187 }, { "epoch": 0.24, - "grad_norm": 0.400390625, - "learning_rate": 1.907368421052632e-05, - "loss": 2.1508, + "grad_norm": 0.3515625, + "learning_rate": 0.00018095238095238095, + "loss": 1.7845, "step": 188 }, { "epoch": 0.24, - "grad_norm": 0.46484375, - "learning_rate": 1.9063157894736843e-05, - "loss": 2.2494, + "grad_norm": 0.306640625, + "learning_rate": 0.00018073593073593075, + "loss": 1.8512, "step": 189 }, { "epoch": 0.24, - "grad_norm": 0.4609375, - "learning_rate": 1.9052631578947368e-05, - "loss": 2.2185, + "grad_norm": 0.328125, + "learning_rate": 0.00018051948051948054, + "loss": 1.8208, "step": 190 }, { "epoch": 0.24, - "grad_norm": 0.44140625, - "learning_rate": 1.9042105263157896e-05, - "loss": 2.234, + "grad_norm": 0.28515625, + "learning_rate": 0.0001803030303030303, + "loss": 1.8353, "step": 191 }, { "epoch": 0.25, - "grad_norm": 0.41015625, - "learning_rate": 1.9031578947368424e-05, - "loss": 2.2324, + "grad_norm": 0.251953125, + "learning_rate": 0.0001800865800865801, + "loss": 1.8627, "step": 192 }, { "epoch": 0.25, - "grad_norm": 0.408203125, - "learning_rate": 1.902105263157895e-05, - "loss": 2.1216, + "grad_norm": 0.275390625, + "learning_rate": 0.00017987012987012987, + "loss": 1.7228, "step": 193 }, { "epoch": 0.25, - "grad_norm": 0.42578125, - "learning_rate": 1.9010526315789476e-05, - "loss": 2.2419, + "grad_norm": 0.279296875, + "learning_rate": 0.00017965367965367966, + "loss": 1.878, "step": 194 }, { "epoch": 0.25, - "grad_norm": 0.431640625, - "learning_rate": 1.9e-05, - "loss": 2.1503, + "grad_norm": 0.255859375, + "learning_rate": 0.00017943722943722945, + "loss": 1.7724, "step": 195 }, { "epoch": 0.25, - "grad_norm": 0.41015625, - "learning_rate": 1.898947368421053e-05, - "loss": 2.1745, + "grad_norm": 0.251953125, + "learning_rate": 0.00017922077922077922, + "loss": 1.7845, "step": 196 }, { "epoch": 0.25, - "grad_norm": 0.4453125, - "learning_rate": 1.8978947368421054e-05, - "loss": 2.1631, + "grad_norm": 0.326171875, + "learning_rate": 0.000179004329004329, + "loss": 1.7848, "step": 197 }, { "epoch": 0.25, - "grad_norm": 0.388671875, - "learning_rate": 1.8968421052631582e-05, - "loss": 2.1721, + "grad_norm": 0.265625, + "learning_rate": 0.0001787878787878788, + "loss": 1.808, "step": 198 }, { "epoch": 0.25, - "grad_norm": 0.404296875, - "learning_rate": 1.8957894736842106e-05, - "loss": 2.1645, + "grad_norm": 0.310546875, + "learning_rate": 0.0001785714285714286, + "loss": 1.834, "step": 199 }, { "epoch": 0.26, - "grad_norm": 0.4921875, - "learning_rate": 1.894736842105263e-05, - "loss": 2.1334, + "grad_norm": 0.275390625, + "learning_rate": 0.00017835497835497836, + "loss": 1.7679, "step": 200 }, { "epoch": 0.26, - "eval_loss": 2.1659653186798096, - "eval_runtime": 126.133, - "eval_samples_per_second": 39.641, - "eval_steps_per_second": 1.245, + "eval_loss": 1.8024407625198364, + "eval_runtime": 125.3989, + "eval_samples_per_second": 39.873, + "eval_steps_per_second": 1.252, "step": 200 }, { "epoch": 0.26, - "grad_norm": 0.427734375, - "learning_rate": 1.893684210526316e-05, - "loss": 2.1689, + "grad_norm": 0.283203125, + "learning_rate": 0.00017813852813852813, + "loss": 1.8003, "step": 201 }, { "epoch": 0.26, - "grad_norm": 0.416015625, - "learning_rate": 1.8926315789473687e-05, - "loss": 2.2114, + "grad_norm": 0.267578125, + "learning_rate": 0.00017792207792207792, + "loss": 1.8544, "step": 202 }, { "epoch": 0.26, - "grad_norm": 0.4453125, - "learning_rate": 1.891578947368421e-05, - "loss": 2.2515, + "grad_norm": 0.259765625, + "learning_rate": 0.00017770562770562772, + "loss": 1.8894, "step": 203 }, { "epoch": 0.26, - "grad_norm": 0.439453125, - "learning_rate": 1.8905263157894736e-05, - "loss": 2.2219, + "grad_norm": 0.28125, + "learning_rate": 0.0001774891774891775, + "loss": 1.8496, "step": 204 }, { "epoch": 0.26, - "grad_norm": 0.404296875, - "learning_rate": 1.8894736842105264e-05, - "loss": 2.0661, + "grad_norm": 0.265625, + "learning_rate": 0.00017727272727272728, + "loss": 1.7319, "step": 205 }, { "epoch": 0.26, - "grad_norm": 0.408203125, - "learning_rate": 1.8884210526315792e-05, - "loss": 2.1477, + "grad_norm": 0.29296875, + "learning_rate": 0.00017705627705627707, + "loss": 1.8156, "step": 206 }, { "epoch": 0.26, - "grad_norm": 0.390625, - "learning_rate": 1.8873684210526317e-05, - "loss": 2.0283, + "grad_norm": 0.294921875, + "learning_rate": 0.00017683982683982684, + "loss": 1.6916, "step": 207 }, { "epoch": 0.27, - "grad_norm": 0.396484375, - "learning_rate": 1.886315789473684e-05, - "loss": 2.1791, + "grad_norm": 0.267578125, + "learning_rate": 0.00017662337662337663, + "loss": 1.826, "step": 208 }, { "epoch": 0.27, - "grad_norm": 0.39453125, - "learning_rate": 1.885263157894737e-05, - "loss": 2.1646, + "grad_norm": 0.3046875, + "learning_rate": 0.00017640692640692642, + "loss": 1.8066, "step": 209 }, { "epoch": 0.27, - "grad_norm": 0.361328125, - "learning_rate": 1.8842105263157898e-05, - "loss": 2.1645, + "grad_norm": 0.32421875, + "learning_rate": 0.0001761904761904762, + "loss": 1.8503, "step": 210 }, { "epoch": 0.27, - "grad_norm": 0.3984375, - "learning_rate": 1.8831578947368422e-05, - "loss": 2.1413, + "grad_norm": 0.279296875, + "learning_rate": 0.00017597402597402598, + "loss": 1.8168, "step": 211 }, { "epoch": 0.27, - "grad_norm": 0.373046875, - "learning_rate": 1.882105263157895e-05, - "loss": 2.1087, + "grad_norm": 0.26953125, + "learning_rate": 0.00017575757575757578, + "loss": 1.7903, "step": 212 }, { "epoch": 0.27, - "grad_norm": 0.345703125, - "learning_rate": 1.8810526315789475e-05, - "loss": 2.1136, + "grad_norm": 0.287109375, + "learning_rate": 0.00017554112554112554, + "loss": 1.8139, "step": 213 }, { "epoch": 0.27, - "grad_norm": 0.380859375, - "learning_rate": 1.88e-05, - "loss": 2.1076, + "grad_norm": 0.26171875, + "learning_rate": 0.00017532467532467534, + "loss": 1.7982, "step": 214 }, { "epoch": 0.28, - "grad_norm": 0.37109375, - "learning_rate": 1.8789473684210528e-05, - "loss": 2.119, + "grad_norm": 0.291015625, + "learning_rate": 0.0001751082251082251, + "loss": 1.7828, "step": 215 }, { "epoch": 0.28, - "grad_norm": 0.37890625, - "learning_rate": 1.8778947368421056e-05, - "loss": 2.0844, + "grad_norm": 0.26953125, + "learning_rate": 0.0001748917748917749, + "loss": 1.7416, "step": 216 }, { "epoch": 0.28, - "grad_norm": 0.43359375, - "learning_rate": 1.876842105263158e-05, - "loss": 2.1748, + "grad_norm": 0.34375, + "learning_rate": 0.0001746753246753247, + "loss": 1.806, "step": 217 }, { "epoch": 0.28, - "grad_norm": 0.375, - "learning_rate": 1.8757894736842105e-05, - "loss": 2.2094, + "grad_norm": 0.265625, + "learning_rate": 0.00017445887445887448, + "loss": 1.8822, "step": 218 }, { "epoch": 0.28, - "grad_norm": 0.376953125, - "learning_rate": 1.8747368421052633e-05, - "loss": 2.1159, + "grad_norm": 0.345703125, + "learning_rate": 0.00017424242424242425, + "loss": 1.7818, "step": 219 }, { "epoch": 0.28, - "grad_norm": 0.353515625, - "learning_rate": 1.873684210526316e-05, - "loss": 2.1141, + "grad_norm": 0.275390625, + "learning_rate": 0.00017402597402597401, + "loss": 1.843, "step": 220 }, { "epoch": 0.28, - "grad_norm": 0.435546875, - "learning_rate": 1.8726315789473686e-05, - "loss": 2.1565, + "grad_norm": 0.33984375, + "learning_rate": 0.00017380952380952383, + "loss": 1.8087, "step": 221 }, { "epoch": 0.28, - "grad_norm": 0.3828125, - "learning_rate": 1.871578947368421e-05, - "loss": 2.1474, + "grad_norm": 0.27734375, + "learning_rate": 0.0001735930735930736, + "loss": 1.8346, "step": 222 }, { "epoch": 0.29, - "grad_norm": 0.33984375, - "learning_rate": 1.8705263157894738e-05, - "loss": 2.0948, + "grad_norm": 0.296875, + "learning_rate": 0.0001733766233766234, + "loss": 1.7843, "step": 223 }, { "epoch": 0.29, - "grad_norm": 0.314453125, - "learning_rate": 1.8694736842105266e-05, - "loss": 2.0956, + "grad_norm": 0.330078125, + "learning_rate": 0.00017316017316017316, + "loss": 1.7974, "step": 224 }, { "epoch": 0.29, - "grad_norm": 0.369140625, - "learning_rate": 1.868421052631579e-05, - "loss": 2.0801, + "grad_norm": 0.26953125, + "learning_rate": 0.00017294372294372295, + "loss": 1.7798, "step": 225 }, { "epoch": 0.29, - "grad_norm": 0.40625, - "learning_rate": 1.8673684210526316e-05, - "loss": 2.0134, + "grad_norm": 0.287109375, + "learning_rate": 0.00017272727272727275, + "loss": 1.7318, "step": 226 }, { "epoch": 0.29, - "grad_norm": 0.3828125, - "learning_rate": 1.8663157894736844e-05, - "loss": 2.1064, + "grad_norm": 0.27734375, + "learning_rate": 0.0001725108225108225, + "loss": 1.8244, "step": 227 }, { "epoch": 0.29, - "grad_norm": 0.388671875, - "learning_rate": 1.8652631578947368e-05, - "loss": 2.1532, + "grad_norm": 0.345703125, + "learning_rate": 0.0001722943722943723, + "loss": 1.8291, "step": 228 }, { "epoch": 0.29, - "grad_norm": 0.34765625, - "learning_rate": 1.8642105263157896e-05, - "loss": 2.0242, + "grad_norm": 0.275390625, + "learning_rate": 0.00017207792207792207, + "loss": 1.7127, "step": 229 }, { "epoch": 0.29, - "grad_norm": 0.33203125, - "learning_rate": 1.8631578947368424e-05, - "loss": 2.1449, + "grad_norm": 0.27734375, + "learning_rate": 0.00017186147186147187, + "loss": 1.8416, "step": 230 }, { "epoch": 0.3, - "grad_norm": 0.359375, - "learning_rate": 1.862105263157895e-05, - "loss": 2.1205, + "grad_norm": 0.283203125, + "learning_rate": 0.00017164502164502166, + "loss": 1.8327, "step": 231 }, { "epoch": 0.3, - "grad_norm": 0.376953125, - "learning_rate": 1.8610526315789473e-05, - "loss": 2.0216, + "grad_norm": 0.29296875, + "learning_rate": 0.00017142857142857143, + "loss": 1.7058, "step": 232 }, { "epoch": 0.3, - "grad_norm": 0.34375, - "learning_rate": 1.86e-05, - "loss": 2.0181, + "grad_norm": 0.3046875, + "learning_rate": 0.00017121212121212122, + "loss": 1.7438, "step": 233 }, { "epoch": 0.3, - "grad_norm": 0.337890625, - "learning_rate": 1.858947368421053e-05, - "loss": 2.1133, + "grad_norm": 0.298828125, + "learning_rate": 0.00017099567099567098, + "loss": 1.822, "step": 234 }, { "epoch": 0.3, - "grad_norm": 0.396484375, - "learning_rate": 1.8578947368421054e-05, - "loss": 2.1258, + "grad_norm": 0.275390625, + "learning_rate": 0.0001707792207792208, + "loss": 1.8308, "step": 235 }, { "epoch": 0.3, - "grad_norm": 0.412109375, - "learning_rate": 1.856842105263158e-05, - "loss": 2.1206, + "grad_norm": 0.283203125, + "learning_rate": 0.00017056277056277057, + "loss": 1.8354, "step": 236 }, { "epoch": 0.3, - "grad_norm": 0.306640625, - "learning_rate": 1.8557894736842107e-05, - "loss": 2.0442, + "grad_norm": 0.265625, + "learning_rate": 0.00017034632034632036, + "loss": 1.7753, "step": 237 }, { "epoch": 0.3, - "grad_norm": 0.318359375, - "learning_rate": 1.8547368421052635e-05, - "loss": 2.1067, + "grad_norm": 0.337890625, + "learning_rate": 0.00017012987012987013, + "loss": 1.8091, "step": 238 }, { "epoch": 0.31, - "grad_norm": 0.296875, - "learning_rate": 1.853684210526316e-05, - "loss": 2.1295, + "grad_norm": 0.3046875, + "learning_rate": 0.00016991341991341992, + "loss": 1.8526, "step": 239 }, { "epoch": 0.31, - "grad_norm": 0.373046875, - "learning_rate": 1.8526315789473684e-05, - "loss": 2.1436, + "grad_norm": 0.33984375, + "learning_rate": 0.00016969696969696972, + "loss": 1.8296, "step": 240 }, { "epoch": 0.31, - "grad_norm": 0.27734375, - "learning_rate": 1.8515789473684212e-05, - "loss": 2.123, + "grad_norm": 0.322265625, + "learning_rate": 0.00016948051948051948, + "loss": 1.8309, "step": 241 }, { "epoch": 0.31, - "grad_norm": 0.359375, - "learning_rate": 1.8505263157894737e-05, - "loss": 2.0278, + "grad_norm": 0.310546875, + "learning_rate": 0.00016926406926406928, + "loss": 1.7423, "step": 242 }, { "epoch": 0.31, - "grad_norm": 0.2890625, - "learning_rate": 1.8494736842105265e-05, - "loss": 2.031, + "grad_norm": 0.310546875, + "learning_rate": 0.00016904761904761904, + "loss": 1.7644, "step": 243 }, { "epoch": 0.31, - "grad_norm": 0.328125, - "learning_rate": 1.8484210526315793e-05, - "loss": 2.0259, + "grad_norm": 0.30859375, + "learning_rate": 0.00016883116883116884, + "loss": 1.7392, "step": 244 }, { "epoch": 0.31, - "grad_norm": 0.275390625, - "learning_rate": 1.8473684210526317e-05, - "loss": 2.1168, + "grad_norm": 0.349609375, + "learning_rate": 0.00016861471861471863, + "loss": 1.8515, "step": 245 }, { "epoch": 0.31, - "grad_norm": 0.294921875, - "learning_rate": 1.8463157894736842e-05, - "loss": 2.028, + "grad_norm": 0.30859375, + "learning_rate": 0.0001683982683982684, + "loss": 1.751, "step": 246 }, { "epoch": 0.32, - "grad_norm": 0.326171875, - "learning_rate": 1.845263157894737e-05, - "loss": 2.0843, + "grad_norm": 0.3046875, + "learning_rate": 0.0001681818181818182, + "loss": 1.8196, "step": 247 }, { "epoch": 0.32, - "grad_norm": 0.2734375, - "learning_rate": 1.8442105263157898e-05, - "loss": 2.1167, + "grad_norm": 0.28515625, + "learning_rate": 0.00016796536796536798, + "loss": 1.8556, "step": 248 }, { "epoch": 0.32, - "grad_norm": 0.302734375, - "learning_rate": 1.8431578947368423e-05, - "loss": 2.0787, + "grad_norm": 0.390625, + "learning_rate": 0.00016774891774891778, + "loss": 1.7934, "step": 249 }, { "epoch": 0.32, - "grad_norm": 0.28515625, - "learning_rate": 1.8421052631578947e-05, - "loss": 2.0313, + "grad_norm": 0.29296875, + "learning_rate": 0.00016753246753246754, + "loss": 1.753, "step": 250 }, { "epoch": 0.32, - "grad_norm": 0.318359375, - "learning_rate": 1.8410526315789475e-05, - "loss": 2.1279, + "eval_loss": 1.7896223068237305, + "eval_runtime": 125.3191, + "eval_samples_per_second": 39.898, + "eval_steps_per_second": 1.253, + "step": 250 + }, + { + "epoch": 0.32, + "grad_norm": 0.2890625, + "learning_rate": 0.0001673160173160173, + "loss": 1.8485, "step": 251 }, { "epoch": 0.32, - "grad_norm": 0.3203125, - "learning_rate": 1.8400000000000003e-05, - "loss": 2.0327, + "grad_norm": 0.33984375, + "learning_rate": 0.0001670995670995671, + "loss": 1.7847, "step": 252 }, { "epoch": 0.32, - "grad_norm": 0.2734375, - "learning_rate": 1.8389473684210528e-05, - "loss": 1.9418, + "grad_norm": 0.294921875, + "learning_rate": 0.0001668831168831169, + "loss": 1.688, "step": 253 }, { "epoch": 0.33, - "grad_norm": 0.310546875, - "learning_rate": 1.8378947368421053e-05, - "loss": 1.9637, + "grad_norm": 0.271484375, + "learning_rate": 0.0001666666666666667, + "loss": 1.7079, "step": 254 }, { "epoch": 0.33, - "grad_norm": 0.2578125, - "learning_rate": 1.836842105263158e-05, - "loss": 2.0559, + "grad_norm": 0.33203125, + "learning_rate": 0.00016645021645021645, + "loss": 1.7858, "step": 255 }, { "epoch": 0.33, - "grad_norm": 0.271484375, - "learning_rate": 1.8357894736842105e-05, - "loss": 2.0528, + "grad_norm": 0.310546875, + "learning_rate": 0.00016623376623376625, + "loss": 1.8125, "step": 256 }, { "epoch": 0.33, - "grad_norm": 0.3359375, - "learning_rate": 1.8347368421052633e-05, - "loss": 2.0434, + "grad_norm": 0.32421875, + "learning_rate": 0.00016601731601731601, + "loss": 1.7826, "step": 257 }, { "epoch": 0.33, - "grad_norm": 0.3046875, - "learning_rate": 1.8336842105263158e-05, - "loss": 2.0086, + "grad_norm": 0.3515625, + "learning_rate": 0.0001658008658008658, + "loss": 1.7352, "step": 258 }, { "epoch": 0.33, - "grad_norm": 0.361328125, - "learning_rate": 1.8326315789473686e-05, - "loss": 2.0458, + "grad_norm": 0.357421875, + "learning_rate": 0.0001655844155844156, + "loss": 1.7863, "step": 259 }, { "epoch": 0.33, - "grad_norm": 0.302734375, - "learning_rate": 1.831578947368421e-05, - "loss": 2.0623, + "grad_norm": 0.283203125, + "learning_rate": 0.00016536796536796537, + "loss": 1.7868, "step": 260 }, { "epoch": 0.33, - "grad_norm": 0.2734375, - "learning_rate": 1.830526315789474e-05, - "loss": 2.0518, + "grad_norm": 0.279296875, + "learning_rate": 0.00016515151515151516, + "loss": 1.7958, "step": 261 }, { "epoch": 0.34, - "grad_norm": 0.291015625, - "learning_rate": 1.8294736842105267e-05, - "loss": 2.0208, + "grad_norm": 0.29296875, + "learning_rate": 0.00016493506493506495, + "loss": 1.7685, "step": 262 }, { "epoch": 0.34, - "grad_norm": 0.30078125, - "learning_rate": 1.828421052631579e-05, - "loss": 1.998, + "grad_norm": 0.279296875, + "learning_rate": 0.00016471861471861472, + "loss": 1.7561, "step": 263 }, { "epoch": 0.34, - "grad_norm": 0.30078125, - "learning_rate": 1.8273684210526316e-05, - "loss": 2.0385, + "grad_norm": 0.2890625, + "learning_rate": 0.0001645021645021645, + "loss": 1.7634, "step": 264 }, { "epoch": 0.34, - "grad_norm": 0.26953125, - "learning_rate": 1.8263157894736844e-05, - "loss": 2.0674, + "grad_norm": 0.296875, + "learning_rate": 0.00016428571428571428, + "loss": 1.8092, "step": 265 }, { "epoch": 0.34, - "grad_norm": 0.2734375, - "learning_rate": 1.8252631578947372e-05, - "loss": 2.1062, + "grad_norm": 0.31640625, + "learning_rate": 0.00016406926406926407, + "loss": 1.8324, "step": 266 }, { "epoch": 0.34, - "grad_norm": 0.294921875, - "learning_rate": 1.8242105263157897e-05, - "loss": 1.9543, + "grad_norm": 0.28125, + "learning_rate": 0.00016385281385281387, + "loss": 1.7221, "step": 267 }, { "epoch": 0.34, - "grad_norm": 0.26171875, - "learning_rate": 1.823157894736842e-05, - "loss": 2.0033, + "grad_norm": 0.298828125, + "learning_rate": 0.00016363636363636366, + "loss": 1.7796, "step": 268 }, { "epoch": 0.34, - "grad_norm": 0.28125, - "learning_rate": 1.822105263157895e-05, - "loss": 2.0031, + "grad_norm": 0.302734375, + "learning_rate": 0.00016341991341991343, + "loss": 1.7713, "step": 269 }, { "epoch": 0.35, - "grad_norm": 0.3046875, - "learning_rate": 1.8210526315789477e-05, - "loss": 2.0491, + "grad_norm": 0.265625, + "learning_rate": 0.0001632034632034632, + "loss": 1.8178, "step": 270 }, { "epoch": 0.35, - "grad_norm": 0.2578125, - "learning_rate": 1.8200000000000002e-05, - "loss": 2.0505, + "grad_norm": 0.404296875, + "learning_rate": 0.000162987012987013, + "loss": 1.8104, "step": 271 }, { "epoch": 0.35, - "grad_norm": 0.27734375, - "learning_rate": 1.8189473684210527e-05, - "loss": 2.0931, + "grad_norm": 0.3125, + "learning_rate": 0.00016277056277056278, + "loss": 1.8387, "step": 272 }, { "epoch": 0.35, - "grad_norm": 0.294921875, - "learning_rate": 1.8178947368421055e-05, - "loss": 2.1664, + "grad_norm": 0.310546875, + "learning_rate": 0.00016255411255411257, + "loss": 1.8897, "step": 273 }, { "epoch": 0.35, - "grad_norm": 0.2734375, - "learning_rate": 1.816842105263158e-05, - "loss": 2.0588, + "grad_norm": 0.28125, + "learning_rate": 0.00016233766233766234, + "loss": 1.8162, "step": 274 }, { "epoch": 0.35, - "grad_norm": 0.3359375, - "learning_rate": 1.8157894736842107e-05, - "loss": 2.0858, + "grad_norm": 0.40625, + "learning_rate": 0.00016212121212121213, + "loss": 1.8086, "step": 275 }, { "epoch": 0.35, - "grad_norm": 0.259765625, - "learning_rate": 1.8147368421052632e-05, - "loss": 2.1721, + "grad_norm": 0.283203125, + "learning_rate": 0.00016190476190476192, + "loss": 1.9186, "step": 276 }, { "epoch": 0.35, - "grad_norm": 0.26953125, - "learning_rate": 1.813684210526316e-05, - "loss": 2.049, + "grad_norm": 0.345703125, + "learning_rate": 0.0001616883116883117, + "loss": 1.8123, "step": 277 }, { "epoch": 0.36, - "grad_norm": 0.251953125, - "learning_rate": 1.8126315789473685e-05, - "loss": 2.0509, + "grad_norm": 0.27734375, + "learning_rate": 0.00016147186147186148, + "loss": 1.8115, "step": 278 }, { "epoch": 0.36, - "grad_norm": 0.29296875, - "learning_rate": 1.8115789473684213e-05, - "loss": 2.1458, + "grad_norm": 0.33203125, + "learning_rate": 0.00016125541125541125, + "loss": 1.8785, "step": 279 }, { "epoch": 0.36, - "grad_norm": 0.25, - "learning_rate": 1.810526315789474e-05, - "loss": 2.049, + "grad_norm": 0.3125, + "learning_rate": 0.00016103896103896104, + "loss": 1.805, "step": 280 }, { "epoch": 0.36, - "grad_norm": 0.2734375, - "learning_rate": 1.8094736842105265e-05, - "loss": 2.0903, + "grad_norm": 0.3046875, + "learning_rate": 0.00016082251082251084, + "loss": 1.8656, "step": 281 }, { "epoch": 0.36, - "grad_norm": 0.302734375, - "learning_rate": 1.808421052631579e-05, - "loss": 2.0443, + "grad_norm": 0.3515625, + "learning_rate": 0.0001606060606060606, + "loss": 1.7848, "step": 282 }, { "epoch": 0.36, - "grad_norm": 0.318359375, - "learning_rate": 1.8073684210526318e-05, - "loss": 2.0505, + "grad_norm": 0.3125, + "learning_rate": 0.0001603896103896104, + "loss": 1.7817, "step": 283 }, { "epoch": 0.36, - "grad_norm": 0.392578125, - "learning_rate": 1.8063157894736846e-05, - "loss": 2.0556, + "grad_norm": 0.30078125, + "learning_rate": 0.00016017316017316016, + "loss": 1.8217, "step": 284 }, { "epoch": 0.36, - "grad_norm": 0.28515625, - "learning_rate": 1.805263157894737e-05, - "loss": 2.1208, + "grad_norm": 0.318359375, + "learning_rate": 0.00015995670995670998, + "loss": 1.8771, "step": 285 }, { "epoch": 0.37, - "grad_norm": 0.28125, - "learning_rate": 1.8042105263157895e-05, - "loss": 2.0509, + "grad_norm": 0.31640625, + "learning_rate": 0.00015974025974025975, + "loss": 1.8028, "step": 286 }, { "epoch": 0.37, - "grad_norm": 0.279296875, - "learning_rate": 1.8031578947368423e-05, - "loss": 2.0647, + "grad_norm": 0.291015625, + "learning_rate": 0.00015952380952380954, + "loss": 1.8331, "step": 287 }, { "epoch": 0.37, - "grad_norm": 0.2890625, - "learning_rate": 1.8021052631578948e-05, - "loss": 2.0011, + "grad_norm": 0.357421875, + "learning_rate": 0.0001593073593073593, + "loss": 1.7528, "step": 288 }, { "epoch": 0.37, - "grad_norm": 0.259765625, - "learning_rate": 1.8010526315789476e-05, - "loss": 2.0282, + "grad_norm": 0.267578125, + "learning_rate": 0.0001590909090909091, + "loss": 1.8012, "step": 289 }, { "epoch": 0.37, - "grad_norm": 0.30859375, - "learning_rate": 1.8e-05, - "loss": 2.0263, + "grad_norm": 0.2734375, + "learning_rate": 0.0001588744588744589, + "loss": 1.7787, "step": 290 }, { "epoch": 0.37, - "grad_norm": 0.296875, - "learning_rate": 1.798947368421053e-05, - "loss": 2.0415, + "grad_norm": 0.3125, + "learning_rate": 0.00015865800865800866, + "loss": 1.803, "step": 291 }, { "epoch": 0.37, - "grad_norm": 0.296875, - "learning_rate": 1.7978947368421053e-05, - "loss": 1.9791, + "grad_norm": 0.4375, + "learning_rate": 0.00015844155844155845, + "loss": 1.7466, "step": 292 }, { "epoch": 0.37, - "grad_norm": 0.263671875, - "learning_rate": 1.7968421052631578e-05, - "loss": 2.1071, + "grad_norm": 0.30078125, + "learning_rate": 0.00015822510822510822, + "loss": 1.8551, "step": 293 }, { "epoch": 0.38, - "grad_norm": 0.208984375, - "learning_rate": 1.795789473684211e-05, - "loss": 2.0227, + "grad_norm": 0.279296875, + "learning_rate": 0.00015800865800865801, + "loss": 1.7932, "step": 294 }, { "epoch": 0.38, - "grad_norm": 0.30078125, - "learning_rate": 1.7947368421052634e-05, - "loss": 1.9584, + "grad_norm": 0.359375, + "learning_rate": 0.0001577922077922078, + "loss": 1.7295, "step": 295 }, { "epoch": 0.38, - "grad_norm": 0.271484375, - "learning_rate": 1.793684210526316e-05, - "loss": 2.0451, + "grad_norm": 0.318359375, + "learning_rate": 0.00015757575757575757, + "loss": 1.7984, "step": 296 }, { "epoch": 0.38, - "grad_norm": 0.310546875, - "learning_rate": 1.7926315789473686e-05, - "loss": 2.1113, + "grad_norm": 0.40234375, + "learning_rate": 0.00015735930735930737, + "loss": 1.8666, "step": 297 }, { "epoch": 0.38, - "grad_norm": 0.267578125, - "learning_rate": 1.7915789473684214e-05, - "loss": 1.9366, + "grad_norm": 0.287109375, + "learning_rate": 0.00015714285714285716, + "loss": 1.7149, "step": 298 }, { "epoch": 0.38, - "grad_norm": 0.255859375, - "learning_rate": 1.790526315789474e-05, - "loss": 1.9567, + "grad_norm": 0.462890625, + "learning_rate": 0.00015692640692640695, + "loss": 1.7447, "step": 299 }, { "epoch": 0.38, - "grad_norm": 0.2890625, - "learning_rate": 1.7894736842105264e-05, - "loss": 2.0307, + "grad_norm": 0.4296875, + "learning_rate": 0.00015670995670995672, + "loss": 1.8199, + "step": 300 + }, + { + "epoch": 0.38, + "eval_loss": 1.7805144786834717, + "eval_runtime": 125.5041, + "eval_samples_per_second": 39.839, + "eval_steps_per_second": 1.251, "step": 300 }, { "epoch": 0.39, - "grad_norm": 0.263671875, - "learning_rate": 1.7884210526315792e-05, - "loss": 2.0251, + "grad_norm": 0.29296875, + "learning_rate": 0.00015649350649350649, + "loss": 1.7937, "step": 301 }, { "epoch": 0.39, - "grad_norm": 0.279296875, - "learning_rate": 1.7873684210526316e-05, - "loss": 2.0157, + "grad_norm": 0.400390625, + "learning_rate": 0.00015627705627705628, + "loss": 1.7617, "step": 302 }, { "epoch": 0.39, - "grad_norm": 0.341796875, - "learning_rate": 1.7863157894736844e-05, - "loss": 2.0085, + "grad_norm": 0.435546875, + "learning_rate": 0.00015606060606060607, + "loss": 1.7755, "step": 303 }, { "epoch": 0.39, - "grad_norm": 0.2333984375, - "learning_rate": 1.785263157894737e-05, - "loss": 2.0893, + "grad_norm": 0.326171875, + "learning_rate": 0.00015584415584415587, + "loss": 1.862, "step": 304 }, { "epoch": 0.39, - "grad_norm": 0.26171875, - "learning_rate": 1.7842105263157897e-05, - "loss": 2.0839, + "grad_norm": 0.365234375, + "learning_rate": 0.00015562770562770563, + "loss": 1.8572, "step": 305 }, { "epoch": 0.39, - "grad_norm": 0.244140625, - "learning_rate": 1.7831578947368422e-05, - "loss": 1.9943, + "grad_norm": 0.330078125, + "learning_rate": 0.00015541125541125543, + "loss": 1.7678, "step": 306 }, { "epoch": 0.39, - "grad_norm": 0.240234375, - "learning_rate": 1.7821052631578946e-05, - "loss": 1.9827, + "grad_norm": 0.359375, + "learning_rate": 0.0001551948051948052, + "loss": 1.7435, "step": 307 }, { "epoch": 0.39, - "grad_norm": 0.228515625, - "learning_rate": 1.7810526315789474e-05, - "loss": 1.965, + "grad_norm": 0.326171875, + "learning_rate": 0.00015497835497835498, + "loss": 1.7535, "step": 308 }, { "epoch": 0.4, - "grad_norm": 0.40234375, - "learning_rate": 1.7800000000000002e-05, - "loss": 1.9748, + "grad_norm": 0.39453125, + "learning_rate": 0.00015476190476190478, + "loss": 1.7495, "step": 309 }, { "epoch": 0.4, - "grad_norm": 0.283203125, - "learning_rate": 1.7789473684210527e-05, - "loss": 1.9507, + "grad_norm": 0.28515625, + "learning_rate": 0.00015454545454545454, + "loss": 1.7107, "step": 310 }, { "epoch": 0.4, - "grad_norm": 0.314453125, - "learning_rate": 1.7778947368421055e-05, - "loss": 1.9727, + "grad_norm": 0.298828125, + "learning_rate": 0.00015432900432900434, + "loss": 1.7419, "step": 311 }, { "epoch": 0.4, - "grad_norm": 0.25, - "learning_rate": 1.7768421052631583e-05, - "loss": 1.9944, + "grad_norm": 0.296875, + "learning_rate": 0.00015411255411255413, + "loss": 1.7466, "step": 312 }, { "epoch": 0.4, - "grad_norm": 0.244140625, - "learning_rate": 1.7757894736842108e-05, - "loss": 2.0249, + "grad_norm": 0.3046875, + "learning_rate": 0.0001538961038961039, + "loss": 1.8045, "step": 313 }, { "epoch": 0.4, - "grad_norm": 0.265625, - "learning_rate": 1.7747368421052632e-05, - "loss": 1.9949, + "grad_norm": 0.369140625, + "learning_rate": 0.0001536796536796537, + "loss": 1.7756, "step": 314 }, { "epoch": 0.4, - "grad_norm": 0.2392578125, - "learning_rate": 1.773684210526316e-05, - "loss": 2.0334, + "grad_norm": 0.35546875, + "learning_rate": 0.00015346320346320346, + "loss": 1.8018, "step": 315 }, { "epoch": 0.4, - "grad_norm": 0.2451171875, - "learning_rate": 1.7726315789473685e-05, - "loss": 2.0362, + "grad_norm": 0.287109375, + "learning_rate": 0.00015324675324675325, + "loss": 1.8246, "step": 316 }, { "epoch": 0.41, - "grad_norm": 0.26171875, - "learning_rate": 1.7715789473684213e-05, - "loss": 1.9895, + "grad_norm": 0.28515625, + "learning_rate": 0.00015303030303030304, + "loss": 1.761, "step": 317 }, { "epoch": 0.41, - "grad_norm": 0.26171875, - "learning_rate": 1.7705263157894738e-05, - "loss": 1.9843, + "grad_norm": 0.29296875, + "learning_rate": 0.00015281385281385284, + "loss": 1.773, "step": 318 }, { "epoch": 0.41, - "grad_norm": 0.271484375, - "learning_rate": 1.7694736842105266e-05, - "loss": 2.0213, + "grad_norm": 0.345703125, + "learning_rate": 0.0001525974025974026, + "loss": 1.7889, "step": 319 }, { "epoch": 0.41, - "grad_norm": 0.283203125, - "learning_rate": 1.768421052631579e-05, - "loss": 2.0384, + "grad_norm": 0.38671875, + "learning_rate": 0.00015238095238095237, + "loss": 1.827, "step": 320 }, { "epoch": 0.41, - "grad_norm": 0.296875, - "learning_rate": 1.7673684210526315e-05, - "loss": 2.0249, + "grad_norm": 0.333984375, + "learning_rate": 0.0001521645021645022, + "loss": 1.7928, "step": 321 }, { "epoch": 0.41, - "grad_norm": 0.240234375, - "learning_rate": 1.7663157894736843e-05, - "loss": 1.9845, + "grad_norm": 0.33203125, + "learning_rate": 0.00015194805194805196, + "loss": 1.7618, "step": 322 }, { "epoch": 0.41, - "grad_norm": 0.263671875, - "learning_rate": 1.765263157894737e-05, - "loss": 1.9788, + "grad_norm": 0.36328125, + "learning_rate": 0.00015173160173160175, + "loss": 1.7576, "step": 323 }, { "epoch": 0.41, - "grad_norm": 0.248046875, - "learning_rate": 1.7642105263157896e-05, - "loss": 1.8875, + "grad_norm": 0.2890625, + "learning_rate": 0.00015151515151515152, + "loss": 1.6683, "step": 324 }, { "epoch": 0.42, - "grad_norm": 0.25390625, - "learning_rate": 1.763157894736842e-05, - "loss": 2.0243, + "grad_norm": 0.27734375, + "learning_rate": 0.0001512987012987013, + "loss": 1.7967, "step": 325 }, { "epoch": 0.42, - "grad_norm": 0.287109375, - "learning_rate": 1.7621052631578948e-05, - "loss": 2.0382, + "grad_norm": 0.28125, + "learning_rate": 0.0001510822510822511, + "loss": 1.8209, "step": 326 }, { "epoch": 0.42, - "grad_norm": 0.275390625, - "learning_rate": 1.7610526315789476e-05, - "loss": 2.0997, + "grad_norm": 0.39453125, + "learning_rate": 0.00015086580086580087, + "loss": 1.8743, "step": 327 }, { "epoch": 0.42, - "grad_norm": 0.3515625, - "learning_rate": 1.76e-05, - "loss": 2.0718, + "grad_norm": 0.27734375, + "learning_rate": 0.00015064935064935066, + "loss": 1.8559, "step": 328 }, { "epoch": 0.42, - "grad_norm": 0.259765625, - "learning_rate": 1.758947368421053e-05, - "loss": 1.9986, + "grad_norm": 0.337890625, + "learning_rate": 0.00015043290043290043, + "loss": 1.7791, "step": 329 }, { "epoch": 0.42, - "grad_norm": 0.263671875, - "learning_rate": 1.7578947368421054e-05, - "loss": 2.0703, + "grad_norm": 0.353515625, + "learning_rate": 0.00015021645021645022, + "loss": 1.8626, "step": 330 }, { "epoch": 0.42, - "grad_norm": 0.2451171875, - "learning_rate": 1.756842105263158e-05, - "loss": 2.025, + "grad_norm": 0.4140625, + "learning_rate": 0.00015000000000000001, + "loss": 1.8169, "step": 331 }, { "epoch": 0.42, - "grad_norm": 0.224609375, - "learning_rate": 1.7557894736842106e-05, - "loss": 2.0197, + "grad_norm": 0.373046875, + "learning_rate": 0.00014978354978354978, + "loss": 1.8006, "step": 332 }, { "epoch": 0.43, - "grad_norm": 0.248046875, - "learning_rate": 1.7547368421052634e-05, - "loss": 2.1094, + "grad_norm": 0.302734375, + "learning_rate": 0.00014956709956709957, + "loss": 1.8818, "step": 333 }, { "epoch": 0.43, - "grad_norm": 0.31640625, - "learning_rate": 1.753684210526316e-05, - "loss": 2.01, + "grad_norm": 0.423828125, + "learning_rate": 0.00014935064935064934, + "loss": 1.7898, "step": 334 }, { "epoch": 0.43, - "grad_norm": 0.236328125, - "learning_rate": 1.7526315789473683e-05, - "loss": 1.9842, + "grad_norm": 0.314453125, + "learning_rate": 0.00014913419913419916, + "loss": 1.7639, "step": 335 }, { "epoch": 0.43, - "grad_norm": 0.30078125, - "learning_rate": 1.751578947368421e-05, - "loss": 2.0515, + "grad_norm": 0.375, + "learning_rate": 0.00014891774891774893, + "loss": 1.8276, "step": 336 }, { "epoch": 0.43, - "grad_norm": 0.2470703125, - "learning_rate": 1.750526315789474e-05, - "loss": 2.0888, + "grad_norm": 0.34375, + "learning_rate": 0.00014870129870129872, + "loss": 1.8718, "step": 337 }, { "epoch": 0.43, - "grad_norm": 0.2451171875, - "learning_rate": 1.7494736842105264e-05, - "loss": 1.9125, + "grad_norm": 0.310546875, + "learning_rate": 0.00014848484848484849, + "loss": 1.7142, "step": 338 }, { "epoch": 0.43, - "grad_norm": 0.2431640625, - "learning_rate": 1.748421052631579e-05, - "loss": 2.0572, + "grad_norm": 0.330078125, + "learning_rate": 0.00014826839826839828, + "loss": 1.8261, "step": 339 }, { "epoch": 0.44, - "grad_norm": 0.27734375, - "learning_rate": 1.7473684210526317e-05, - "loss": 2.0348, + "grad_norm": 0.328125, + "learning_rate": 0.00014805194805194807, + "loss": 1.8177, "step": 340 }, { "epoch": 0.44, - "grad_norm": 0.251953125, - "learning_rate": 1.7463157894736845e-05, - "loss": 1.9546, + "grad_norm": 0.357421875, + "learning_rate": 0.00014783549783549784, + "loss": 1.739, "step": 341 }, { "epoch": 0.44, - "grad_norm": 0.255859375, - "learning_rate": 1.745263157894737e-05, - "loss": 1.9928, + "grad_norm": 0.291015625, + "learning_rate": 0.00014761904761904763, + "loss": 1.7867, "step": 342 }, { "epoch": 0.44, - "grad_norm": 0.23046875, - "learning_rate": 1.7442105263157894e-05, - "loss": 1.9963, + "grad_norm": 0.380859375, + "learning_rate": 0.0001474025974025974, + "loss": 1.7825, "step": 343 }, { "epoch": 0.44, - "grad_norm": 0.263671875, - "learning_rate": 1.7431578947368422e-05, - "loss": 2.0041, + "grad_norm": 0.29296875, + "learning_rate": 0.0001471861471861472, + "loss": 1.7784, "step": 344 }, { "epoch": 0.44, - "grad_norm": 0.251953125, - "learning_rate": 1.742105263157895e-05, - "loss": 2.0166, + "grad_norm": 0.2890625, + "learning_rate": 0.00014696969696969698, + "loss": 1.8077, "step": 345 }, { "epoch": 0.44, - "grad_norm": 0.236328125, - "learning_rate": 1.7410526315789475e-05, - "loss": 1.9628, + "grad_norm": 0.296875, + "learning_rate": 0.00014675324675324675, + "loss": 1.7616, "step": 346 }, { "epoch": 0.44, - "grad_norm": 0.263671875, - "learning_rate": 1.7400000000000003e-05, - "loss": 1.9719, + "grad_norm": 0.2890625, + "learning_rate": 0.00014653679653679654, + "loss": 1.7738, "step": 347 }, { "epoch": 0.45, - "grad_norm": 0.3046875, - "learning_rate": 1.7389473684210527e-05, - "loss": 1.9378, + "grad_norm": 0.345703125, + "learning_rate": 0.00014632034632034634, + "loss": 1.7019, "step": 348 }, { "epoch": 0.45, - "grad_norm": 0.2578125, - "learning_rate": 1.7378947368421052e-05, - "loss": 1.9047, + "grad_norm": 0.33984375, + "learning_rate": 0.00014610389610389613, + "loss": 1.6911, "step": 349 }, { "epoch": 0.45, - "grad_norm": 0.2451171875, - "learning_rate": 1.736842105263158e-05, - "loss": 1.991, + "grad_norm": 0.3359375, + "learning_rate": 0.0001458874458874459, + "loss": 1.7708, "step": 350 }, { "epoch": 0.45, - "grad_norm": 0.326171875, - "learning_rate": 1.7357894736842108e-05, - "loss": 1.9697, + "eval_loss": 1.7709890604019165, + "eval_runtime": 125.3163, + "eval_samples_per_second": 39.899, + "eval_steps_per_second": 1.253, + "step": 350 + }, + { + "epoch": 0.45, + "grad_norm": 0.349609375, + "learning_rate": 0.00014567099567099566, + "loss": 1.76, "step": 351 }, { "epoch": 0.45, - "grad_norm": 0.255859375, - "learning_rate": 1.7347368421052633e-05, - "loss": 2.0127, + "grad_norm": 0.38671875, + "learning_rate": 0.00014545454545454546, + "loss": 1.7913, "step": 352 }, { "epoch": 0.45, - "grad_norm": 0.248046875, - "learning_rate": 1.7336842105263157e-05, - "loss": 2.0216, + "grad_norm": 0.421875, + "learning_rate": 0.00014523809523809525, + "loss": 1.8083, "step": 353 }, { "epoch": 0.45, - "grad_norm": 0.21875, - "learning_rate": 1.7326315789473685e-05, - "loss": 1.9968, + "grad_norm": 0.388671875, + "learning_rate": 0.00014502164502164504, + "loss": 1.7886, "step": 354 }, { "epoch": 0.45, - "grad_norm": 0.2333984375, - "learning_rate": 1.7315789473684213e-05, - "loss": 2.0262, + "grad_norm": 0.265625, + "learning_rate": 0.0001448051948051948, + "loss": 1.8125, "step": 355 }, { "epoch": 0.46, - "grad_norm": 0.33203125, - "learning_rate": 1.7305263157894738e-05, - "loss": 1.928, + "grad_norm": 0.353515625, + "learning_rate": 0.00014458874458874458, + "loss": 1.6922, "step": 356 }, { "epoch": 0.46, - "grad_norm": 0.2490234375, - "learning_rate": 1.7294736842105263e-05, - "loss": 1.8397, + "grad_norm": 0.41796875, + "learning_rate": 0.00014437229437229437, + "loss": 1.6483, "step": 357 }, { "epoch": 0.46, - "grad_norm": 0.2890625, - "learning_rate": 1.728421052631579e-05, - "loss": 2.0124, + "grad_norm": 0.38671875, + "learning_rate": 0.00014415584415584416, + "loss": 1.805, "step": 358 }, { "epoch": 0.46, - "grad_norm": 0.2265625, - "learning_rate": 1.727368421052632e-05, - "loss": 1.9703, + "grad_norm": 0.29296875, + "learning_rate": 0.00014393939393939396, + "loss": 1.7458, "step": 359 }, { "epoch": 0.46, - "grad_norm": 0.267578125, - "learning_rate": 1.7263157894736843e-05, - "loss": 1.9995, + "grad_norm": 0.416015625, + "learning_rate": 0.00014372294372294372, + "loss": 1.7989, "step": 360 }, { "epoch": 0.46, - "grad_norm": 0.2578125, - "learning_rate": 1.725263157894737e-05, - "loss": 1.9216, + "grad_norm": 0.404296875, + "learning_rate": 0.00014350649350649352, + "loss": 1.7271, "step": 361 }, { "epoch": 0.46, - "grad_norm": 0.2333984375, - "learning_rate": 1.7242105263157896e-05, - "loss": 2.0042, + "grad_norm": 0.296875, + "learning_rate": 0.0001432900432900433, + "loss": 1.7804, "step": 362 }, { "epoch": 0.46, - "grad_norm": 0.2421875, - "learning_rate": 1.723157894736842e-05, - "loss": 2.056, + "grad_norm": 0.296875, + "learning_rate": 0.00014307359307359307, + "loss": 1.8581, "step": 363 }, { "epoch": 0.47, - "grad_norm": 0.224609375, - "learning_rate": 1.722105263157895e-05, - "loss": 1.9541, + "grad_norm": 0.353515625, + "learning_rate": 0.00014285714285714287, + "loss": 1.757, "step": 364 }, { "epoch": 0.47, - "grad_norm": 0.2734375, - "learning_rate": 1.7210526315789477e-05, - "loss": 1.9504, + "grad_norm": 0.3203125, + "learning_rate": 0.00014264069264069263, + "loss": 1.7536, "step": 365 }, { "epoch": 0.47, - "grad_norm": 0.2412109375, - "learning_rate": 1.72e-05, - "loss": 1.9347, + "grad_norm": 0.3203125, + "learning_rate": 0.00014242424242424243, + "loss": 1.7191, "step": 366 }, { "epoch": 0.47, - "grad_norm": 0.283203125, - "learning_rate": 1.7189473684210526e-05, - "loss": 1.9838, + "grad_norm": 0.384765625, + "learning_rate": 0.00014220779220779222, + "loss": 1.7741, "step": 367 }, { "epoch": 0.47, - "grad_norm": 0.236328125, - "learning_rate": 1.7178947368421054e-05, - "loss": 1.9544, + "grad_norm": 0.3359375, + "learning_rate": 0.00014199134199134201, + "loss": 1.7626, "step": 368 }, { "epoch": 0.47, - "grad_norm": 0.265625, - "learning_rate": 1.7168421052631582e-05, - "loss": 2.0481, + "grad_norm": 0.4140625, + "learning_rate": 0.00014177489177489178, + "loss": 1.8351, "step": 369 }, { "epoch": 0.47, - "grad_norm": 0.263671875, - "learning_rate": 1.7157894736842107e-05, - "loss": 2.0103, + "grad_norm": 0.4921875, + "learning_rate": 0.00014155844155844155, + "loss": 1.7943, "step": 370 }, { "epoch": 0.47, - "grad_norm": 0.220703125, - "learning_rate": 1.714736842105263e-05, - "loss": 1.968, + "grad_norm": 0.28515625, + "learning_rate": 0.00014134199134199137, + "loss": 1.7729, "step": 371 }, { "epoch": 0.48, - "grad_norm": 0.232421875, - "learning_rate": 1.713684210526316e-05, - "loss": 2.0555, + "grad_norm": 0.36328125, + "learning_rate": 0.00014112554112554113, + "loss": 1.853, "step": 372 }, { "epoch": 0.48, - "grad_norm": 0.23046875, - "learning_rate": 1.7126315789473687e-05, - "loss": 1.991, + "grad_norm": 0.373046875, + "learning_rate": 0.00014090909090909093, + "loss": 1.7805, "step": 373 }, { "epoch": 0.48, - "grad_norm": 0.255859375, - "learning_rate": 1.7115789473684212e-05, - "loss": 2.0144, + "grad_norm": 0.375, + "learning_rate": 0.0001406926406926407, + "loss": 1.805, "step": 374 }, { "epoch": 0.48, - "grad_norm": 0.234375, - "learning_rate": 1.7105263157894737e-05, - "loss": 1.9675, + "grad_norm": 0.365234375, + "learning_rate": 0.00014047619047619049, + "loss": 1.7717, "step": 375 }, { "epoch": 0.48, - "grad_norm": 0.2353515625, - "learning_rate": 1.7094736842105265e-05, - "loss": 1.9817, + "grad_norm": 0.375, + "learning_rate": 0.00014025974025974028, + "loss": 1.793, "step": 376 }, { "epoch": 0.48, - "grad_norm": 0.216796875, - "learning_rate": 1.708421052631579e-05, - "loss": 2.0047, + "grad_norm": 0.353515625, + "learning_rate": 0.00014004329004329005, + "loss": 1.8017, "step": 377 }, { "epoch": 0.48, - "grad_norm": 0.263671875, - "learning_rate": 1.7073684210526317e-05, - "loss": 1.9948, + "grad_norm": 0.359375, + "learning_rate": 0.00013982683982683984, + "loss": 1.758, "step": 378 }, { "epoch": 0.48, - "grad_norm": 0.2255859375, - "learning_rate": 1.7063157894736845e-05, - "loss": 1.9905, + "grad_norm": 0.35546875, + "learning_rate": 0.0001396103896103896, + "loss": 1.7761, "step": 379 }, { "epoch": 0.49, - "grad_norm": 0.3125, - "learning_rate": 1.705263157894737e-05, - "loss": 1.9718, + "grad_norm": 0.33203125, + "learning_rate": 0.0001393939393939394, + "loss": 1.7626, "step": 380 }, { "epoch": 0.49, - "grad_norm": 0.2119140625, - "learning_rate": 1.7042105263157895e-05, - "loss": 2.0456, + "grad_norm": 0.361328125, + "learning_rate": 0.0001391774891774892, + "loss": 1.8463, "step": 381 }, { "epoch": 0.49, - "grad_norm": 0.25, - "learning_rate": 1.7031578947368423e-05, - "loss": 1.9281, + "grad_norm": 0.4140625, + "learning_rate": 0.00013896103896103896, + "loss": 1.7563, "step": 382 }, { "epoch": 0.49, - "grad_norm": 0.29296875, - "learning_rate": 1.702105263157895e-05, - "loss": 1.9111, + "grad_norm": 0.318359375, + "learning_rate": 0.00013874458874458875, + "loss": 1.7085, "step": 383 }, { "epoch": 0.49, - "grad_norm": 0.28125, - "learning_rate": 1.7010526315789475e-05, - "loss": 2.0309, + "grad_norm": 0.337890625, + "learning_rate": 0.00013852813852813852, + "loss": 1.8202, "step": 384 }, { "epoch": 0.49, - "grad_norm": 0.23828125, - "learning_rate": 1.7e-05, - "loss": 1.984, + "grad_norm": 0.30078125, + "learning_rate": 0.00013831168831168834, + "loss": 1.7618, "step": 385 }, { "epoch": 0.49, - "grad_norm": 0.2109375, - "learning_rate": 1.6989473684210528e-05, - "loss": 2.0281, + "grad_norm": 0.3046875, + "learning_rate": 0.0001380952380952381, + "loss": 1.8142, "step": 386 }, { "epoch": 0.5, - "grad_norm": 0.2353515625, - "learning_rate": 1.6978947368421056e-05, - "loss": 1.8723, + "grad_norm": 0.318359375, + "learning_rate": 0.0001378787878787879, + "loss": 1.6778, "step": 387 }, { "epoch": 0.5, - "grad_norm": 0.220703125, - "learning_rate": 1.696842105263158e-05, - "loss": 1.9403, + "grad_norm": 0.287109375, + "learning_rate": 0.00013766233766233766, + "loss": 1.749, "step": 388 }, { "epoch": 0.5, - "grad_norm": 0.271484375, - "learning_rate": 1.6957894736842105e-05, - "loss": 1.9185, + "grad_norm": 0.310546875, + "learning_rate": 0.00013744588744588746, + "loss": 1.7236, "step": 389 }, { "epoch": 0.5, - "grad_norm": 0.251953125, - "learning_rate": 1.6947368421052633e-05, - "loss": 1.9107, + "grad_norm": 0.322265625, + "learning_rate": 0.00013722943722943725, + "loss": 1.7329, "step": 390 }, { "epoch": 0.5, - "grad_norm": 0.2578125, - "learning_rate": 1.6936842105263158e-05, - "loss": 2.035, + "grad_norm": 0.384765625, + "learning_rate": 0.00013701298701298702, + "loss": 1.8287, "step": 391 }, { "epoch": 0.5, - "grad_norm": 0.259765625, - "learning_rate": 1.6926315789473686e-05, - "loss": 2.0138, + "grad_norm": 0.318359375, + "learning_rate": 0.0001367965367965368, + "loss": 1.8173, "step": 392 }, { "epoch": 0.5, - "grad_norm": 0.267578125, - "learning_rate": 1.691578947368421e-05, - "loss": 2.1113, + "grad_norm": 0.5234375, + "learning_rate": 0.00013658008658008658, + "loss": 1.8881, "step": 393 }, { "epoch": 0.5, - "grad_norm": 0.275390625, - "learning_rate": 1.690526315789474e-05, - "loss": 1.9751, + "grad_norm": 0.4453125, + "learning_rate": 0.00013636363636363637, + "loss": 1.7895, "step": 394 }, { "epoch": 0.51, - "grad_norm": 0.216796875, - "learning_rate": 1.6894736842105263e-05, - "loss": 1.9502, + "grad_norm": 0.310546875, + "learning_rate": 0.00013614718614718616, + "loss": 1.7531, "step": 395 }, { "epoch": 0.51, - "grad_norm": 0.271484375, - "learning_rate": 1.688421052631579e-05, - "loss": 1.9788, + "grad_norm": 0.337890625, + "learning_rate": 0.00013593073593073593, + "loss": 1.7601, "step": 396 }, { "epoch": 0.51, - "grad_norm": 0.2578125, - "learning_rate": 1.687368421052632e-05, - "loss": 1.8612, + "grad_norm": 0.353515625, + "learning_rate": 0.00013571428571428572, + "loss": 1.6574, "step": 397 }, { "epoch": 0.51, - "grad_norm": 0.25390625, - "learning_rate": 1.6863157894736844e-05, - "loss": 1.9603, + "grad_norm": 0.345703125, + "learning_rate": 0.0001354978354978355, + "loss": 1.7577, "step": 398 }, { "epoch": 0.51, - "grad_norm": 0.2353515625, - "learning_rate": 1.685263157894737e-05, - "loss": 1.8944, + "grad_norm": 0.455078125, + "learning_rate": 0.0001352813852813853, + "loss": 1.6886, "step": 399 }, { "epoch": 0.51, - "grad_norm": 0.267578125, - "learning_rate": 1.6842105263157896e-05, - "loss": 1.9853, + "grad_norm": 0.337890625, + "learning_rate": 0.00013506493506493507, + "loss": 1.7786, "step": 400 }, { "epoch": 0.51, - "eval_loss": 1.9595069885253906, - "eval_runtime": 126.0386, - "eval_samples_per_second": 39.67, - "eval_steps_per_second": 1.246, + "eval_loss": 1.7625454664230347, + "eval_runtime": 125.3037, + "eval_samples_per_second": 39.903, + "eval_steps_per_second": 1.253, "step": 400 }, { "epoch": 0.51, - "grad_norm": 0.2236328125, - "learning_rate": 1.6831578947368424e-05, - "loss": 1.9316, + "grad_norm": 0.392578125, + "learning_rate": 0.00013484848484848484, + "loss": 1.7421, "step": 401 }, { "epoch": 0.51, - "grad_norm": 0.26171875, - "learning_rate": 1.682105263157895e-05, - "loss": 1.9619, + "grad_norm": 0.53125, + "learning_rate": 0.00013463203463203463, + "loss": 1.7626, "step": 402 }, { "epoch": 0.52, - "grad_norm": 0.26171875, - "learning_rate": 1.6810526315789474e-05, - "loss": 2.0236, + "grad_norm": 0.34765625, + "learning_rate": 0.00013441558441558443, + "loss": 1.7943, "step": 403 }, { "epoch": 0.52, - "grad_norm": 0.2177734375, - "learning_rate": 1.6800000000000002e-05, - "loss": 1.9843, + "grad_norm": 0.435546875, + "learning_rate": 0.00013419913419913422, + "loss": 1.8116, "step": 404 }, { "epoch": 0.52, - "grad_norm": 0.29296875, - "learning_rate": 1.6789473684210526e-05, - "loss": 1.9651, + "grad_norm": 0.419921875, + "learning_rate": 0.000133982683982684, + "loss": 1.7532, "step": 405 }, { "epoch": 0.52, - "grad_norm": 0.2177734375, - "learning_rate": 1.6778947368421054e-05, - "loss": 1.932, + "grad_norm": 0.345703125, + "learning_rate": 0.00013376623376623375, + "loss": 1.7478, "step": 406 }, { "epoch": 0.52, - "grad_norm": 0.259765625, - "learning_rate": 1.676842105263158e-05, - "loss": 2.0235, + "grad_norm": 0.40234375, + "learning_rate": 0.00013354978354978355, + "loss": 1.8208, "step": 407 }, { "epoch": 0.52, - "grad_norm": 0.2353515625, - "learning_rate": 1.6757894736842107e-05, - "loss": 1.9685, + "grad_norm": 0.314453125, + "learning_rate": 0.00013333333333333334, + "loss": 1.7518, "step": 408 }, { "epoch": 0.52, - "grad_norm": 0.2177734375, - "learning_rate": 1.6747368421052632e-05, - "loss": 2.0068, + "grad_norm": 0.470703125, + "learning_rate": 0.00013311688311688313, + "loss": 1.8063, "step": 409 }, { "epoch": 0.52, - "grad_norm": 0.287109375, - "learning_rate": 1.673684210526316e-05, - "loss": 1.9278, + "grad_norm": 0.419921875, + "learning_rate": 0.0001329004329004329, + "loss": 1.7366, "step": 410 }, { "epoch": 0.53, - "grad_norm": 0.314453125, - "learning_rate": 1.6726315789473684e-05, - "loss": 1.9738, + "grad_norm": 0.326171875, + "learning_rate": 0.0001326839826839827, + "loss": 1.7605, "step": 411 }, { "epoch": 0.53, - "grad_norm": 0.212890625, - "learning_rate": 1.6715789473684212e-05, - "loss": 2.0359, + "grad_norm": 0.287109375, + "learning_rate": 0.00013246753246753249, + "loss": 1.8454, "step": 412 }, { "epoch": 0.53, - "grad_norm": 0.216796875, - "learning_rate": 1.6705263157894737e-05, - "loss": 1.9557, + "grad_norm": 0.376953125, + "learning_rate": 0.00013225108225108225, + "loss": 1.7342, "step": 413 }, { "epoch": 0.53, - "grad_norm": 0.251953125, - "learning_rate": 1.6694736842105265e-05, - "loss": 1.8912, + "grad_norm": 0.416015625, + "learning_rate": 0.00013203463203463205, + "loss": 1.7031, "step": 414 }, { "epoch": 0.53, - "grad_norm": 0.255859375, - "learning_rate": 1.6684210526315793e-05, - "loss": 2.0441, + "grad_norm": 0.3984375, + "learning_rate": 0.0001318181818181818, + "loss": 1.8397, "step": 415 }, { "epoch": 0.53, - "grad_norm": 0.2373046875, - "learning_rate": 1.6673684210526318e-05, - "loss": 2.0091, + "grad_norm": 0.369140625, + "learning_rate": 0.0001316017316017316, + "loss": 1.8038, "step": 416 }, { "epoch": 0.53, - "grad_norm": 0.345703125, - "learning_rate": 1.6663157894736842e-05, - "loss": 1.959, + "grad_norm": 0.369140625, + "learning_rate": 0.0001313852813852814, + "loss": 1.7529, "step": 417 }, { "epoch": 0.53, - "grad_norm": 0.2109375, - "learning_rate": 1.665263157894737e-05, - "loss": 1.9557, + "grad_norm": 0.396484375, + "learning_rate": 0.0001311688311688312, + "loss": 1.7746, "step": 418 }, { "epoch": 0.54, - "grad_norm": 0.2060546875, - "learning_rate": 1.66421052631579e-05, - "loss": 2.0337, + "grad_norm": 0.31640625, + "learning_rate": 0.00013095238095238096, + "loss": 1.8441, "step": 419 }, { "epoch": 0.54, - "grad_norm": 0.2412109375, - "learning_rate": 1.6631578947368423e-05, - "loss": 1.9243, + "grad_norm": 0.287109375, + "learning_rate": 0.00013073593073593072, + "loss": 1.7302, "step": 420 }, { "epoch": 0.54, - "grad_norm": 0.25, - "learning_rate": 1.6621052631578948e-05, - "loss": 1.9255, + "grad_norm": 0.470703125, + "learning_rate": 0.00013051948051948052, + "loss": 1.7149, "step": 421 }, { "epoch": 0.54, - "grad_norm": 0.2158203125, - "learning_rate": 1.6610526315789476e-05, - "loss": 1.8838, + "grad_norm": 0.330078125, + "learning_rate": 0.0001303030303030303, + "loss": 1.6881, "step": 422 }, { "epoch": 0.54, - "grad_norm": 0.279296875, - "learning_rate": 1.66e-05, - "loss": 1.9987, + "grad_norm": 0.28515625, + "learning_rate": 0.0001300865800865801, + "loss": 1.7627, "step": 423 }, { "epoch": 0.54, - "grad_norm": 0.2158203125, - "learning_rate": 1.658947368421053e-05, - "loss": 1.9341, + "grad_norm": 0.310546875, + "learning_rate": 0.00012987012987012987, + "loss": 1.7386, "step": 424 }, { "epoch": 0.54, - "grad_norm": 0.216796875, - "learning_rate": 1.6578947368421053e-05, - "loss": 1.9327, + "grad_norm": 0.310546875, + "learning_rate": 0.00012965367965367964, + "loss": 1.7331, "step": 425 }, { "epoch": 0.55, - "grad_norm": 0.30078125, - "learning_rate": 1.656842105263158e-05, - "loss": 1.9685, + "grad_norm": 0.384765625, + "learning_rate": 0.00012943722943722946, + "loss": 1.766, "step": 426 }, { "epoch": 0.55, - "grad_norm": 0.23046875, - "learning_rate": 1.6557894736842106e-05, - "loss": 1.9199, + "grad_norm": 0.3828125, + "learning_rate": 0.00012922077922077922, + "loss": 1.7111, "step": 427 }, { "epoch": 0.55, - "grad_norm": 0.318359375, - "learning_rate": 1.6547368421052634e-05, - "loss": 1.958, + "grad_norm": 0.37890625, + "learning_rate": 0.00012900432900432902, + "loss": 1.7874, "step": 428 }, { "epoch": 0.55, - "grad_norm": 0.2265625, - "learning_rate": 1.653684210526316e-05, - "loss": 1.9791, + "grad_norm": 0.380859375, + "learning_rate": 0.00012878787878787878, + "loss": 1.7565, "step": 429 }, { "epoch": 0.55, - "grad_norm": 0.2109375, - "learning_rate": 1.6526315789473686e-05, - "loss": 2.0316, + "grad_norm": 0.37109375, + "learning_rate": 0.00012857142857142858, + "loss": 1.8451, "step": 430 }, { "epoch": 0.55, - "grad_norm": 0.2255859375, - "learning_rate": 1.651578947368421e-05, - "loss": 1.8716, + "grad_norm": 0.419921875, + "learning_rate": 0.00012835497835497837, + "loss": 1.7024, "step": 431 }, { "epoch": 0.55, - "grad_norm": 0.2392578125, - "learning_rate": 1.650526315789474e-05, - "loss": 1.8782, + "grad_norm": 0.294921875, + "learning_rate": 0.00012813852813852814, + "loss": 1.7037, "step": 432 }, { "epoch": 0.55, - "grad_norm": 0.2333984375, - "learning_rate": 1.6494736842105267e-05, - "loss": 1.8991, + "grad_norm": 0.310546875, + "learning_rate": 0.00012792207792207793, + "loss": 1.7166, "step": 433 }, { "epoch": 0.56, - "grad_norm": 0.228515625, - "learning_rate": 1.648421052631579e-05, - "loss": 1.9401, + "grad_norm": 0.404296875, + "learning_rate": 0.0001277056277056277, + "loss": 1.7534, "step": 434 }, { "epoch": 0.56, - "grad_norm": 0.2314453125, - "learning_rate": 1.6473684210526316e-05, - "loss": 1.919, + "grad_norm": 0.3359375, + "learning_rate": 0.00012748917748917752, + "loss": 1.7315, "step": 435 }, { "epoch": 0.56, - "grad_norm": 0.26171875, - "learning_rate": 1.6463157894736844e-05, - "loss": 1.9507, + "grad_norm": 0.349609375, + "learning_rate": 0.00012727272727272728, + "loss": 1.7707, "step": 436 }, { "epoch": 0.56, - "grad_norm": 0.2099609375, - "learning_rate": 1.645263157894737e-05, - "loss": 2.0071, + "grad_norm": 0.328125, + "learning_rate": 0.00012705627705627707, + "loss": 1.811, "step": 437 }, { "epoch": 0.56, - "grad_norm": 0.2314453125, - "learning_rate": 1.6442105263157897e-05, - "loss": 2.0168, + "grad_norm": 0.2890625, + "learning_rate": 0.00012683982683982684, + "loss": 1.8211, "step": 438 }, { "epoch": 0.56, - "grad_norm": 0.248046875, - "learning_rate": 1.643157894736842e-05, - "loss": 1.8663, + "grad_norm": 0.330078125, + "learning_rate": 0.00012662337662337663, + "loss": 1.6992, "step": 439 }, { "epoch": 0.56, - "grad_norm": 0.248046875, - "learning_rate": 1.642105263157895e-05, - "loss": 2.0558, + "grad_norm": 0.33203125, + "learning_rate": 0.00012640692640692643, + "loss": 1.8294, "step": 440 }, { "epoch": 0.56, - "grad_norm": 0.2734375, - "learning_rate": 1.6410526315789474e-05, - "loss": 1.9572, + "grad_norm": 0.326171875, + "learning_rate": 0.0001261904761904762, + "loss": 1.7351, "step": 441 }, { "epoch": 0.57, - "grad_norm": 0.26953125, - "learning_rate": 1.64e-05, - "loss": 1.95, + "grad_norm": 0.384765625, + "learning_rate": 0.000125974025974026, + "loss": 1.732, "step": 442 }, { "epoch": 0.57, - "grad_norm": 0.23046875, - "learning_rate": 1.6389473684210527e-05, - "loss": 2.0, + "grad_norm": 0.41796875, + "learning_rate": 0.00012575757575757575, + "loss": 1.7839, "step": 443 }, { "epoch": 0.57, - "grad_norm": 0.25, - "learning_rate": 1.6378947368421055e-05, - "loss": 2.0094, + "grad_norm": 0.34375, + "learning_rate": 0.00012554112554112555, + "loss": 1.7892, "step": 444 }, { "epoch": 0.57, - "grad_norm": 0.291015625, - "learning_rate": 1.636842105263158e-05, - "loss": 1.94, + "grad_norm": 0.37890625, + "learning_rate": 0.00012532467532467534, + "loss": 1.7428, "step": 445 }, { "epoch": 0.57, - "grad_norm": 0.251953125, - "learning_rate": 1.6357894736842108e-05, - "loss": 2.0211, + "grad_norm": 0.3359375, + "learning_rate": 0.0001251082251082251, + "loss": 1.8055, "step": 446 }, { "epoch": 0.57, - "grad_norm": 0.2578125, - "learning_rate": 1.6347368421052636e-05, - "loss": 2.0493, + "grad_norm": 0.384765625, + "learning_rate": 0.0001248917748917749, + "loss": 1.8227, "step": 447 }, { "epoch": 0.57, - "grad_norm": 0.259765625, - "learning_rate": 1.633684210526316e-05, - "loss": 1.9497, + "grad_norm": 0.349609375, + "learning_rate": 0.00012467532467532467, + "loss": 1.7273, "step": 448 }, { "epoch": 0.57, - "grad_norm": 0.2392578125, - "learning_rate": 1.6326315789473685e-05, - "loss": 2.0445, + "grad_norm": 0.328125, + "learning_rate": 0.00012445887445887449, + "loss": 1.8338, "step": 449 }, { "epoch": 0.58, - "grad_norm": 0.265625, - "learning_rate": 1.6315789473684213e-05, - "loss": 1.9021, + "grad_norm": 0.4140625, + "learning_rate": 0.00012424242424242425, + "loss": 1.71, "step": 450 }, { "epoch": 0.58, - "grad_norm": 0.259765625, - "learning_rate": 1.6305263157894737e-05, - "loss": 1.8826, + "eval_loss": 1.7571938037872314, + "eval_runtime": 125.7305, + "eval_samples_per_second": 39.768, + "eval_steps_per_second": 1.249, + "step": 450 + }, + { + "epoch": 0.58, + "grad_norm": 0.408203125, + "learning_rate": 0.00012402597402597402, + "loss": 1.6852, "step": 451 }, { "epoch": 0.58, - "grad_norm": 0.216796875, - "learning_rate": 1.6294736842105265e-05, - "loss": 1.9832, + "grad_norm": 0.3125, + "learning_rate": 0.0001238095238095238, + "loss": 1.7876, "step": 452 }, { "epoch": 0.58, - "grad_norm": 0.255859375, - "learning_rate": 1.628421052631579e-05, - "loss": 1.9787, + "grad_norm": 0.314453125, + "learning_rate": 0.0001235930735930736, + "loss": 1.7676, "step": 453 }, { "epoch": 0.58, - "grad_norm": 0.27734375, - "learning_rate": 1.6273684210526318e-05, - "loss": 1.9508, + "grad_norm": 0.439453125, + "learning_rate": 0.0001233766233766234, + "loss": 1.7567, "step": 454 }, { "epoch": 0.58, - "grad_norm": 0.275390625, - "learning_rate": 1.6263157894736843e-05, - "loss": 1.9811, + "grad_norm": 0.294921875, + "learning_rate": 0.00012316017316017316, + "loss": 1.7835, "step": 455 }, { "epoch": 0.58, - "grad_norm": 0.2333984375, - "learning_rate": 1.6252631578947367e-05, - "loss": 2.0144, + "grad_norm": 0.3046875, + "learning_rate": 0.00012294372294372293, + "loss": 1.8254, "step": 456 }, { "epoch": 0.58, - "grad_norm": 0.228515625, - "learning_rate": 1.6242105263157895e-05, - "loss": 1.9287, + "grad_norm": 0.30078125, + "learning_rate": 0.00012272727272727272, + "loss": 1.7415, "step": 457 }, { "epoch": 0.59, - "grad_norm": 0.291015625, - "learning_rate": 1.6231578947368423e-05, - "loss": 1.9812, + "grad_norm": 0.3359375, + "learning_rate": 0.00012251082251082252, + "loss": 1.7718, "step": 458 }, { "epoch": 0.59, - "grad_norm": 0.2177734375, - "learning_rate": 1.6221052631578948e-05, - "loss": 1.9838, + "grad_norm": 0.361328125, + "learning_rate": 0.0001222943722943723, + "loss": 1.7734, "step": 459 }, { "epoch": 0.59, - "grad_norm": 0.2392578125, - "learning_rate": 1.6210526315789473e-05, - "loss": 1.9178, + "grad_norm": 0.3125, + "learning_rate": 0.00012207792207792208, + "loss": 1.7413, "step": 460 }, { "epoch": 0.59, - "grad_norm": 0.2578125, - "learning_rate": 1.62e-05, - "loss": 2.0835, + "grad_norm": 0.283203125, + "learning_rate": 0.00012186147186147187, + "loss": 1.8832, "step": 461 }, { "epoch": 0.59, - "grad_norm": 0.2177734375, - "learning_rate": 1.618947368421053e-05, - "loss": 1.9134, + "grad_norm": 0.328125, + "learning_rate": 0.00012164502164502165, + "loss": 1.71, "step": 462 }, { "epoch": 0.59, - "grad_norm": 0.2177734375, - "learning_rate": 1.6178947368421053e-05, - "loss": 1.9788, + "grad_norm": 0.3203125, + "learning_rate": 0.00012142857142857143, + "loss": 1.7611, "step": 463 }, { "epoch": 0.59, - "grad_norm": 0.23046875, - "learning_rate": 1.616842105263158e-05, - "loss": 1.9825, + "grad_norm": 0.373046875, + "learning_rate": 0.00012121212121212122, + "loss": 1.809, "step": 464 }, { "epoch": 0.6, - "grad_norm": 0.228515625, - "learning_rate": 1.6157894736842106e-05, - "loss": 1.8493, + "grad_norm": 0.29296875, + "learning_rate": 0.00012099567099567099, + "loss": 1.6685, "step": 465 }, { "epoch": 0.6, - "grad_norm": 0.236328125, - "learning_rate": 1.6147368421052634e-05, - "loss": 1.9302, + "grad_norm": 0.310546875, + "learning_rate": 0.0001207792207792208, + "loss": 1.7246, "step": 466 }, { "epoch": 0.6, - "grad_norm": 0.2373046875, - "learning_rate": 1.613684210526316e-05, - "loss": 1.9387, + "grad_norm": 0.30859375, + "learning_rate": 0.00012056277056277056, + "loss": 1.7187, "step": 467 }, { "epoch": 0.6, - "grad_norm": 0.26953125, - "learning_rate": 1.6126315789473687e-05, - "loss": 1.9132, + "grad_norm": 0.296875, + "learning_rate": 0.00012034632034632037, + "loss": 1.7165, "step": 468 }, { "epoch": 0.6, - "grad_norm": 0.306640625, - "learning_rate": 1.611578947368421e-05, - "loss": 1.9371, + "grad_norm": 0.353515625, + "learning_rate": 0.00012012987012987014, + "loss": 1.7451, "step": 469 }, { "epoch": 0.6, - "grad_norm": 0.23046875, - "learning_rate": 1.6105263157894736e-05, - "loss": 1.9414, + "grad_norm": 0.33203125, + "learning_rate": 0.00011991341991341991, + "loss": 1.7889, "step": 470 }, { "epoch": 0.6, - "grad_norm": 0.23046875, - "learning_rate": 1.6094736842105264e-05, - "loss": 2.0439, + "grad_norm": 0.34375, + "learning_rate": 0.00011969696969696971, + "loss": 1.8299, "step": 471 }, { "epoch": 0.6, - "grad_norm": 0.248046875, - "learning_rate": 1.6084210526315792e-05, - "loss": 1.947, + "grad_norm": 0.314453125, + "learning_rate": 0.00011948051948051949, + "loss": 1.7567, "step": 472 }, { "epoch": 0.61, - "grad_norm": 0.263671875, - "learning_rate": 1.6073684210526317e-05, - "loss": 1.9361, + "grad_norm": 0.3359375, + "learning_rate": 0.00011926406926406928, + "loss": 1.7335, "step": 473 }, { "epoch": 0.61, - "grad_norm": 0.2080078125, - "learning_rate": 1.606315789473684e-05, - "loss": 1.9466, + "grad_norm": 0.30859375, + "learning_rate": 0.00011904761904761905, + "loss": 1.7394, "step": 474 }, { "epoch": 0.61, - "grad_norm": 0.2490234375, - "learning_rate": 1.605263157894737e-05, - "loss": 1.9929, + "grad_norm": 0.34765625, + "learning_rate": 0.00011883116883116883, + "loss": 1.7834, "step": 475 }, { "epoch": 0.61, - "grad_norm": 0.224609375, - "learning_rate": 1.6042105263157897e-05, - "loss": 1.9829, + "grad_norm": 0.322265625, + "learning_rate": 0.00011861471861471862, + "loss": 1.7819, "step": 476 }, { "epoch": 0.61, - "grad_norm": 0.189453125, - "learning_rate": 1.6031578947368422e-05, - "loss": 1.9354, + "grad_norm": 0.271484375, + "learning_rate": 0.0001183982683982684, + "loss": 1.7574, "step": 477 }, { "epoch": 0.61, - "grad_norm": 0.255859375, - "learning_rate": 1.6021052631578947e-05, - "loss": 1.8911, + "grad_norm": 0.330078125, + "learning_rate": 0.0001181818181818182, + "loss": 1.7031, "step": 478 }, { "epoch": 0.61, - "grad_norm": 0.23828125, - "learning_rate": 1.6010526315789475e-05, - "loss": 1.9281, + "grad_norm": 0.30859375, + "learning_rate": 0.00011796536796536797, + "loss": 1.7361, "step": 479 }, { "epoch": 0.61, - "grad_norm": 0.228515625, - "learning_rate": 1.6000000000000003e-05, - "loss": 1.9434, + "grad_norm": 0.333984375, + "learning_rate": 0.00011774891774891777, + "loss": 1.7638, "step": 480 }, { "epoch": 0.62, - "grad_norm": 0.30078125, - "learning_rate": 1.5989473684210527e-05, - "loss": 1.9577, + "grad_norm": 0.34375, + "learning_rate": 0.00011753246753246753, + "loss": 1.7476, "step": 481 }, { "epoch": 0.62, - "grad_norm": 0.251953125, - "learning_rate": 1.5978947368421055e-05, - "loss": 1.8733, + "grad_norm": 0.333984375, + "learning_rate": 0.00011731601731601731, + "loss": 1.6884, "step": 482 }, { "epoch": 0.62, - "grad_norm": 0.216796875, - "learning_rate": 1.596842105263158e-05, - "loss": 1.8984, + "grad_norm": 0.296875, + "learning_rate": 0.0001170995670995671, + "loss": 1.6966, "step": 483 }, { "epoch": 0.62, - "grad_norm": 0.2353515625, - "learning_rate": 1.5957894736842105e-05, - "loss": 1.9113, + "grad_norm": 0.306640625, + "learning_rate": 0.00011688311688311689, + "loss": 1.7041, "step": 484 }, { "epoch": 0.62, - "grad_norm": 0.2890625, - "learning_rate": 1.5947368421052633e-05, - "loss": 1.978, + "grad_norm": 0.33203125, + "learning_rate": 0.00011666666666666668, + "loss": 1.7788, "step": 485 }, { "epoch": 0.62, - "grad_norm": 0.2275390625, - "learning_rate": 1.593684210526316e-05, - "loss": 1.9808, + "grad_norm": 0.328125, + "learning_rate": 0.00011645021645021646, + "loss": 1.7744, "step": 486 }, { "epoch": 0.62, - "grad_norm": 0.267578125, - "learning_rate": 1.5926315789473685e-05, - "loss": 1.9299, + "grad_norm": 0.423828125, + "learning_rate": 0.00011623376623376625, + "loss": 1.7492, "step": 487 }, { "epoch": 0.62, - "grad_norm": 0.20703125, - "learning_rate": 1.591578947368421e-05, - "loss": 1.9296, + "grad_norm": 0.318359375, + "learning_rate": 0.00011601731601731602, + "loss": 1.7537, "step": 488 }, { "epoch": 0.63, - "grad_norm": 0.2119140625, - "learning_rate": 1.5905263157894738e-05, - "loss": 1.921, + "grad_norm": 0.337890625, + "learning_rate": 0.0001158008658008658, + "loss": 1.7363, "step": 489 }, { "epoch": 0.63, - "grad_norm": 0.267578125, - "learning_rate": 1.5894736842105266e-05, - "loss": 1.9749, + "grad_norm": 0.349609375, + "learning_rate": 0.00011558441558441559, + "loss": 1.7846, "step": 490 }, { "epoch": 0.63, - "grad_norm": 0.2265625, - "learning_rate": 1.588421052631579e-05, - "loss": 1.8742, + "grad_norm": 0.310546875, + "learning_rate": 0.00011536796536796537, + "loss": 1.7081, "step": 491 }, { "epoch": 0.63, - "grad_norm": 0.294921875, - "learning_rate": 1.5873684210526315e-05, - "loss": 1.9215, + "grad_norm": 0.33203125, + "learning_rate": 0.00011515151515151516, + "loss": 1.7275, "step": 492 }, { "epoch": 0.63, - "grad_norm": 0.248046875, - "learning_rate": 1.5863157894736843e-05, - "loss": 1.9104, + "grad_norm": 0.33203125, + "learning_rate": 0.00011493506493506494, + "loss": 1.7196, "step": 493 }, { "epoch": 0.63, - "grad_norm": 0.23046875, - "learning_rate": 1.585263157894737e-05, - "loss": 1.8637, + "grad_norm": 0.30078125, + "learning_rate": 0.00011471861471861471, + "loss": 1.6845, "step": 494 }, { "epoch": 0.63, - "grad_norm": 0.2236328125, - "learning_rate": 1.5842105263157896e-05, - "loss": 1.9424, + "grad_norm": 0.337890625, + "learning_rate": 0.00011450216450216452, + "loss": 1.7443, "step": 495 }, { "epoch": 0.63, - "grad_norm": 0.2119140625, - "learning_rate": 1.5831578947368424e-05, - "loss": 1.8857, + "grad_norm": 0.3359375, + "learning_rate": 0.00011428571428571428, + "loss": 1.7104, "step": 496 }, { "epoch": 0.64, - "grad_norm": 0.244140625, - "learning_rate": 1.582105263157895e-05, - "loss": 1.9293, + "grad_norm": 0.380859375, + "learning_rate": 0.00011406926406926408, + "loss": 1.7307, "step": 497 }, { "epoch": 0.64, - "grad_norm": 0.240234375, - "learning_rate": 1.5810526315789473e-05, - "loss": 2.0248, + "grad_norm": 0.361328125, + "learning_rate": 0.00011385281385281386, + "loss": 1.8338, "step": 498 }, { "epoch": 0.64, - "grad_norm": 0.2578125, - "learning_rate": 1.58e-05, - "loss": 1.9169, + "grad_norm": 0.296875, + "learning_rate": 0.00011363636363636365, + "loss": 1.7235, "step": 499 }, { "epoch": 0.64, - "grad_norm": 0.205078125, - "learning_rate": 1.578947368421053e-05, - "loss": 2.0431, + "grad_norm": 0.291015625, + "learning_rate": 0.00011341991341991343, + "loss": 1.8493, "step": 500 }, { "epoch": 0.64, - "grad_norm": 0.29296875, - "learning_rate": 1.5778947368421054e-05, - "loss": 1.9378, + "eval_loss": 1.7474199533462524, + "eval_runtime": 125.2746, + "eval_samples_per_second": 39.912, + "eval_steps_per_second": 1.253, + "step": 500 + }, + { + "epoch": 0.64, + "grad_norm": 0.33203125, + "learning_rate": 0.0001132034632034632, + "loss": 1.7369, "step": 501 }, { "epoch": 0.64, - "grad_norm": 0.240234375, - "learning_rate": 1.576842105263158e-05, - "loss": 1.9339, + "grad_norm": 0.322265625, + "learning_rate": 0.000112987012987013, + "loss": 1.7215, "step": 502 }, { "epoch": 0.64, - "grad_norm": 0.2177734375, - "learning_rate": 1.5757894736842107e-05, - "loss": 1.9931, + "grad_norm": 0.36328125, + "learning_rate": 0.00011277056277056277, + "loss": 1.7981, "step": 503 }, { "epoch": 0.64, - "grad_norm": 0.2431640625, - "learning_rate": 1.5747368421052635e-05, - "loss": 1.9111, + "grad_norm": 0.318359375, + "learning_rate": 0.00011255411255411256, + "loss": 1.6949, "step": 504 }, { "epoch": 0.65, - "grad_norm": 0.287109375, - "learning_rate": 1.573684210526316e-05, - "loss": 1.9611, + "grad_norm": 0.38671875, + "learning_rate": 0.00011233766233766234, + "loss": 1.7577, "step": 505 }, { "epoch": 0.65, - "grad_norm": 0.3046875, - "learning_rate": 1.5726315789473684e-05, - "loss": 1.9067, + "grad_norm": 0.52734375, + "learning_rate": 0.00011212121212121212, + "loss": 1.7396, "step": 506 }, { "epoch": 0.65, - "grad_norm": 0.224609375, - "learning_rate": 1.5715789473684212e-05, - "loss": 1.8983, + "grad_norm": 0.3359375, + "learning_rate": 0.00011190476190476191, + "loss": 1.7094, "step": 507 }, { "epoch": 0.65, - "grad_norm": 0.228515625, - "learning_rate": 1.570526315789474e-05, - "loss": 1.929, + "grad_norm": 0.314453125, + "learning_rate": 0.00011168831168831168, + "loss": 1.7354, "step": 508 }, { "epoch": 0.65, - "grad_norm": 0.201171875, - "learning_rate": 1.5694736842105264e-05, - "loss": 1.8899, + "grad_norm": 0.37109375, + "learning_rate": 0.00011147186147186149, + "loss": 1.7109, "step": 509 }, { "epoch": 0.65, - "grad_norm": 0.2138671875, - "learning_rate": 1.568421052631579e-05, - "loss": 1.9552, + "grad_norm": 0.4140625, + "learning_rate": 0.00011125541125541125, + "loss": 1.7809, "step": 510 }, { "epoch": 0.65, - "grad_norm": 0.2431640625, - "learning_rate": 1.5673684210526317e-05, - "loss": 1.9897, + "grad_norm": 0.455078125, + "learning_rate": 0.00011103896103896105, + "loss": 1.7939, "step": 511 }, { "epoch": 0.66, - "grad_norm": 0.287109375, - "learning_rate": 1.5663157894736842e-05, - "loss": 1.9904, + "grad_norm": 0.291015625, + "learning_rate": 0.00011082251082251083, + "loss": 1.7942, "step": 512 }, { "epoch": 0.66, - "grad_norm": 0.2216796875, - "learning_rate": 1.565263157894737e-05, - "loss": 1.9099, + "grad_norm": 0.609375, + "learning_rate": 0.00011060606060606061, + "loss": 1.7189, "step": 513 }, { "epoch": 0.66, - "grad_norm": 0.28125, - "learning_rate": 1.5642105263157898e-05, - "loss": 1.9848, + "grad_norm": 0.67578125, + "learning_rate": 0.0001103896103896104, + "loss": 1.7662, "step": 514 }, { "epoch": 0.66, - "grad_norm": 0.341796875, - "learning_rate": 1.5631578947368422e-05, - "loss": 1.9001, + "grad_norm": 0.353515625, + "learning_rate": 0.00011017316017316017, + "loss": 1.7288, "step": 515 }, { "epoch": 0.66, - "grad_norm": 0.212890625, - "learning_rate": 1.5621052631578947e-05, - "loss": 1.9287, + "grad_norm": 0.322265625, + "learning_rate": 0.00010995670995670997, + "loss": 1.715, "step": 516 }, { "epoch": 0.66, - "grad_norm": 0.2119140625, - "learning_rate": 1.5610526315789475e-05, - "loss": 1.8714, + "grad_norm": 0.40625, + "learning_rate": 0.00010974025974025974, + "loss": 1.6725, "step": 517 }, { "epoch": 0.66, - "grad_norm": 0.2314453125, - "learning_rate": 1.5600000000000003e-05, - "loss": 1.958, + "grad_norm": 0.52734375, + "learning_rate": 0.00010952380952380953, + "loss": 1.7752, "step": 518 }, { "epoch": 0.66, - "grad_norm": 0.298828125, - "learning_rate": 1.5589473684210528e-05, - "loss": 1.9089, + "grad_norm": 0.48828125, + "learning_rate": 0.00010930735930735931, + "loss": 1.7345, "step": 519 }, { "epoch": 0.67, - "grad_norm": 0.2060546875, - "learning_rate": 1.5578947368421052e-05, - "loss": 1.8629, + "grad_norm": 0.3203125, + "learning_rate": 0.00010909090909090909, + "loss": 1.7054, "step": 520 }, { "epoch": 0.67, - "grad_norm": 0.2236328125, - "learning_rate": 1.556842105263158e-05, - "loss": 2.0567, + "grad_norm": 0.45703125, + "learning_rate": 0.00010887445887445889, + "loss": 1.8508, "step": 521 }, { "epoch": 0.67, - "grad_norm": 0.248046875, - "learning_rate": 1.555789473684211e-05, - "loss": 1.9075, + "grad_norm": 0.50390625, + "learning_rate": 0.00010865800865800865, + "loss": 1.723, "step": 522 }, { "epoch": 0.67, - "grad_norm": 0.2255859375, - "learning_rate": 1.5547368421052633e-05, - "loss": 1.8856, + "grad_norm": 0.361328125, + "learning_rate": 0.00010844155844155846, + "loss": 1.7008, "step": 523 }, { "epoch": 0.67, - "grad_norm": 0.2451171875, - "learning_rate": 1.5536842105263158e-05, - "loss": 1.8793, + "grad_norm": 0.330078125, + "learning_rate": 0.00010822510822510823, + "loss": 1.6947, "step": 524 }, { "epoch": 0.67, - "grad_norm": 0.22265625, - "learning_rate": 1.5526315789473686e-05, - "loss": 1.9754, + "grad_norm": 0.439453125, + "learning_rate": 0.000108008658008658, + "loss": 1.8002, "step": 525 }, { "epoch": 0.67, - "grad_norm": 0.23046875, - "learning_rate": 1.551578947368421e-05, - "loss": 1.9667, + "grad_norm": 0.419921875, + "learning_rate": 0.0001077922077922078, + "loss": 1.7645, "step": 526 }, { "epoch": 0.67, - "grad_norm": 0.2734375, - "learning_rate": 1.550526315789474e-05, - "loss": 1.9586, + "grad_norm": 0.392578125, + "learning_rate": 0.00010757575757575758, + "loss": 1.7568, "step": 527 }, { "epoch": 0.68, - "grad_norm": 0.267578125, - "learning_rate": 1.5494736842105263e-05, - "loss": 1.9621, + "grad_norm": 0.34375, + "learning_rate": 0.00010735930735930737, + "loss": 1.7684, "step": 528 }, { "epoch": 0.68, - "grad_norm": 0.236328125, - "learning_rate": 1.548421052631579e-05, - "loss": 1.9793, + "grad_norm": 0.337890625, + "learning_rate": 0.00010714285714285715, + "loss": 1.7795, "step": 529 }, { "epoch": 0.68, - "grad_norm": 0.267578125, - "learning_rate": 1.5473684210526316e-05, - "loss": 1.9415, + "grad_norm": 0.375, + "learning_rate": 0.00010692640692640694, + "loss": 1.7471, "step": 530 }, { "epoch": 0.68, - "grad_norm": 0.259765625, - "learning_rate": 1.5463157894736844e-05, - "loss": 1.9964, + "grad_norm": 0.369140625, + "learning_rate": 0.00010670995670995671, + "loss": 1.8058, "step": 531 }, { "epoch": 0.68, - "grad_norm": 0.2294921875, - "learning_rate": 1.545263157894737e-05, - "loss": 1.87, + "grad_norm": 0.326171875, + "learning_rate": 0.00010649350649350649, + "loss": 1.673, "step": 532 }, { "epoch": 0.68, - "grad_norm": 0.25390625, - "learning_rate": 1.5442105263157896e-05, - "loss": 1.9316, + "grad_norm": 0.365234375, + "learning_rate": 0.00010627705627705628, + "loss": 1.728, "step": 533 }, { "epoch": 0.68, - "grad_norm": 0.21484375, - "learning_rate": 1.543157894736842e-05, - "loss": 1.8487, + "grad_norm": 0.423828125, + "learning_rate": 0.00010606060606060606, + "loss": 1.6786, "step": 534 }, { "epoch": 0.68, - "grad_norm": 0.263671875, - "learning_rate": 1.542105263157895e-05, - "loss": 1.9245, + "grad_norm": 0.37890625, + "learning_rate": 0.00010584415584415586, + "loss": 1.733, "step": 535 }, { "epoch": 0.69, - "grad_norm": 0.21875, - "learning_rate": 1.5410526315789477e-05, - "loss": 1.9137, + "grad_norm": 0.34375, + "learning_rate": 0.00010562770562770564, + "loss": 1.7435, "step": 536 }, { "epoch": 0.69, - "grad_norm": 0.197265625, - "learning_rate": 1.54e-05, - "loss": 1.9481, + "grad_norm": 0.3125, + "learning_rate": 0.00010541125541125543, + "loss": 1.7421, "step": 537 }, { "epoch": 0.69, - "grad_norm": 0.259765625, - "learning_rate": 1.5389473684210526e-05, - "loss": 2.0154, + "grad_norm": 0.376953125, + "learning_rate": 0.0001051948051948052, + "loss": 1.8177, "step": 538 }, { "epoch": 0.69, - "grad_norm": 0.318359375, - "learning_rate": 1.5378947368421054e-05, - "loss": 2.0653, + "grad_norm": 0.390625, + "learning_rate": 0.00010497835497835498, + "loss": 1.853, "step": 539 }, { "epoch": 0.69, - "grad_norm": 0.20703125, - "learning_rate": 1.536842105263158e-05, - "loss": 1.9154, + "grad_norm": 0.384765625, + "learning_rate": 0.00010476190476190477, + "loss": 1.7235, "step": 540 }, { "epoch": 0.69, - "grad_norm": 0.2197265625, - "learning_rate": 1.5357894736842107e-05, - "loss": 1.8704, + "grad_norm": 0.35546875, + "learning_rate": 0.00010454545454545455, + "loss": 1.6917, "step": 541 }, { "epoch": 0.69, - "grad_norm": 0.271484375, - "learning_rate": 1.534736842105263e-05, - "loss": 1.955, + "grad_norm": 0.314453125, + "learning_rate": 0.00010432900432900434, + "loss": 1.7533, "step": 542 }, { "epoch": 0.69, - "grad_norm": 0.224609375, - "learning_rate": 1.533684210526316e-05, - "loss": 1.8977, + "grad_norm": 0.390625, + "learning_rate": 0.00010411255411255412, + "loss": 1.699, "step": 543 }, { "epoch": 0.7, - "grad_norm": 0.2421875, - "learning_rate": 1.5326315789473684e-05, - "loss": 1.9117, + "grad_norm": 0.32421875, + "learning_rate": 0.00010389610389610389, + "loss": 1.7324, "step": 544 }, { "epoch": 0.7, - "grad_norm": 0.2265625, - "learning_rate": 1.5315789473684212e-05, - "loss": 1.9683, + "grad_norm": 0.3515625, + "learning_rate": 0.00010367965367965368, + "loss": 1.7698, "step": 545 }, { "epoch": 0.7, - "grad_norm": 0.25, - "learning_rate": 1.530526315789474e-05, - "loss": 1.9521, + "grad_norm": 0.294921875, + "learning_rate": 0.00010346320346320346, + "loss": 1.7495, "step": 546 }, { "epoch": 0.7, - "grad_norm": 0.26171875, - "learning_rate": 1.5294736842105265e-05, - "loss": 1.9898, + "grad_norm": 0.318359375, + "learning_rate": 0.00010324675324675325, + "loss": 1.7885, "step": 547 }, { "epoch": 0.7, - "grad_norm": 0.220703125, - "learning_rate": 1.528421052631579e-05, - "loss": 1.9669, + "grad_norm": 0.310546875, + "learning_rate": 0.00010303030303030303, + "loss": 1.7624, "step": 548 }, { "epoch": 0.7, - "grad_norm": 0.2314453125, - "learning_rate": 1.5273684210526318e-05, - "loss": 1.9327, + "grad_norm": 0.306640625, + "learning_rate": 0.00010281385281385283, + "loss": 1.7373, "step": 549 }, { "epoch": 0.7, - "grad_norm": 0.216796875, - "learning_rate": 1.5263157894736846e-05, - "loss": 1.9592, + "grad_norm": 0.353515625, + "learning_rate": 0.00010259740259740261, + "loss": 1.7663, + "step": 550 + }, + { + "epoch": 0.7, + "eval_loss": 1.7420332431793213, + "eval_runtime": 125.4197, + "eval_samples_per_second": 39.866, + "eval_steps_per_second": 1.252, "step": 550 }, { "epoch": 0.71, - "grad_norm": 0.25, - "learning_rate": 1.525263157894737e-05, - "loss": 1.9977, + "grad_norm": 0.326171875, + "learning_rate": 0.00010238095238095237, + "loss": 1.8104, "step": 551 }, { "epoch": 0.71, - "grad_norm": 0.22265625, - "learning_rate": 1.5242105263157897e-05, - "loss": 1.9629, + "grad_norm": 0.34375, + "learning_rate": 0.00010216450216450218, + "loss": 1.7708, "step": 552 }, { "epoch": 0.71, - "grad_norm": 0.328125, - "learning_rate": 1.5231578947368421e-05, - "loss": 1.8902, + "grad_norm": 0.3984375, + "learning_rate": 0.00010194805194805195, + "loss": 1.7057, "step": 553 }, { "epoch": 0.71, - "grad_norm": 0.275390625, - "learning_rate": 1.5221052631578948e-05, - "loss": 1.8989, + "grad_norm": 0.328125, + "learning_rate": 0.00010173160173160174, + "loss": 1.718, "step": 554 }, { "epoch": 0.71, - "grad_norm": 0.302734375, - "learning_rate": 1.5210526315789476e-05, - "loss": 1.8974, + "grad_norm": 0.3671875, + "learning_rate": 0.00010151515151515152, + "loss": 1.6913, "step": 555 }, { "epoch": 0.71, - "grad_norm": 0.2490234375, - "learning_rate": 1.5200000000000002e-05, - "loss": 1.9185, + "grad_norm": 0.5078125, + "learning_rate": 0.0001012987012987013, + "loss": 1.7444, "step": 556 }, { "epoch": 0.71, - "grad_norm": 0.2275390625, - "learning_rate": 1.5189473684210526e-05, - "loss": 1.9483, + "grad_norm": 0.357421875, + "learning_rate": 0.00010108225108225109, + "loss": 1.7699, "step": 557 }, { "epoch": 0.71, - "grad_norm": 0.2314453125, - "learning_rate": 1.5178947368421053e-05, - "loss": 1.9989, + "grad_norm": 0.423828125, + "learning_rate": 0.00010086580086580086, + "loss": 1.8066, "step": 558 }, { "epoch": 0.72, - "grad_norm": 0.2470703125, - "learning_rate": 1.516842105263158e-05, - "loss": 1.9074, + "grad_norm": 0.333984375, + "learning_rate": 0.00010064935064935067, + "loss": 1.7054, "step": 559 }, { "epoch": 0.72, - "grad_norm": 0.2275390625, - "learning_rate": 1.5157894736842107e-05, - "loss": 1.9551, + "grad_norm": 0.41796875, + "learning_rate": 0.00010043290043290043, + "loss": 1.7562, "step": 560 }, { "epoch": 0.72, - "grad_norm": 0.2373046875, - "learning_rate": 1.5147368421052633e-05, - "loss": 1.8873, + "grad_norm": 0.41015625, + "learning_rate": 0.00010021645021645023, + "loss": 1.6979, "step": 561 }, { "epoch": 0.72, - "grad_norm": 0.236328125, - "learning_rate": 1.5136842105263158e-05, - "loss": 1.9896, + "grad_norm": 0.34765625, + "learning_rate": 0.0001, + "loss": 1.8067, "step": 562 }, { "epoch": 0.72, - "grad_norm": 0.2734375, - "learning_rate": 1.5126315789473684e-05, - "loss": 1.9767, + "grad_norm": 0.31640625, + "learning_rate": 9.978354978354978e-05, + "loss": 1.7718, "step": 563 }, { "epoch": 0.72, - "grad_norm": 0.271484375, - "learning_rate": 1.5115789473684212e-05, - "loss": 2.0132, + "grad_norm": 0.36328125, + "learning_rate": 9.956709956709958e-05, + "loss": 1.8256, "step": 564 }, { "epoch": 0.72, - "grad_norm": 0.23046875, - "learning_rate": 1.5105263157894739e-05, - "loss": 1.9348, + "grad_norm": 0.33984375, + "learning_rate": 9.935064935064936e-05, + "loss": 1.7328, "step": 565 }, { "epoch": 0.72, - "grad_norm": 0.2060546875, - "learning_rate": 1.5094736842105263e-05, - "loss": 1.8899, + "grad_norm": 0.361328125, + "learning_rate": 9.913419913419914e-05, + "loss": 1.6943, "step": 566 }, { "epoch": 0.73, - "grad_norm": 0.2578125, - "learning_rate": 1.508421052631579e-05, - "loss": 1.9548, + "grad_norm": 0.380859375, + "learning_rate": 9.891774891774892e-05, + "loss": 1.7398, "step": 567 }, { "epoch": 0.73, - "grad_norm": 0.2392578125, - "learning_rate": 1.5073684210526316e-05, - "loss": 1.8136, + "grad_norm": 0.39453125, + "learning_rate": 9.870129870129871e-05, + "loss": 1.6486, "step": 568 }, { "epoch": 0.73, - "grad_norm": 0.3203125, - "learning_rate": 1.5063157894736844e-05, - "loss": 1.8954, + "grad_norm": 0.388671875, + "learning_rate": 9.848484848484849e-05, + "loss": 1.7058, "step": 569 }, { "epoch": 0.73, - "grad_norm": 0.2041015625, - "learning_rate": 1.505263157894737e-05, - "loss": 1.9079, + "grad_norm": 0.30859375, + "learning_rate": 9.826839826839827e-05, + "loss": 1.7299, "step": 570 }, { "epoch": 0.73, - "grad_norm": 0.2373046875, - "learning_rate": 1.5042105263157895e-05, - "loss": 1.9658, + "grad_norm": 0.30078125, + "learning_rate": 9.805194805194806e-05, + "loss": 1.7634, "step": 571 }, { "epoch": 0.73, - "grad_norm": 0.232421875, - "learning_rate": 1.5031578947368421e-05, - "loss": 1.9341, + "grad_norm": 0.34375, + "learning_rate": 9.783549783549783e-05, + "loss": 1.7341, "step": 572 }, { "epoch": 0.73, - "grad_norm": 0.228515625, - "learning_rate": 1.502105263157895e-05, - "loss": 1.9714, + "grad_norm": 0.337890625, + "learning_rate": 9.761904761904762e-05, + "loss": 1.7888, "step": 573 }, { "epoch": 0.73, - "grad_norm": 0.25, - "learning_rate": 1.5010526315789476e-05, - "loss": 1.9933, + "grad_norm": 0.34375, + "learning_rate": 9.74025974025974e-05, + "loss": 1.7765, "step": 574 }, { "epoch": 0.74, - "grad_norm": 0.3359375, - "learning_rate": 1.5000000000000002e-05, - "loss": 1.9723, + "grad_norm": 0.55078125, + "learning_rate": 9.71861471861472e-05, + "loss": 1.7764, "step": 575 }, { "epoch": 0.74, - "grad_norm": 0.2431640625, - "learning_rate": 1.4989473684210527e-05, - "loss": 1.9474, + "grad_norm": 0.314453125, + "learning_rate": 9.696969696969698e-05, + "loss": 1.745, "step": 576 }, { "epoch": 0.74, - "grad_norm": 0.2099609375, - "learning_rate": 1.4978947368421053e-05, - "loss": 1.9314, + "grad_norm": 0.357421875, + "learning_rate": 9.675324675324677e-05, + "loss": 1.7382, "step": 577 }, { "epoch": 0.74, - "grad_norm": 0.2373046875, - "learning_rate": 1.4968421052631581e-05, - "loss": 1.8863, + "grad_norm": 0.384765625, + "learning_rate": 9.653679653679654e-05, + "loss": 1.7118, "step": 578 }, { "epoch": 0.74, - "grad_norm": 0.2314453125, - "learning_rate": 1.4957894736842107e-05, - "loss": 1.8646, + "grad_norm": 0.380859375, + "learning_rate": 9.632034632034633e-05, + "loss": 1.6819, "step": 579 }, { "epoch": 0.74, - "grad_norm": 0.2333984375, - "learning_rate": 1.4947368421052632e-05, - "loss": 2.005, + "grad_norm": 0.326171875, + "learning_rate": 9.610389610389611e-05, + "loss": 1.8091, "step": 580 }, { "epoch": 0.74, - "grad_norm": 0.224609375, - "learning_rate": 1.4936842105263158e-05, - "loss": 1.9775, + "grad_norm": 0.380859375, + "learning_rate": 9.588744588744589e-05, + "loss": 1.796, "step": 581 }, { "epoch": 0.74, - "grad_norm": 0.2158203125, - "learning_rate": 1.4926315789473686e-05, - "loss": 1.8805, + "grad_norm": 0.380859375, + "learning_rate": 9.567099567099568e-05, + "loss": 1.7027, "step": 582 }, { "epoch": 0.75, - "grad_norm": 0.248046875, - "learning_rate": 1.4915789473684213e-05, - "loss": 1.9357, + "grad_norm": 0.43359375, + "learning_rate": 9.545454545454546e-05, + "loss": 1.736, "step": 583 }, { "epoch": 0.75, - "grad_norm": 0.201171875, - "learning_rate": 1.4905263157894739e-05, - "loss": 1.931, + "grad_norm": 0.337890625, + "learning_rate": 9.523809523809524e-05, + "loss": 1.738, "step": 584 }, { "epoch": 0.75, - "grad_norm": 0.24609375, - "learning_rate": 1.4894736842105264e-05, - "loss": 1.9545, + "grad_norm": 0.369140625, + "learning_rate": 9.502164502164502e-05, + "loss": 1.7425, "step": 585 }, { "epoch": 0.75, - "grad_norm": 0.244140625, - "learning_rate": 1.488421052631579e-05, - "loss": 1.8772, + "grad_norm": 0.333984375, + "learning_rate": 9.480519480519481e-05, + "loss": 1.6909, "step": 586 }, { "epoch": 0.75, - "grad_norm": 0.212890625, - "learning_rate": 1.4873684210526318e-05, - "loss": 1.9458, + "grad_norm": 0.50390625, + "learning_rate": 9.45887445887446e-05, + "loss": 1.7845, "step": 587 }, { "epoch": 0.75, - "grad_norm": 0.296875, - "learning_rate": 1.4863157894736844e-05, - "loss": 1.9185, + "grad_norm": 0.44140625, + "learning_rate": 9.437229437229437e-05, + "loss": 1.7345, "step": 588 }, { "epoch": 0.75, - "grad_norm": 0.2236328125, - "learning_rate": 1.4852631578947369e-05, - "loss": 1.9933, + "grad_norm": 0.283203125, + "learning_rate": 9.415584415584417e-05, + "loss": 1.7717, "step": 589 }, { "epoch": 0.75, - "grad_norm": 0.23828125, - "learning_rate": 1.4842105263157895e-05, - "loss": 1.9395, + "grad_norm": 0.44921875, + "learning_rate": 9.393939393939395e-05, + "loss": 1.7243, "step": 590 }, { "epoch": 0.76, - "grad_norm": 0.2216796875, - "learning_rate": 1.4831578947368422e-05, - "loss": 1.9606, + "grad_norm": 0.431640625, + "learning_rate": 9.372294372294373e-05, + "loss": 1.7662, "step": 591 }, { "epoch": 0.76, - "grad_norm": 0.2275390625, - "learning_rate": 1.482105263157895e-05, - "loss": 1.8837, + "grad_norm": 0.40625, + "learning_rate": 9.35064935064935e-05, + "loss": 1.7209, "step": 592 }, { "epoch": 0.76, - "grad_norm": 0.21484375, - "learning_rate": 1.4810526315789476e-05, - "loss": 1.8854, + "grad_norm": 0.3828125, + "learning_rate": 9.32900432900433e-05, + "loss": 1.6923, "step": 593 }, { "epoch": 0.76, - "grad_norm": 0.234375, - "learning_rate": 1.48e-05, - "loss": 2.009, + "grad_norm": 0.34375, + "learning_rate": 9.307359307359308e-05, + "loss": 1.8138, "step": 594 }, { "epoch": 0.76, - "grad_norm": 0.2470703125, - "learning_rate": 1.4789473684210527e-05, - "loss": 1.9204, + "grad_norm": 0.3515625, + "learning_rate": 9.285714285714286e-05, + "loss": 1.7474, "step": 595 }, { "epoch": 0.76, - "grad_norm": 0.259765625, - "learning_rate": 1.4778947368421055e-05, - "loss": 1.9478, + "grad_norm": 0.421875, + "learning_rate": 9.264069264069265e-05, + "loss": 1.776, "step": 596 }, { "epoch": 0.76, - "grad_norm": 0.2333984375, - "learning_rate": 1.4768421052631581e-05, - "loss": 1.962, + "grad_norm": 0.361328125, + "learning_rate": 9.242424242424242e-05, + "loss": 1.7624, "step": 597 }, { "epoch": 0.77, - "grad_norm": 0.3046875, - "learning_rate": 1.4757894736842106e-05, - "loss": 1.921, + "grad_norm": 0.341796875, + "learning_rate": 9.220779220779221e-05, + "loss": 1.733, "step": 598 }, { "epoch": 0.77, - "grad_norm": 0.248046875, - "learning_rate": 1.4747368421052632e-05, - "loss": 1.9596, + "grad_norm": 0.326171875, + "learning_rate": 9.199134199134199e-05, + "loss": 1.7617, "step": 599 }, { "epoch": 0.77, - "grad_norm": 0.2734375, - "learning_rate": 1.4736842105263159e-05, - "loss": 1.8835, + "grad_norm": 0.326171875, + "learning_rate": 9.177489177489178e-05, + "loss": 1.6983, "step": 600 }, { "epoch": 0.77, - "eval_loss": 1.921275019645691, - "eval_runtime": 126.2352, - "eval_samples_per_second": 39.609, - "eval_steps_per_second": 1.244, + "eval_loss": 1.7379374504089355, + "eval_runtime": 124.9706, + "eval_samples_per_second": 40.009, + "eval_steps_per_second": 1.256, "step": 600 }, { "epoch": 0.77, - "grad_norm": 0.267578125, - "learning_rate": 1.4726315789473687e-05, - "loss": 1.8885, + "grad_norm": 0.34375, + "learning_rate": 9.155844155844156e-05, + "loss": 1.7099, "step": 601 }, { "epoch": 0.77, - "grad_norm": 0.25, - "learning_rate": 1.4715789473684213e-05, - "loss": 1.8686, + "grad_norm": 0.32421875, + "learning_rate": 9.134199134199136e-05, + "loss": 1.671, "step": 602 }, { "epoch": 0.77, - "grad_norm": 0.2158203125, - "learning_rate": 1.4705263157894738e-05, - "loss": 1.9035, + "grad_norm": 0.302734375, + "learning_rate": 9.112554112554112e-05, + "loss": 1.7249, "step": 603 }, { "epoch": 0.77, - "grad_norm": 0.25, - "learning_rate": 1.4694736842105264e-05, - "loss": 2.0034, + "grad_norm": 0.318359375, + "learning_rate": 9.090909090909092e-05, + "loss": 1.7958, "step": 604 }, { "epoch": 0.77, - "grad_norm": 0.259765625, - "learning_rate": 1.468421052631579e-05, - "loss": 1.8919, + "grad_norm": 0.37109375, + "learning_rate": 9.06926406926407e-05, + "loss": 1.7202, "step": 605 }, { "epoch": 0.78, - "grad_norm": 0.2099609375, - "learning_rate": 1.4673684210526318e-05, - "loss": 1.9118, + "grad_norm": 0.33203125, + "learning_rate": 9.047619047619048e-05, + "loss": 1.7261, "step": 606 }, { "epoch": 0.78, - "grad_norm": 0.2470703125, - "learning_rate": 1.4663157894736843e-05, - "loss": 1.9975, + "grad_norm": 0.32421875, + "learning_rate": 9.025974025974027e-05, + "loss": 1.8016, "step": 607 }, { "epoch": 0.78, - "grad_norm": 0.24609375, - "learning_rate": 1.465263157894737e-05, - "loss": 1.9883, + "grad_norm": 0.310546875, + "learning_rate": 9.004329004329005e-05, + "loss": 1.7957, "step": 608 }, { "epoch": 0.78, - "grad_norm": 0.224609375, - "learning_rate": 1.4642105263157896e-05, - "loss": 1.9589, + "grad_norm": 0.3125, + "learning_rate": 8.982683982683983e-05, + "loss": 1.7856, "step": 609 }, { "epoch": 0.78, - "grad_norm": 0.2138671875, - "learning_rate": 1.4631578947368424e-05, - "loss": 1.9553, + "grad_norm": 0.341796875, + "learning_rate": 8.961038961038961e-05, + "loss": 1.7807, "step": 610 }, { "epoch": 0.78, - "grad_norm": 0.220703125, - "learning_rate": 1.462105263157895e-05, - "loss": 1.9468, + "grad_norm": 0.341796875, + "learning_rate": 8.93939393939394e-05, + "loss": 1.7536, "step": 611 }, { "epoch": 0.78, - "grad_norm": 0.2041015625, - "learning_rate": 1.4610526315789474e-05, - "loss": 1.9506, + "grad_norm": 0.3125, + "learning_rate": 8.917748917748918e-05, + "loss": 1.7499, "step": 612 }, { "epoch": 0.78, - "grad_norm": 0.29296875, - "learning_rate": 1.46e-05, - "loss": 1.961, + "grad_norm": 0.365234375, + "learning_rate": 8.896103896103896e-05, + "loss": 1.7726, "step": 613 }, { "epoch": 0.79, - "grad_norm": 0.212890625, - "learning_rate": 1.4589473684210527e-05, - "loss": 1.9417, + "grad_norm": 0.37109375, + "learning_rate": 8.874458874458876e-05, + "loss": 1.7558, "step": 614 }, { "epoch": 0.79, - "grad_norm": 0.2080078125, - "learning_rate": 1.4578947368421055e-05, - "loss": 1.9096, + "grad_norm": 0.298828125, + "learning_rate": 8.852813852813854e-05, + "loss": 1.726, "step": 615 }, { "epoch": 0.79, - "grad_norm": 0.25, - "learning_rate": 1.456842105263158e-05, - "loss": 1.9419, + "grad_norm": 0.330078125, + "learning_rate": 8.831168831168831e-05, + "loss": 1.7632, "step": 616 }, { "epoch": 0.79, - "grad_norm": 0.32421875, - "learning_rate": 1.4557894736842106e-05, - "loss": 1.9151, + "grad_norm": 0.361328125, + "learning_rate": 8.80952380952381e-05, + "loss": 1.7293, "step": 617 }, { "epoch": 0.79, - "grad_norm": 0.220703125, - "learning_rate": 1.4547368421052632e-05, - "loss": 1.9091, + "grad_norm": 0.337890625, + "learning_rate": 8.787878787878789e-05, + "loss": 1.7263, "step": 618 }, { "epoch": 0.79, - "grad_norm": 0.2333984375, - "learning_rate": 1.4536842105263159e-05, - "loss": 2.0308, + "grad_norm": 0.35546875, + "learning_rate": 8.766233766233767e-05, + "loss": 1.8403, "step": 619 }, { "epoch": 0.79, - "grad_norm": 0.265625, - "learning_rate": 1.4526315789473687e-05, - "loss": 1.8765, + "grad_norm": 0.33984375, + "learning_rate": 8.744588744588745e-05, + "loss": 1.6896, "step": 620 }, { "epoch": 0.79, - "grad_norm": 0.294921875, - "learning_rate": 1.4515789473684211e-05, - "loss": 1.9832, + "grad_norm": 0.328125, + "learning_rate": 8.722943722943724e-05, + "loss": 1.797, "step": 621 }, { "epoch": 0.8, - "grad_norm": 0.244140625, - "learning_rate": 1.4505263157894738e-05, - "loss": 1.9988, + "grad_norm": 0.439453125, + "learning_rate": 8.701298701298701e-05, + "loss": 1.7971, "step": 622 }, { "epoch": 0.8, - "grad_norm": 0.2470703125, - "learning_rate": 1.4494736842105264e-05, - "loss": 1.9625, + "grad_norm": 0.39453125, + "learning_rate": 8.67965367965368e-05, + "loss": 1.7697, "step": 623 }, { "epoch": 0.8, - "grad_norm": 0.2890625, - "learning_rate": 1.4484210526315792e-05, - "loss": 1.9392, + "grad_norm": 0.326171875, + "learning_rate": 8.658008658008658e-05, + "loss": 1.7567, "step": 624 }, { "epoch": 0.8, - "grad_norm": 0.2177734375, - "learning_rate": 1.4473684210526317e-05, - "loss": 1.8315, + "grad_norm": 0.3203125, + "learning_rate": 8.636363636363637e-05, + "loss": 1.659, "step": 625 }, { "epoch": 0.8, - "grad_norm": 0.2421875, - "learning_rate": 1.4463157894736843e-05, - "loss": 1.9505, + "grad_norm": 0.3671875, + "learning_rate": 8.614718614718615e-05, + "loss": 1.7546, "step": 626 }, { "epoch": 0.8, - "grad_norm": 0.2119140625, - "learning_rate": 1.445263157894737e-05, - "loss": 1.8436, + "grad_norm": 0.32421875, + "learning_rate": 8.593073593073593e-05, + "loss": 1.6824, "step": 627 }, { "epoch": 0.8, - "grad_norm": 0.267578125, - "learning_rate": 1.4442105263157896e-05, - "loss": 1.8915, + "grad_norm": 0.42578125, + "learning_rate": 8.571428571428571e-05, + "loss": 1.7049, "step": 628 }, { "epoch": 0.8, - "grad_norm": 0.208984375, - "learning_rate": 1.4431578947368424e-05, - "loss": 1.8428, + "grad_norm": 0.3125, + "learning_rate": 8.549783549783549e-05, + "loss": 1.6639, "step": 629 }, { "epoch": 0.81, - "grad_norm": 0.2119140625, - "learning_rate": 1.4421052631578948e-05, - "loss": 1.9378, + "grad_norm": 0.337890625, + "learning_rate": 8.528138528138529e-05, + "loss": 1.7667, "step": 630 }, { "epoch": 0.81, - "grad_norm": 0.2197265625, - "learning_rate": 1.4410526315789475e-05, - "loss": 1.961, + "grad_norm": 0.326171875, + "learning_rate": 8.506493506493507e-05, + "loss": 1.7703, "step": 631 }, { "epoch": 0.81, - "grad_norm": 0.2392578125, - "learning_rate": 1.4400000000000001e-05, - "loss": 1.9327, + "grad_norm": 0.388671875, + "learning_rate": 8.484848484848486e-05, + "loss": 1.7526, "step": 632 }, { "epoch": 0.81, - "grad_norm": 0.271484375, - "learning_rate": 1.4389473684210526e-05, - "loss": 1.9544, + "grad_norm": 0.376953125, + "learning_rate": 8.463203463203464e-05, + "loss": 1.7658, "step": 633 }, { "epoch": 0.81, - "grad_norm": 0.2275390625, - "learning_rate": 1.4378947368421054e-05, - "loss": 1.9896, + "grad_norm": 0.318359375, + "learning_rate": 8.441558441558442e-05, + "loss": 1.7888, "step": 634 }, { "epoch": 0.81, - "grad_norm": 0.2490234375, - "learning_rate": 1.436842105263158e-05, - "loss": 1.8814, + "grad_norm": 0.3125, + "learning_rate": 8.41991341991342e-05, + "loss": 1.7197, "step": 635 }, { "epoch": 0.81, - "grad_norm": 0.2138671875, - "learning_rate": 1.4357894736842106e-05, - "loss": 1.8679, + "grad_norm": 0.328125, + "learning_rate": 8.398268398268399e-05, + "loss": 1.6745, "step": 636 }, { "epoch": 0.82, - "grad_norm": 0.2275390625, - "learning_rate": 1.4347368421052633e-05, - "loss": 1.894, + "grad_norm": 0.369140625, + "learning_rate": 8.376623376623377e-05, + "loss": 1.7168, "step": 637 }, { "epoch": 0.82, - "grad_norm": 0.21875, - "learning_rate": 1.433684210526316e-05, - "loss": 1.8991, + "grad_norm": 0.33984375, + "learning_rate": 8.354978354978355e-05, + "loss": 1.6904, "step": 638 }, { "epoch": 0.82, - "grad_norm": 0.2734375, - "learning_rate": 1.4326315789473685e-05, - "loss": 1.8635, + "grad_norm": 0.396484375, + "learning_rate": 8.333333333333334e-05, + "loss": 1.6792, "step": 639 }, { "epoch": 0.82, - "grad_norm": 0.2373046875, - "learning_rate": 1.4315789473684212e-05, - "loss": 1.9113, + "grad_norm": 0.34375, + "learning_rate": 8.311688311688312e-05, + "loss": 1.7299, "step": 640 }, { "epoch": 0.82, - "grad_norm": 0.244140625, - "learning_rate": 1.4305263157894738e-05, - "loss": 1.9089, + "grad_norm": 0.365234375, + "learning_rate": 8.29004329004329e-05, + "loss": 1.7334, "step": 641 }, { "epoch": 0.82, - "grad_norm": 0.1923828125, - "learning_rate": 1.4294736842105263e-05, - "loss": 1.8984, + "grad_norm": 0.32421875, + "learning_rate": 8.268398268398268e-05, + "loss": 1.7236, "step": 642 }, { "epoch": 0.82, - "grad_norm": 0.21875, - "learning_rate": 1.4284210526315792e-05, - "loss": 1.9174, + "grad_norm": 0.408203125, + "learning_rate": 8.246753246753248e-05, + "loss": 1.7398, "step": 643 }, { "epoch": 0.82, - "grad_norm": 0.2412109375, - "learning_rate": 1.4273684210526317e-05, - "loss": 1.9419, + "grad_norm": 0.337890625, + "learning_rate": 8.225108225108226e-05, + "loss": 1.7407, "step": 644 }, { "epoch": 0.83, - "grad_norm": 0.232421875, - "learning_rate": 1.4263157894736843e-05, - "loss": 2.0331, + "grad_norm": 0.34375, + "learning_rate": 8.203463203463204e-05, + "loss": 1.8405, "step": 645 }, { "epoch": 0.83, - "grad_norm": 0.216796875, - "learning_rate": 1.425263157894737e-05, - "loss": 1.8914, + "grad_norm": 0.306640625, + "learning_rate": 8.181818181818183e-05, + "loss": 1.7225, "step": 646 }, { "epoch": 0.83, - "grad_norm": 0.21875, - "learning_rate": 1.4242105263157894e-05, - "loss": 2.0047, + "grad_norm": 0.318359375, + "learning_rate": 8.16017316017316e-05, + "loss": 1.8285, "step": 647 }, { "epoch": 0.83, - "grad_norm": 0.267578125, - "learning_rate": 1.4231578947368422e-05, - "loss": 1.955, + "grad_norm": 0.4453125, + "learning_rate": 8.138528138528139e-05, + "loss": 1.7646, "step": 648 }, { "epoch": 0.83, - "grad_norm": 0.2470703125, - "learning_rate": 1.4221052631578949e-05, - "loss": 1.9201, + "grad_norm": 0.3203125, + "learning_rate": 8.116883116883117e-05, + "loss": 1.7172, "step": 649 }, { "epoch": 0.83, - "grad_norm": 0.283203125, - "learning_rate": 1.4210526315789475e-05, - "loss": 1.9968, + "grad_norm": 0.39453125, + "learning_rate": 8.095238095238096e-05, + "loss": 1.8033, "step": 650 }, { "epoch": 0.83, - "grad_norm": 0.2275390625, - "learning_rate": 1.4200000000000001e-05, - "loss": 1.939, + "eval_loss": 1.7356816530227661, + "eval_runtime": 125.4294, + "eval_samples_per_second": 39.863, + "eval_steps_per_second": 1.252, + "step": 650 + }, + { + "epoch": 0.83, + "grad_norm": 0.30859375, + "learning_rate": 8.073593073593074e-05, + "loss": 1.7494, "step": 651 }, { "epoch": 0.83, - "grad_norm": 0.2412109375, - "learning_rate": 1.418947368421053e-05, - "loss": 1.9347, + "grad_norm": 0.40234375, + "learning_rate": 8.051948051948052e-05, + "loss": 1.7573, "step": 652 }, { "epoch": 0.84, - "grad_norm": 0.2421875, - "learning_rate": 1.4178947368421054e-05, - "loss": 1.9764, + "grad_norm": 0.46484375, + "learning_rate": 8.03030303030303e-05, + "loss": 1.7712, "step": 653 }, { "epoch": 0.84, - "grad_norm": 0.265625, - "learning_rate": 1.416842105263158e-05, - "loss": 1.897, + "grad_norm": 0.40234375, + "learning_rate": 8.008658008658008e-05, + "loss": 1.6894, "step": 654 }, { "epoch": 0.84, - "grad_norm": 0.2177734375, - "learning_rate": 1.4157894736842107e-05, - "loss": 1.8547, + "grad_norm": 0.361328125, + "learning_rate": 7.987012987012987e-05, + "loss": 1.6714, "step": 655 }, { "epoch": 0.84, - "grad_norm": 0.224609375, - "learning_rate": 1.4147368421052631e-05, - "loss": 1.9668, + "grad_norm": 0.33984375, + "learning_rate": 7.965367965367965e-05, + "loss": 1.801, "step": 656 }, { "epoch": 0.84, - "grad_norm": 0.255859375, - "learning_rate": 1.413684210526316e-05, - "loss": 1.9344, + "grad_norm": 0.41015625, + "learning_rate": 7.943722943722945e-05, + "loss": 1.75, "step": 657 }, { "epoch": 0.84, - "grad_norm": 0.302734375, - "learning_rate": 1.4126315789473686e-05, - "loss": 1.8808, + "grad_norm": 0.421875, + "learning_rate": 7.922077922077923e-05, + "loss": 1.6875, "step": 658 }, { "epoch": 0.84, - "grad_norm": 0.2294921875, - "learning_rate": 1.4115789473684212e-05, - "loss": 1.8283, + "grad_norm": 0.421875, + "learning_rate": 7.900432900432901e-05, + "loss": 1.6635, "step": 659 }, { "epoch": 0.84, - "grad_norm": 0.21875, - "learning_rate": 1.4105263157894738e-05, - "loss": 1.9727, + "grad_norm": 0.302734375, + "learning_rate": 7.878787878787879e-05, + "loss": 1.782, "step": 660 }, { "epoch": 0.85, - "grad_norm": 0.275390625, - "learning_rate": 1.4094736842105263e-05, - "loss": 1.8578, + "grad_norm": 0.3671875, + "learning_rate": 7.857142857142858e-05, + "loss": 1.6838, "step": 661 }, { "epoch": 0.85, - "grad_norm": 0.2275390625, - "learning_rate": 1.4084210526315791e-05, - "loss": 1.8984, + "grad_norm": 0.341796875, + "learning_rate": 7.835497835497836e-05, + "loss": 1.6957, "step": 662 }, { "epoch": 0.85, - "grad_norm": 0.2255859375, - "learning_rate": 1.4073684210526317e-05, - "loss": 1.9318, + "grad_norm": 0.3359375, + "learning_rate": 7.813852813852814e-05, + "loss": 1.7366, "step": 663 }, { "epoch": 0.85, - "grad_norm": 0.279296875, - "learning_rate": 1.4063157894736844e-05, - "loss": 1.87, + "grad_norm": 0.404296875, + "learning_rate": 7.792207792207793e-05, + "loss": 1.6709, "step": 664 }, { "epoch": 0.85, - "grad_norm": 0.2333984375, - "learning_rate": 1.4052631578947368e-05, - "loss": 1.9685, + "grad_norm": 0.373046875, + "learning_rate": 7.770562770562771e-05, + "loss": 1.7653, "step": 665 }, { "epoch": 0.85, - "grad_norm": 0.2265625, - "learning_rate": 1.4042105263157896e-05, - "loss": 1.9345, + "grad_norm": 0.375, + "learning_rate": 7.748917748917749e-05, + "loss": 1.7467, "step": 666 }, { "epoch": 0.85, - "grad_norm": 0.224609375, - "learning_rate": 1.4031578947368423e-05, - "loss": 1.9365, + "grad_norm": 0.3046875, + "learning_rate": 7.727272727272727e-05, + "loss": 1.7499, "step": 667 }, { "epoch": 0.85, - "grad_norm": 0.234375, - "learning_rate": 1.4021052631578949e-05, - "loss": 1.9529, + "grad_norm": 0.337890625, + "learning_rate": 7.705627705627707e-05, + "loss": 1.7462, "step": 668 }, { "epoch": 0.86, - "grad_norm": 0.236328125, - "learning_rate": 1.4010526315789475e-05, - "loss": 1.8801, + "grad_norm": 0.3828125, + "learning_rate": 7.683982683982685e-05, + "loss": 1.6883, "step": 669 }, { "epoch": 0.86, - "grad_norm": 0.2265625, - "learning_rate": 1.4e-05, - "loss": 1.9642, + "grad_norm": 0.337890625, + "learning_rate": 7.662337662337662e-05, + "loss": 1.7674, "step": 670 }, { "epoch": 0.86, - "grad_norm": 0.2392578125, - "learning_rate": 1.3989473684210528e-05, - "loss": 1.9285, + "grad_norm": 0.37890625, + "learning_rate": 7.640692640692642e-05, + "loss": 1.7388, "step": 671 }, { "epoch": 0.86, - "grad_norm": 0.2333984375, - "learning_rate": 1.3978947368421054e-05, - "loss": 1.9131, + "grad_norm": 0.33203125, + "learning_rate": 7.619047619047618e-05, + "loss": 1.7206, "step": 672 }, { "epoch": 0.86, - "grad_norm": 0.21875, - "learning_rate": 1.396842105263158e-05, - "loss": 1.9516, + "grad_norm": 0.341796875, + "learning_rate": 7.597402597402598e-05, + "loss": 1.749, "step": 673 }, { "epoch": 0.86, - "grad_norm": 0.2392578125, - "learning_rate": 1.3957894736842105e-05, - "loss": 1.9452, + "grad_norm": 0.361328125, + "learning_rate": 7.575757575757576e-05, + "loss": 1.7318, "step": 674 }, { "epoch": 0.86, - "grad_norm": 0.322265625, - "learning_rate": 1.3947368421052631e-05, - "loss": 1.9174, + "grad_norm": 0.333984375, + "learning_rate": 7.554112554112555e-05, + "loss": 1.7234, "step": 675 }, { "epoch": 0.87, - "grad_norm": 0.220703125, - "learning_rate": 1.393684210526316e-05, - "loss": 1.8182, + "grad_norm": 0.3359375, + "learning_rate": 7.532467532467533e-05, + "loss": 1.6407, "step": 676 }, { "epoch": 0.87, - "grad_norm": 0.2431640625, - "learning_rate": 1.3926315789473686e-05, - "loss": 2.0103, + "grad_norm": 0.431640625, + "learning_rate": 7.510822510822511e-05, + "loss": 1.7951, "step": 677 }, { "epoch": 0.87, - "grad_norm": 0.345703125, - "learning_rate": 1.3915789473684212e-05, - "loss": 2.0103, + "grad_norm": 0.49609375, + "learning_rate": 7.489177489177489e-05, + "loss": 1.7958, "step": 678 }, { "epoch": 0.87, - "grad_norm": 0.2490234375, - "learning_rate": 1.3905263157894737e-05, - "loss": 1.9405, + "grad_norm": 0.345703125, + "learning_rate": 7.467532467532467e-05, + "loss": 1.745, "step": 679 }, { "epoch": 0.87, - "grad_norm": 0.20703125, - "learning_rate": 1.3894736842105265e-05, - "loss": 1.8172, + "grad_norm": 0.396484375, + "learning_rate": 7.445887445887446e-05, + "loss": 1.6285, "step": 680 }, { "epoch": 0.87, - "grad_norm": 0.21484375, - "learning_rate": 1.3884210526315791e-05, - "loss": 1.9886, + "grad_norm": 0.390625, + "learning_rate": 7.424242424242424e-05, + "loss": 1.792, "step": 681 }, { "epoch": 0.87, - "grad_norm": 0.259765625, - "learning_rate": 1.3873684210526317e-05, - "loss": 2.0317, + "grad_norm": 0.466796875, + "learning_rate": 7.402597402597404e-05, + "loss": 1.8207, "step": 682 }, { "epoch": 0.87, - "grad_norm": 0.275390625, - "learning_rate": 1.3863157894736842e-05, - "loss": 1.889, + "grad_norm": 0.4609375, + "learning_rate": 7.380952380952382e-05, + "loss": 1.7013, "step": 683 }, { "epoch": 0.88, - "grad_norm": 0.232421875, - "learning_rate": 1.3852631578947368e-05, - "loss": 1.9249, + "grad_norm": 0.349609375, + "learning_rate": 7.35930735930736e-05, + "loss": 1.759, "step": 684 }, { "epoch": 0.88, - "grad_norm": 0.2294921875, - "learning_rate": 1.3842105263157896e-05, - "loss": 1.939, + "grad_norm": 0.3671875, + "learning_rate": 7.337662337662338e-05, + "loss": 1.7398, "step": 685 }, { "epoch": 0.88, - "grad_norm": 0.2119140625, - "learning_rate": 1.3831578947368423e-05, - "loss": 1.9113, + "grad_norm": 0.318359375, + "learning_rate": 7.316017316017317e-05, + "loss": 1.6995, "step": 686 }, { "epoch": 0.88, - "grad_norm": 0.224609375, - "learning_rate": 1.3821052631578949e-05, - "loss": 1.9513, + "grad_norm": 0.4296875, + "learning_rate": 7.294372294372295e-05, + "loss": 1.7617, "step": 687 }, { "epoch": 0.88, - "grad_norm": 0.2490234375, - "learning_rate": 1.3810526315789474e-05, - "loss": 1.9086, + "grad_norm": 0.431640625, + "learning_rate": 7.272727272727273e-05, + "loss": 1.6959, "step": 688 }, { "epoch": 0.88, - "grad_norm": 0.2392578125, - "learning_rate": 1.38e-05, - "loss": 1.9919, + "grad_norm": 0.322265625, + "learning_rate": 7.251082251082252e-05, + "loss": 1.7879, "step": 689 }, { "epoch": 0.88, - "grad_norm": 0.21484375, - "learning_rate": 1.3789473684210528e-05, - "loss": 1.8392, + "grad_norm": 0.35546875, + "learning_rate": 7.229437229437229e-05, + "loss": 1.6613, "step": 690 }, { "epoch": 0.88, - "grad_norm": 0.20703125, - "learning_rate": 1.3778947368421054e-05, - "loss": 1.8287, + "grad_norm": 0.3515625, + "learning_rate": 7.207792207792208e-05, + "loss": 1.6485, "step": 691 }, { "epoch": 0.89, - "grad_norm": 0.267578125, - "learning_rate": 1.3768421052631579e-05, - "loss": 1.9327, + "grad_norm": 0.400390625, + "learning_rate": 7.186147186147186e-05, + "loss": 1.719, "step": 692 }, { "epoch": 0.89, - "grad_norm": 0.236328125, - "learning_rate": 1.3757894736842105e-05, - "loss": 1.9435, + "grad_norm": 0.32421875, + "learning_rate": 7.164502164502165e-05, + "loss": 1.7316, "step": 693 }, { "epoch": 0.89, - "grad_norm": 0.2265625, - "learning_rate": 1.3747368421052633e-05, - "loss": 1.9514, + "grad_norm": 0.33984375, + "learning_rate": 7.142857142857143e-05, + "loss": 1.7509, "step": 694 }, { "epoch": 0.89, - "grad_norm": 0.283203125, - "learning_rate": 1.373684210526316e-05, - "loss": 1.9966, + "grad_norm": 0.375, + "learning_rate": 7.121212121212121e-05, + "loss": 1.785, "step": 695 }, { "epoch": 0.89, - "grad_norm": 0.240234375, - "learning_rate": 1.3726315789473686e-05, - "loss": 1.9452, + "grad_norm": 0.328125, + "learning_rate": 7.099567099567101e-05, + "loss": 1.7643, "step": 696 }, { "epoch": 0.89, - "grad_norm": 0.21484375, - "learning_rate": 1.371578947368421e-05, - "loss": 1.8826, + "grad_norm": 0.341796875, + "learning_rate": 7.077922077922077e-05, + "loss": 1.7058, "step": 697 }, { "epoch": 0.89, - "grad_norm": 0.224609375, - "learning_rate": 1.3705263157894737e-05, - "loss": 1.9868, + "grad_norm": 0.34375, + "learning_rate": 7.056277056277057e-05, + "loss": 1.807, "step": 698 }, { "epoch": 0.89, - "grad_norm": 0.26171875, - "learning_rate": 1.3694736842105265e-05, - "loss": 1.8164, + "grad_norm": 0.349609375, + "learning_rate": 7.034632034632035e-05, + "loss": 1.6489, "step": 699 }, { "epoch": 0.9, - "grad_norm": 0.2333984375, - "learning_rate": 1.3684210526315791e-05, - "loss": 2.013, + "grad_norm": 0.359375, + "learning_rate": 7.012987012987014e-05, + "loss": 1.8146, "step": 700 }, { "epoch": 0.9, - "grad_norm": 0.208984375, - "learning_rate": 1.3673684210526316e-05, - "loss": 1.8337, + "eval_loss": 1.7302906513214111, + "eval_runtime": 125.8875, + "eval_samples_per_second": 39.718, + "eval_steps_per_second": 1.247, + "step": 700 + }, + { + "epoch": 0.9, + "grad_norm": 0.298828125, + "learning_rate": 6.991341991341992e-05, + "loss": 1.6505, "step": 701 }, { "epoch": 0.9, - "grad_norm": 0.23828125, - "learning_rate": 1.3663157894736842e-05, - "loss": 1.9138, + "grad_norm": 0.39453125, + "learning_rate": 6.96969696969697e-05, + "loss": 1.7259, "step": 702 }, { "epoch": 0.9, - "grad_norm": 0.24609375, - "learning_rate": 1.3652631578947369e-05, - "loss": 2.0003, + "grad_norm": 0.314453125, + "learning_rate": 6.948051948051948e-05, + "loss": 1.81, "step": 703 }, { "epoch": 0.9, - "grad_norm": 0.220703125, - "learning_rate": 1.3642105263157897e-05, - "loss": 1.9673, + "grad_norm": 0.3515625, + "learning_rate": 6.926406926406926e-05, + "loss": 1.7801, "step": 704 }, { "epoch": 0.9, - "grad_norm": 0.255859375, - "learning_rate": 1.3631578947368423e-05, - "loss": 1.9059, + "grad_norm": 0.3359375, + "learning_rate": 6.904761904761905e-05, + "loss": 1.7213, "step": 705 }, { "epoch": 0.9, - "grad_norm": 0.2001953125, - "learning_rate": 1.3621052631578948e-05, - "loss": 1.9658, + "grad_norm": 0.3125, + "learning_rate": 6.883116883116883e-05, + "loss": 1.8054, "step": 706 }, { "epoch": 0.9, - "grad_norm": 0.2109375, - "learning_rate": 1.3610526315789474e-05, - "loss": 1.9275, + "grad_norm": 0.31640625, + "learning_rate": 6.861471861471862e-05, + "loss": 1.7339, "step": 707 }, { "epoch": 0.91, - "grad_norm": 0.2333984375, - "learning_rate": 1.3600000000000002e-05, - "loss": 1.9687, + "grad_norm": 0.328125, + "learning_rate": 6.83982683982684e-05, + "loss": 1.79, "step": 708 }, { "epoch": 0.91, - "grad_norm": 0.265625, - "learning_rate": 1.3589473684210528e-05, - "loss": 1.8378, + "grad_norm": 0.36328125, + "learning_rate": 6.818181818181818e-05, + "loss": 1.6615, "step": 709 }, { "epoch": 0.91, - "grad_norm": 0.2138671875, - "learning_rate": 1.3578947368421055e-05, - "loss": 2.0384, + "grad_norm": 0.306640625, + "learning_rate": 6.796536796536796e-05, + "loss": 1.8298, "step": 710 }, { "epoch": 0.91, - "grad_norm": 0.21875, - "learning_rate": 1.356842105263158e-05, - "loss": 1.8796, + "grad_norm": 0.298828125, + "learning_rate": 6.774891774891774e-05, + "loss": 1.6835, "step": 711 }, { "epoch": 0.91, - "grad_norm": 0.2265625, - "learning_rate": 1.3557894736842106e-05, - "loss": 1.9842, + "grad_norm": 0.32421875, + "learning_rate": 6.753246753246754e-05, + "loss": 1.7739, "step": 712 }, { "epoch": 0.91, - "grad_norm": 0.2412109375, - "learning_rate": 1.3547368421052634e-05, - "loss": 1.9426, + "grad_norm": 0.337890625, + "learning_rate": 6.731601731601732e-05, + "loss": 1.7431, "step": 713 }, { "epoch": 0.91, - "grad_norm": 0.27734375, - "learning_rate": 1.353684210526316e-05, - "loss": 1.9857, + "grad_norm": 0.337890625, + "learning_rate": 6.709956709956711e-05, + "loss": 1.7816, "step": 714 }, { "epoch": 0.91, - "grad_norm": 0.2021484375, - "learning_rate": 1.3526315789473685e-05, - "loss": 1.9319, + "grad_norm": 0.32421875, + "learning_rate": 6.688311688311688e-05, + "loss": 1.7516, "step": 715 }, { "epoch": 0.92, - "grad_norm": 0.21875, - "learning_rate": 1.3515789473684211e-05, - "loss": 1.9212, + "grad_norm": 0.341796875, + "learning_rate": 6.666666666666667e-05, + "loss": 1.7276, "step": 716 }, { "epoch": 0.92, - "grad_norm": 0.283203125, - "learning_rate": 1.3505263157894737e-05, - "loss": 1.9374, + "grad_norm": 0.330078125, + "learning_rate": 6.645021645021645e-05, + "loss": 1.7521, "step": 717 }, { "epoch": 0.92, - "grad_norm": 0.26171875, - "learning_rate": 1.3494736842105265e-05, - "loss": 1.9455, + "grad_norm": 0.36328125, + "learning_rate": 6.623376623376624e-05, + "loss": 1.7368, "step": 718 }, { "epoch": 0.92, - "grad_norm": 0.265625, - "learning_rate": 1.3484210526315792e-05, - "loss": 2.0084, + "grad_norm": 0.3828125, + "learning_rate": 6.601731601731602e-05, + "loss": 1.7896, "step": 719 }, { "epoch": 0.92, - "grad_norm": 0.2373046875, - "learning_rate": 1.3473684210526316e-05, - "loss": 2.0203, + "grad_norm": 0.365234375, + "learning_rate": 6.58008658008658e-05, + "loss": 1.8057, "step": 720 }, { "epoch": 0.92, - "grad_norm": 0.267578125, - "learning_rate": 1.3463157894736842e-05, - "loss": 1.9512, + "grad_norm": 0.3828125, + "learning_rate": 6.55844155844156e-05, + "loss": 1.7596, "step": 721 }, { "epoch": 0.92, - "grad_norm": 0.2890625, - "learning_rate": 1.345263157894737e-05, - "loss": 1.9631, + "grad_norm": 0.40625, + "learning_rate": 6.536796536796536e-05, + "loss": 1.7793, "step": 722 }, { "epoch": 0.93, - "grad_norm": 0.263671875, - "learning_rate": 1.3442105263157897e-05, - "loss": 1.8664, + "grad_norm": 0.365234375, + "learning_rate": 6.515151515151516e-05, + "loss": 1.6966, "step": 723 }, { "epoch": 0.93, - "grad_norm": 0.2177734375, - "learning_rate": 1.3431578947368421e-05, - "loss": 1.8441, + "grad_norm": 0.32421875, + "learning_rate": 6.493506493506494e-05, + "loss": 1.6619, "step": 724 }, { "epoch": 0.93, - "grad_norm": 0.2177734375, - "learning_rate": 1.3421052631578948e-05, - "loss": 1.9354, + "grad_norm": 0.3359375, + "learning_rate": 6.471861471861473e-05, + "loss": 1.7305, "step": 725 }, { "epoch": 0.93, - "grad_norm": 0.232421875, - "learning_rate": 1.3410526315789474e-05, - "loss": 1.9086, + "grad_norm": 0.365234375, + "learning_rate": 6.450216450216451e-05, + "loss": 1.7337, "step": 726 }, { "epoch": 0.93, - "grad_norm": 0.2490234375, - "learning_rate": 1.3400000000000002e-05, - "loss": 1.923, + "grad_norm": 0.37890625, + "learning_rate": 6.428571428571429e-05, + "loss": 1.7255, "step": 727 }, { "epoch": 0.93, - "grad_norm": 0.2333984375, - "learning_rate": 1.3389473684210528e-05, - "loss": 1.9298, + "grad_norm": 0.345703125, + "learning_rate": 6.406926406926407e-05, + "loss": 1.75, "step": 728 }, { "epoch": 0.93, - "grad_norm": 0.2216796875, - "learning_rate": 1.3378947368421053e-05, - "loss": 1.9055, + "grad_norm": 0.337890625, + "learning_rate": 6.385281385281385e-05, + "loss": 1.7166, "step": 729 }, { "epoch": 0.93, - "grad_norm": 0.2314453125, - "learning_rate": 1.336842105263158e-05, - "loss": 1.9323, + "grad_norm": 0.3203125, + "learning_rate": 6.363636363636364e-05, + "loss": 1.7407, "step": 730 }, { "epoch": 0.94, - "grad_norm": 0.359375, - "learning_rate": 1.3357894736842106e-05, - "loss": 1.8457, + "grad_norm": 0.349609375, + "learning_rate": 6.341991341991342e-05, + "loss": 1.6759, "step": 731 }, { "epoch": 0.94, - "grad_norm": 0.26953125, - "learning_rate": 1.3347368421052634e-05, - "loss": 1.9374, + "grad_norm": 0.33203125, + "learning_rate": 6.320346320346321e-05, + "loss": 1.7485, "step": 732 }, { "epoch": 0.94, - "grad_norm": 0.234375, - "learning_rate": 1.3336842105263158e-05, - "loss": 1.9759, + "grad_norm": 0.37890625, + "learning_rate": 6.2987012987013e-05, + "loss": 1.8, "step": 733 }, { "epoch": 0.94, - "grad_norm": 0.232421875, - "learning_rate": 1.3326315789473685e-05, - "loss": 1.8957, + "grad_norm": 0.404296875, + "learning_rate": 6.277056277056277e-05, + "loss": 1.6966, "step": 734 }, { "epoch": 0.94, - "grad_norm": 0.234375, - "learning_rate": 1.3315789473684211e-05, - "loss": 1.9306, + "grad_norm": 0.37890625, + "learning_rate": 6.255411255411255e-05, + "loss": 1.7524, "step": 735 }, { "epoch": 0.94, - "grad_norm": 0.216796875, - "learning_rate": 1.3305263157894739e-05, - "loss": 1.8852, + "grad_norm": 0.3515625, + "learning_rate": 6.233766233766233e-05, + "loss": 1.7087, "step": 736 }, { "epoch": 0.94, - "grad_norm": 0.263671875, - "learning_rate": 1.3294736842105265e-05, - "loss": 2.0118, + "grad_norm": 0.37109375, + "learning_rate": 6.212121212121213e-05, + "loss": 1.7963, "step": 737 }, { "epoch": 0.94, - "grad_norm": 0.2255859375, - "learning_rate": 1.328421052631579e-05, - "loss": 1.8816, + "grad_norm": 0.361328125, + "learning_rate": 6.19047619047619e-05, + "loss": 1.7043, "step": 738 }, { "epoch": 0.95, - "grad_norm": 0.337890625, - "learning_rate": 1.3273684210526316e-05, - "loss": 1.8517, + "grad_norm": 0.3046875, + "learning_rate": 6.16883116883117e-05, + "loss": 1.6797, "step": 739 }, { "epoch": 0.95, - "grad_norm": 0.2294921875, - "learning_rate": 1.3263157894736843e-05, - "loss": 1.8561, + "grad_norm": 0.396484375, + "learning_rate": 6.147186147186147e-05, + "loss": 1.6791, "step": 740 }, { "epoch": 0.95, - "grad_norm": 0.27734375, - "learning_rate": 1.325263157894737e-05, - "loss": 1.9006, + "grad_norm": 0.330078125, + "learning_rate": 6.125541125541126e-05, + "loss": 1.7069, "step": 741 }, { "epoch": 0.95, - "grad_norm": 0.2138671875, - "learning_rate": 1.3242105263157895e-05, - "loss": 1.8789, + "grad_norm": 0.33984375, + "learning_rate": 6.103896103896104e-05, + "loss": 1.6997, "step": 742 }, { "epoch": 0.95, - "grad_norm": 0.26171875, - "learning_rate": 1.3231578947368422e-05, - "loss": 1.9199, + "grad_norm": 0.376953125, + "learning_rate": 6.0822510822510825e-05, + "loss": 1.7395, "step": 743 }, { "epoch": 0.95, - "grad_norm": 0.251953125, - "learning_rate": 1.3221052631578948e-05, - "loss": 1.9281, + "grad_norm": 0.388671875, + "learning_rate": 6.060606060606061e-05, + "loss": 1.7289, "step": 744 }, { "epoch": 0.95, - "grad_norm": 0.2294921875, - "learning_rate": 1.3210526315789476e-05, - "loss": 1.9982, + "grad_norm": 0.375, + "learning_rate": 6.03896103896104e-05, + "loss": 1.7963, "step": 745 }, { "epoch": 0.95, - "grad_norm": 0.255859375, - "learning_rate": 1.3200000000000002e-05, - "loss": 2.0343, + "grad_norm": 0.318359375, + "learning_rate": 6.0173160173160184e-05, + "loss": 1.8322, "step": 746 }, { "epoch": 0.96, - "grad_norm": 0.2265625, - "learning_rate": 1.3189473684210527e-05, - "loss": 1.8725, + "grad_norm": 0.3984375, + "learning_rate": 5.995670995670996e-05, + "loss": 1.6923, "step": 747 }, { "epoch": 0.96, - "grad_norm": 0.2158203125, - "learning_rate": 1.3178947368421053e-05, - "loss": 1.9595, + "grad_norm": 0.306640625, + "learning_rate": 5.9740259740259744e-05, + "loss": 1.7534, "step": 748 }, { "epoch": 0.96, - "grad_norm": 0.2421875, - "learning_rate": 1.316842105263158e-05, - "loss": 1.9805, + "grad_norm": 0.326171875, + "learning_rate": 5.9523809523809524e-05, + "loss": 1.7843, "step": 749 }, { "epoch": 0.96, - "grad_norm": 0.25, - "learning_rate": 1.3157894736842108e-05, - "loss": 1.9848, + "grad_norm": 0.330078125, + "learning_rate": 5.930735930735931e-05, + "loss": 1.7837, "step": 750 }, { "epoch": 0.96, - "grad_norm": 0.2451171875, - "learning_rate": 1.3147368421052632e-05, - "loss": 2.028, + "eval_loss": 1.728518009185791, + "eval_runtime": 125.8224, + "eval_samples_per_second": 39.739, + "eval_steps_per_second": 1.248, + "step": 750 + }, + { + "epoch": 0.96, + "grad_norm": 0.34375, + "learning_rate": 5.90909090909091e-05, + "loss": 1.8208, "step": 751 }, { "epoch": 0.96, - "grad_norm": 0.326171875, - "learning_rate": 1.3136842105263159e-05, - "loss": 1.9331, + "grad_norm": 0.39453125, + "learning_rate": 5.887445887445888e-05, + "loss": 1.7173, "step": 752 }, { "epoch": 0.96, - "grad_norm": 0.236328125, - "learning_rate": 1.3126315789473685e-05, - "loss": 1.9034, + "grad_norm": 0.37109375, + "learning_rate": 5.8658008658008656e-05, + "loss": 1.6923, "step": 753 }, { "epoch": 0.96, - "grad_norm": 0.28125, - "learning_rate": 1.3115789473684211e-05, - "loss": 1.9612, + "grad_norm": 0.41015625, + "learning_rate": 5.844155844155844e-05, + "loss": 1.7685, "step": 754 }, { "epoch": 0.97, - "grad_norm": 0.24609375, - "learning_rate": 1.310526315789474e-05, - "loss": 1.8469, + "grad_norm": 0.3125, + "learning_rate": 5.822510822510823e-05, + "loss": 1.677, "step": 755 }, { "epoch": 0.97, - "grad_norm": 0.28125, - "learning_rate": 1.3094736842105264e-05, - "loss": 1.8542, + "grad_norm": 0.33203125, + "learning_rate": 5.800865800865801e-05, + "loss": 1.6898, "step": 756 }, { "epoch": 0.97, - "grad_norm": 0.232421875, - "learning_rate": 1.308421052631579e-05, - "loss": 1.9493, + "grad_norm": 0.3671875, + "learning_rate": 5.7792207792207796e-05, + "loss": 1.7492, "step": 757 }, { "epoch": 0.97, - "grad_norm": 0.2353515625, - "learning_rate": 1.3073684210526317e-05, - "loss": 1.9714, + "grad_norm": 0.478515625, + "learning_rate": 5.757575757575758e-05, + "loss": 1.7804, "step": 758 }, { "epoch": 0.97, - "grad_norm": 0.2197265625, - "learning_rate": 1.3063157894736845e-05, - "loss": 1.9209, + "grad_norm": 0.388671875, + "learning_rate": 5.7359307359307355e-05, + "loss": 1.7285, "step": 759 }, { "epoch": 0.97, - "grad_norm": 0.2265625, - "learning_rate": 1.305263157894737e-05, - "loss": 1.9021, + "grad_norm": 0.34765625, + "learning_rate": 5.714285714285714e-05, + "loss": 1.7054, "step": 760 }, { "epoch": 0.97, - "grad_norm": 0.236328125, - "learning_rate": 1.3042105263157896e-05, - "loss": 1.9865, + "grad_norm": 0.361328125, + "learning_rate": 5.692640692640693e-05, + "loss": 1.7949, "step": 761 }, { "epoch": 0.98, - "grad_norm": 0.271484375, - "learning_rate": 1.3031578947368422e-05, - "loss": 1.917, + "grad_norm": 0.3828125, + "learning_rate": 5.6709956709956715e-05, + "loss": 1.6984, "step": 762 }, { "epoch": 0.98, - "grad_norm": 0.2216796875, - "learning_rate": 1.3021052631578948e-05, - "loss": 1.7588, + "grad_norm": 0.3359375, + "learning_rate": 5.64935064935065e-05, + "loss": 1.5885, "step": 763 }, { "epoch": 0.98, - "grad_norm": 0.2265625, - "learning_rate": 1.3010526315789476e-05, - "loss": 2.0155, + "grad_norm": 0.373046875, + "learning_rate": 5.627705627705628e-05, + "loss": 1.8189, "step": 764 }, { "epoch": 0.98, - "grad_norm": 0.2177734375, - "learning_rate": 1.3000000000000001e-05, - "loss": 1.9234, + "grad_norm": 0.3359375, + "learning_rate": 5.606060606060606e-05, + "loss": 1.7489, "step": 765 }, { "epoch": 0.98, - "grad_norm": 0.23046875, - "learning_rate": 1.2989473684210527e-05, - "loss": 1.8854, + "grad_norm": 0.318359375, + "learning_rate": 5.584415584415584e-05, + "loss": 1.692, "step": 766 }, { "epoch": 0.98, - "grad_norm": 0.2265625, - "learning_rate": 1.2978947368421054e-05, - "loss": 1.9098, + "grad_norm": 0.337890625, + "learning_rate": 5.562770562770563e-05, + "loss": 1.7128, "step": 767 }, { "epoch": 0.98, - "grad_norm": 0.2578125, - "learning_rate": 1.2968421052631578e-05, - "loss": 1.8813, + "grad_norm": 0.369140625, + "learning_rate": 5.5411255411255414e-05, + "loss": 1.7056, "step": 768 }, { "epoch": 0.98, - "grad_norm": 0.22265625, - "learning_rate": 1.2957894736842108e-05, - "loss": 2.0128, + "grad_norm": 0.345703125, + "learning_rate": 5.51948051948052e-05, + "loss": 1.8224, "step": 769 }, { "epoch": 0.99, - "grad_norm": 0.2021484375, - "learning_rate": 1.2947368421052633e-05, - "loss": 1.9101, + "grad_norm": 0.310546875, + "learning_rate": 5.497835497835499e-05, + "loss": 1.7297, "step": 770 }, { "epoch": 0.99, - "grad_norm": 0.236328125, - "learning_rate": 1.2936842105263159e-05, - "loss": 1.9053, + "grad_norm": 0.34375, + "learning_rate": 5.4761904761904766e-05, + "loss": 1.7245, "step": 771 }, { "epoch": 0.99, - "grad_norm": 0.25390625, - "learning_rate": 1.2926315789473685e-05, - "loss": 1.9096, + "grad_norm": 0.330078125, + "learning_rate": 5.4545454545454546e-05, + "loss": 1.7286, "step": 772 }, { "epoch": 0.99, - "grad_norm": 0.2333984375, - "learning_rate": 1.2915789473684213e-05, - "loss": 1.7933, + "grad_norm": 0.357421875, + "learning_rate": 5.4329004329004326e-05, + "loss": 1.616, "step": 773 }, { "epoch": 0.99, - "grad_norm": 0.25, - "learning_rate": 1.2905263157894738e-05, - "loss": 1.9649, + "grad_norm": 0.361328125, + "learning_rate": 5.411255411255411e-05, + "loss": 1.7709, "step": 774 }, { "epoch": 0.99, - "grad_norm": 0.2490234375, - "learning_rate": 1.2894736842105264e-05, - "loss": 1.9713, + "grad_norm": 0.423828125, + "learning_rate": 5.38961038961039e-05, + "loss": 1.7746, "step": 775 }, { "epoch": 0.99, - "grad_norm": 0.2275390625, - "learning_rate": 1.288421052631579e-05, - "loss": 1.892, + "grad_norm": 0.34765625, + "learning_rate": 5.3679653679653686e-05, + "loss": 1.7077, "step": 776 }, { "epoch": 0.99, - "grad_norm": 0.1953125, - "learning_rate": 1.2873684210526317e-05, - "loss": 1.8099, + "grad_norm": 0.3515625, + "learning_rate": 5.346320346320347e-05, + "loss": 1.6281, "step": 777 }, { "epoch": 1.0, - "grad_norm": 0.2392578125, - "learning_rate": 1.2863157894736845e-05, - "loss": 1.9483, + "grad_norm": 0.361328125, + "learning_rate": 5.3246753246753245e-05, + "loss": 1.7438, "step": 778 }, { "epoch": 1.0, - "grad_norm": 0.234375, - "learning_rate": 1.285263157894737e-05, - "loss": 1.8699, + "grad_norm": 0.353515625, + "learning_rate": 5.303030303030303e-05, + "loss": 1.6973, "step": 779 }, { "epoch": 1.0, - "grad_norm": 0.26171875, - "learning_rate": 1.2842105263157896e-05, - "loss": 1.925, + "grad_norm": 0.380859375, + "learning_rate": 5.281385281385282e-05, + "loss": 1.7698, "step": 780 }, { "epoch": 1.0, - "grad_norm": 0.275390625, - "learning_rate": 1.2831578947368422e-05, - "loss": 1.9613, + "grad_norm": 0.40234375, + "learning_rate": 5.25974025974026e-05, + "loss": 1.7707, "step": 781 }, { "epoch": 1.0, - "grad_norm": 0.267578125, - "learning_rate": 1.2821052631578947e-05, - "loss": 1.9032, + "grad_norm": 0.375, + "learning_rate": 5.2380952380952384e-05, + "loss": 1.7133, "step": 782 }, { "epoch": 1.0, - "grad_norm": 0.2255859375, - "learning_rate": 1.2810526315789475e-05, - "loss": 1.9717, + "grad_norm": 0.34765625, + "learning_rate": 5.216450216450217e-05, + "loss": 1.7593, "step": 783 }, { "epoch": 1.0, - "grad_norm": 0.224609375, - "learning_rate": 1.2800000000000001e-05, - "loss": 1.9219, + "grad_norm": 0.349609375, + "learning_rate": 5.1948051948051944e-05, + "loss": 1.7286, "step": 784 }, { "epoch": 1.0, - "grad_norm": 0.2197265625, - "learning_rate": 1.2789473684210527e-05, - "loss": 1.8999, + "grad_norm": 0.33984375, + "learning_rate": 5.173160173160173e-05, + "loss": 1.724, "step": 785 }, { "epoch": 1.01, - "grad_norm": 0.2451171875, - "learning_rate": 1.2778947368421054e-05, - "loss": 1.9006, + "grad_norm": 0.3359375, + "learning_rate": 5.151515151515152e-05, + "loss": 1.6978, "step": 786 }, { "epoch": 1.01, - "grad_norm": 0.2275390625, - "learning_rate": 1.2768421052631582e-05, - "loss": 2.0137, + "grad_norm": 0.32421875, + "learning_rate": 5.1298701298701304e-05, + "loss": 1.8041, "step": 787 }, { "epoch": 1.01, - "grad_norm": 0.1943359375, - "learning_rate": 1.2757894736842106e-05, - "loss": 1.8891, + "grad_norm": 0.287109375, + "learning_rate": 5.108225108225109e-05, + "loss": 1.7141, "step": 788 }, { "epoch": 1.01, - "grad_norm": 0.224609375, - "learning_rate": 1.2747368421052633e-05, - "loss": 1.8521, + "grad_norm": 0.326171875, + "learning_rate": 5.086580086580087e-05, + "loss": 1.6585, "step": 789 }, { "epoch": 1.01, - "grad_norm": 0.228515625, - "learning_rate": 1.2736842105263159e-05, - "loss": 1.9314, + "grad_norm": 0.33203125, + "learning_rate": 5.064935064935065e-05, + "loss": 1.733, "step": 790 }, { "epoch": 1.01, - "grad_norm": 0.228515625, - "learning_rate": 1.2726315789473684e-05, - "loss": 1.9582, + "grad_norm": 0.388671875, + "learning_rate": 5.043290043290043e-05, + "loss": 1.7602, "step": 791 }, { "epoch": 1.01, - "grad_norm": 0.255859375, - "learning_rate": 1.2715789473684212e-05, - "loss": 2.0024, + "grad_norm": 0.310546875, + "learning_rate": 5.0216450216450216e-05, + "loss": 1.804, "step": 792 }, { "epoch": 1.01, - "grad_norm": 0.23046875, - "learning_rate": 1.2705263157894738e-05, - "loss": 1.873, + "grad_norm": 0.302734375, + "learning_rate": 5e-05, + "loss": 1.6816, "step": 793 }, { "epoch": 1.02, - "grad_norm": 0.2578125, - "learning_rate": 1.2694736842105264e-05, - "loss": 1.9445, + "grad_norm": 0.34375, + "learning_rate": 4.978354978354979e-05, + "loss": 1.7515, "step": 794 }, { "epoch": 1.02, - "grad_norm": 0.2236328125, - "learning_rate": 1.268421052631579e-05, - "loss": 1.8764, + "grad_norm": 0.33203125, + "learning_rate": 4.956709956709957e-05, + "loss": 1.6996, "step": 795 }, { "epoch": 1.02, - "grad_norm": 0.228515625, - "learning_rate": 1.2673684210526315e-05, - "loss": 1.9222, + "grad_norm": 0.333984375, + "learning_rate": 4.9350649350649355e-05, + "loss": 1.7045, "step": 796 }, { "epoch": 1.02, - "grad_norm": 0.26171875, - "learning_rate": 1.2663157894736843e-05, - "loss": 1.9025, + "grad_norm": 0.357421875, + "learning_rate": 4.9134199134199135e-05, + "loss": 1.7026, "step": 797 }, { "epoch": 1.02, - "grad_norm": 0.1962890625, - "learning_rate": 1.265263157894737e-05, - "loss": 1.8435, + "grad_norm": 0.30078125, + "learning_rate": 4.8917748917748915e-05, + "loss": 1.6444, "step": 798 }, { "epoch": 1.02, - "grad_norm": 0.23046875, - "learning_rate": 1.2642105263157896e-05, - "loss": 1.9704, + "grad_norm": 0.345703125, + "learning_rate": 4.87012987012987e-05, + "loss": 1.7705, "step": 799 }, { "epoch": 1.02, - "grad_norm": 0.26171875, - "learning_rate": 1.263157894736842e-05, - "loss": 1.9179, + "grad_norm": 0.35546875, + "learning_rate": 4.848484848484849e-05, + "loss": 1.7178, "step": 800 }, { "epoch": 1.02, - "eval_loss": 1.9045851230621338, - "eval_runtime": 126.0218, - "eval_samples_per_second": 39.676, - "eval_steps_per_second": 1.246, + "eval_loss": 1.7301437854766846, + "eval_runtime": 124.4902, + "eval_samples_per_second": 40.164, + "eval_steps_per_second": 1.261, "step": 800 }, { "epoch": 1.02, - "grad_norm": 0.2138671875, - "learning_rate": 1.2621052631578949e-05, - "loss": 1.941, + "grad_norm": 0.33203125, + "learning_rate": 4.826839826839827e-05, + "loss": 1.7434, "step": 801 }, { "epoch": 1.03, - "grad_norm": 0.244140625, - "learning_rate": 1.2610526315789475e-05, - "loss": 1.9228, + "grad_norm": 0.365234375, + "learning_rate": 4.8051948051948054e-05, + "loss": 1.7148, "step": 802 }, { "epoch": 1.03, - "grad_norm": 0.2421875, - "learning_rate": 1.2600000000000001e-05, - "loss": 1.9792, + "grad_norm": 0.330078125, + "learning_rate": 4.783549783549784e-05, + "loss": 1.7685, "step": 803 }, { "epoch": 1.03, - "grad_norm": 0.23828125, - "learning_rate": 1.2589473684210528e-05, - "loss": 2.0107, + "grad_norm": 0.3515625, + "learning_rate": 4.761904761904762e-05, + "loss": 1.8048, "step": 804 }, { "epoch": 1.03, - "grad_norm": 0.28125, - "learning_rate": 1.2578947368421052e-05, - "loss": 1.9363, + "grad_norm": 0.35546875, + "learning_rate": 4.740259740259741e-05, + "loss": 1.7467, "step": 805 }, { "epoch": 1.03, - "grad_norm": 0.2109375, - "learning_rate": 1.256842105263158e-05, - "loss": 1.8723, + "grad_norm": 0.33203125, + "learning_rate": 4.718614718614719e-05, + "loss": 1.6937, "step": 806 }, { "epoch": 1.03, - "grad_norm": 0.2021484375, - "learning_rate": 1.2557894736842107e-05, - "loss": 1.832, + "grad_norm": 0.306640625, + "learning_rate": 4.696969696969697e-05, + "loss": 1.6432, "step": 807 }, { "epoch": 1.03, - "grad_norm": 0.2451171875, - "learning_rate": 1.2547368421052633e-05, - "loss": 1.8576, + "grad_norm": 0.3515625, + "learning_rate": 4.675324675324675e-05, + "loss": 1.6467, "step": 808 }, { "epoch": 1.04, - "grad_norm": 0.2578125, - "learning_rate": 1.2536842105263158e-05, - "loss": 1.9772, + "grad_norm": 0.373046875, + "learning_rate": 4.653679653679654e-05, + "loss": 1.7572, "step": 809 }, { "epoch": 1.04, - "grad_norm": 0.244140625, - "learning_rate": 1.2526315789473684e-05, - "loss": 1.9748, + "grad_norm": 0.365234375, + "learning_rate": 4.6320346320346326e-05, + "loss": 1.7623, "step": 810 }, { "epoch": 1.04, - "grad_norm": 0.283203125, - "learning_rate": 1.2515789473684212e-05, - "loss": 1.9135, + "grad_norm": 0.333984375, + "learning_rate": 4.6103896103896106e-05, + "loss": 1.6898, "step": 811 }, { "epoch": 1.04, - "grad_norm": 0.265625, - "learning_rate": 1.2505263157894738e-05, - "loss": 1.8852, + "grad_norm": 0.412109375, + "learning_rate": 4.588744588744589e-05, + "loss": 1.6788, "step": 812 }, { "epoch": 1.04, - "grad_norm": 0.2138671875, - "learning_rate": 1.2494736842105265e-05, - "loss": 1.86, + "grad_norm": 0.373046875, + "learning_rate": 4.567099567099568e-05, + "loss": 1.6631, "step": 813 }, { "epoch": 1.04, - "grad_norm": 0.2138671875, - "learning_rate": 1.248421052631579e-05, - "loss": 1.841, + "grad_norm": 0.33203125, + "learning_rate": 4.545454545454546e-05, + "loss": 1.6311, "step": 814 }, { "epoch": 1.04, - "grad_norm": 0.283203125, - "learning_rate": 1.2473684210526317e-05, - "loss": 1.8751, + "grad_norm": 0.408203125, + "learning_rate": 4.523809523809524e-05, + "loss": 1.6636, "step": 815 }, { "epoch": 1.04, - "grad_norm": 0.2314453125, - "learning_rate": 1.2463157894736844e-05, - "loss": 2.0172, + "grad_norm": 0.357421875, + "learning_rate": 4.5021645021645025e-05, + "loss": 1.8099, "step": 816 }, { "epoch": 1.05, - "grad_norm": 0.234375, - "learning_rate": 1.245263157894737e-05, - "loss": 1.8799, + "grad_norm": 0.359375, + "learning_rate": 4.4805194805194805e-05, + "loss": 1.6849, "step": 817 }, { "epoch": 1.05, - "grad_norm": 0.2412109375, - "learning_rate": 1.2442105263157895e-05, - "loss": 1.8544, + "grad_norm": 0.36328125, + "learning_rate": 4.458874458874459e-05, + "loss": 1.6601, "step": 818 }, { "epoch": 1.05, - "grad_norm": 0.2333984375, - "learning_rate": 1.2431578947368421e-05, - "loss": 2.0246, + "grad_norm": 0.365234375, + "learning_rate": 4.437229437229438e-05, + "loss": 1.7999, "step": 819 }, { "epoch": 1.05, - "grad_norm": 0.2890625, - "learning_rate": 1.2421052631578949e-05, - "loss": 1.8201, + "grad_norm": 0.396484375, + "learning_rate": 4.415584415584416e-05, + "loss": 1.6323, "step": 820 }, { "epoch": 1.05, - "grad_norm": 0.240234375, - "learning_rate": 1.2410526315789475e-05, - "loss": 1.941, + "grad_norm": 0.3203125, + "learning_rate": 4.3939393939393944e-05, + "loss": 1.7141, "step": 821 }, { "epoch": 1.05, - "grad_norm": 0.23046875, - "learning_rate": 1.2400000000000002e-05, - "loss": 1.9203, + "grad_norm": 0.3359375, + "learning_rate": 4.3722943722943724e-05, + "loss": 1.7166, "step": 822 }, { "epoch": 1.05, - "grad_norm": 0.283203125, - "learning_rate": 1.2389473684210526e-05, - "loss": 1.8849, + "grad_norm": 0.421875, + "learning_rate": 4.3506493506493503e-05, + "loss": 1.6969, "step": 823 }, { "epoch": 1.05, - "grad_norm": 0.25, - "learning_rate": 1.2378947368421053e-05, - "loss": 1.872, + "grad_norm": 0.333984375, + "learning_rate": 4.329004329004329e-05, + "loss": 1.674, "step": 824 }, { "epoch": 1.06, - "grad_norm": 0.232421875, - "learning_rate": 1.236842105263158e-05, - "loss": 1.8936, + "grad_norm": 0.32421875, + "learning_rate": 4.3073593073593077e-05, + "loss": 1.6893, "step": 825 }, { "epoch": 1.06, - "grad_norm": 0.259765625, - "learning_rate": 1.2357894736842107e-05, - "loss": 1.8907, + "grad_norm": 0.36328125, + "learning_rate": 4.2857142857142856e-05, + "loss": 1.6985, "step": 826 }, { "epoch": 1.06, - "grad_norm": 0.2197265625, - "learning_rate": 1.2347368421052631e-05, - "loss": 1.9216, + "grad_norm": 0.328125, + "learning_rate": 4.264069264069264e-05, + "loss": 1.7245, "step": 827 }, { "epoch": 1.06, - "grad_norm": 0.21484375, - "learning_rate": 1.2336842105263158e-05, - "loss": 1.9218, + "grad_norm": 0.35546875, + "learning_rate": 4.242424242424243e-05, + "loss": 1.7277, "step": 828 }, { "epoch": 1.06, - "grad_norm": 0.2294921875, - "learning_rate": 1.2326315789473686e-05, - "loss": 1.8758, + "grad_norm": 0.3671875, + "learning_rate": 4.220779220779221e-05, + "loss": 1.6765, "step": 829 }, { "epoch": 1.06, - "grad_norm": 0.2158203125, - "learning_rate": 1.2315789473684212e-05, - "loss": 1.9823, + "grad_norm": 0.36328125, + "learning_rate": 4.1991341991341996e-05, + "loss": 1.7863, "step": 830 }, { "epoch": 1.06, - "grad_norm": 0.2236328125, - "learning_rate": 1.2305263157894739e-05, - "loss": 1.8659, + "grad_norm": 0.33984375, + "learning_rate": 4.1774891774891775e-05, + "loss": 1.6689, "step": 831 }, { "epoch": 1.06, - "grad_norm": 0.2421875, - "learning_rate": 1.2294736842105263e-05, - "loss": 1.9785, + "grad_norm": 0.365234375, + "learning_rate": 4.155844155844156e-05, + "loss": 1.7957, "step": 832 }, { "epoch": 1.07, - "grad_norm": 0.2138671875, - "learning_rate": 1.228421052631579e-05, - "loss": 1.829, + "grad_norm": 0.298828125, + "learning_rate": 4.134199134199134e-05, + "loss": 1.6454, "step": 833 }, { "epoch": 1.07, - "grad_norm": 0.2490234375, - "learning_rate": 1.2273684210526317e-05, - "loss": 1.859, + "grad_norm": 0.341796875, + "learning_rate": 4.112554112554113e-05, + "loss": 1.6748, "step": 834 }, { "epoch": 1.07, - "grad_norm": 0.2265625, - "learning_rate": 1.2263157894736844e-05, - "loss": 1.9899, + "grad_norm": 0.380859375, + "learning_rate": 4.0909090909090915e-05, + "loss": 1.8057, "step": 835 }, { "epoch": 1.07, - "grad_norm": 0.2431640625, - "learning_rate": 1.225263157894737e-05, - "loss": 1.9428, + "grad_norm": 0.38671875, + "learning_rate": 4.0692640692640695e-05, + "loss": 1.7343, "step": 836 }, { "epoch": 1.07, - "grad_norm": 0.240234375, - "learning_rate": 1.2242105263157895e-05, - "loss": 1.8889, + "grad_norm": 0.353515625, + "learning_rate": 4.047619047619048e-05, + "loss": 1.6868, "step": 837 }, { "epoch": 1.07, - "grad_norm": 0.232421875, - "learning_rate": 1.2231578947368421e-05, - "loss": 1.8662, + "grad_norm": 0.365234375, + "learning_rate": 4.025974025974026e-05, + "loss": 1.6663, "step": 838 }, { "epoch": 1.07, - "grad_norm": 0.2119140625, - "learning_rate": 1.2221052631578949e-05, - "loss": 1.872, + "grad_norm": 0.3359375, + "learning_rate": 4.004329004329004e-05, + "loss": 1.6779, "step": 839 }, { "epoch": 1.07, - "grad_norm": 0.228515625, - "learning_rate": 1.2210526315789475e-05, - "loss": 1.9284, + "grad_norm": 0.353515625, + "learning_rate": 3.982683982683983e-05, + "loss": 1.7324, "step": 840 }, { "epoch": 1.08, - "grad_norm": 0.265625, - "learning_rate": 1.22e-05, - "loss": 1.9453, + "grad_norm": 0.458984375, + "learning_rate": 3.9610389610389614e-05, + "loss": 1.7377, "step": 841 }, { "epoch": 1.08, - "grad_norm": 0.2265625, - "learning_rate": 1.2189473684210526e-05, - "loss": 1.9634, + "grad_norm": 0.3515625, + "learning_rate": 3.939393939393939e-05, + "loss": 1.7468, "step": 842 }, { "epoch": 1.08, - "grad_norm": 0.251953125, - "learning_rate": 1.2178947368421054e-05, - "loss": 1.9177, + "grad_norm": 0.3359375, + "learning_rate": 3.917748917748918e-05, + "loss": 1.703, "step": 843 }, { "epoch": 1.08, - "grad_norm": 0.306640625, - "learning_rate": 1.216842105263158e-05, - "loss": 1.923, + "grad_norm": 0.38671875, + "learning_rate": 3.8961038961038966e-05, + "loss": 1.7124, "step": 844 }, { "epoch": 1.08, - "grad_norm": 0.2333984375, - "learning_rate": 1.2157894736842107e-05, - "loss": 1.8434, + "grad_norm": 0.341796875, + "learning_rate": 3.8744588744588746e-05, + "loss": 1.6444, "step": 845 }, { "epoch": 1.08, - "grad_norm": 0.3359375, - "learning_rate": 1.2147368421052632e-05, - "loss": 1.96, + "grad_norm": 0.341796875, + "learning_rate": 3.852813852813853e-05, + "loss": 1.7465, "step": 846 }, { "epoch": 1.08, - "grad_norm": 0.275390625, - "learning_rate": 1.2136842105263158e-05, - "loss": 1.9316, + "grad_norm": 0.41015625, + "learning_rate": 3.831168831168831e-05, + "loss": 1.7153, "step": 847 }, { "epoch": 1.09, - "grad_norm": 0.2431640625, - "learning_rate": 1.2126315789473686e-05, - "loss": 1.949, + "grad_norm": 0.359375, + "learning_rate": 3.809523809523809e-05, + "loss": 1.756, "step": 848 }, { "epoch": 1.09, - "grad_norm": 0.224609375, - "learning_rate": 1.2115789473684212e-05, - "loss": 1.8787, + "grad_norm": 0.37890625, + "learning_rate": 3.787878787878788e-05, + "loss": 1.6992, "step": 849 }, { "epoch": 1.09, - "grad_norm": 0.2255859375, - "learning_rate": 1.2105263157894737e-05, - "loss": 1.9637, + "grad_norm": 0.33203125, + "learning_rate": 3.7662337662337665e-05, + "loss": 1.7487, "step": 850 }, { "epoch": 1.09, - "grad_norm": 0.296875, - "learning_rate": 1.2094736842105263e-05, - "loss": 1.9801, + "eval_loss": 1.725529670715332, + "eval_runtime": 124.4606, + "eval_samples_per_second": 40.173, + "eval_steps_per_second": 1.261, + "step": 850 + }, + { + "epoch": 1.09, + "grad_norm": 0.3828125, + "learning_rate": 3.7445887445887445e-05, + "loss": 1.7665, "step": 851 }, { "epoch": 1.09, - "grad_norm": 0.2275390625, - "learning_rate": 1.208421052631579e-05, - "loss": 1.9438, + "grad_norm": 0.349609375, + "learning_rate": 3.722943722943723e-05, + "loss": 1.7517, "step": 852 }, { "epoch": 1.09, - "grad_norm": 0.248046875, - "learning_rate": 1.2073684210526318e-05, - "loss": 1.9827, + "grad_norm": 0.353515625, + "learning_rate": 3.701298701298702e-05, + "loss": 1.7896, "step": 853 }, { "epoch": 1.09, - "grad_norm": 0.2392578125, - "learning_rate": 1.2063157894736844e-05, - "loss": 1.9302, + "grad_norm": 0.37109375, + "learning_rate": 3.67965367965368e-05, + "loss": 1.7554, "step": 854 }, { "epoch": 1.09, - "grad_norm": 0.2734375, - "learning_rate": 1.2052631578947369e-05, - "loss": 1.987, + "grad_norm": 0.3515625, + "learning_rate": 3.6580086580086584e-05, + "loss": 1.7757, "step": 855 }, { "epoch": 1.1, - "grad_norm": 0.271484375, - "learning_rate": 1.2042105263157895e-05, - "loss": 1.9986, + "grad_norm": 0.390625, + "learning_rate": 3.6363636363636364e-05, + "loss": 1.7977, "step": 856 }, { "epoch": 1.1, - "grad_norm": 0.228515625, - "learning_rate": 1.2031578947368423e-05, - "loss": 1.8948, + "grad_norm": 0.376953125, + "learning_rate": 3.6147186147186144e-05, + "loss": 1.7066, "step": 857 }, { "epoch": 1.1, - "grad_norm": 0.287109375, - "learning_rate": 1.202105263157895e-05, - "loss": 1.9146, + "grad_norm": 0.3671875, + "learning_rate": 3.593073593073593e-05, + "loss": 1.7159, "step": 858 }, { "epoch": 1.1, - "grad_norm": 0.263671875, - "learning_rate": 1.2010526315789474e-05, - "loss": 1.8112, + "grad_norm": 0.3515625, + "learning_rate": 3.571428571428572e-05, + "loss": 1.5972, "step": 859 }, { "epoch": 1.1, - "grad_norm": 0.2392578125, - "learning_rate": 1.2e-05, - "loss": 1.913, + "grad_norm": 0.365234375, + "learning_rate": 3.5497835497835503e-05, + "loss": 1.7102, "step": 860 }, { "epoch": 1.1, - "grad_norm": 0.275390625, - "learning_rate": 1.1989473684210527e-05, - "loss": 1.8582, + "grad_norm": 0.369140625, + "learning_rate": 3.528138528138528e-05, + "loss": 1.661, "step": 861 }, { "epoch": 1.1, - "grad_norm": 0.2490234375, - "learning_rate": 1.1978947368421055e-05, - "loss": 1.8888, + "grad_norm": 0.47265625, + "learning_rate": 3.506493506493507e-05, + "loss": 1.691, "step": 862 }, { "epoch": 1.1, - "grad_norm": 0.22265625, - "learning_rate": 1.1968421052631581e-05, - "loss": 1.9347, + "grad_norm": 0.3125, + "learning_rate": 3.484848484848485e-05, + "loss": 1.7247, "step": 863 }, { "epoch": 1.11, - "grad_norm": 0.287109375, - "learning_rate": 1.1957894736842106e-05, - "loss": 1.8314, + "grad_norm": 0.390625, + "learning_rate": 3.463203463203463e-05, + "loss": 1.6295, "step": 864 }, { "epoch": 1.11, - "grad_norm": 0.23828125, - "learning_rate": 1.1947368421052632e-05, - "loss": 1.9537, + "grad_norm": 0.361328125, + "learning_rate": 3.4415584415584416e-05, + "loss": 1.7508, "step": 865 }, { "epoch": 1.11, - "grad_norm": 0.20703125, - "learning_rate": 1.1936842105263158e-05, - "loss": 1.8119, + "grad_norm": 0.330078125, + "learning_rate": 3.41991341991342e-05, + "loss": 1.6085, "step": 866 }, { "epoch": 1.11, - "grad_norm": 0.2216796875, - "learning_rate": 1.1926315789473686e-05, - "loss": 1.9582, + "grad_norm": 0.35546875, + "learning_rate": 3.398268398268398e-05, + "loss": 1.7507, "step": 867 }, { "epoch": 1.11, - "grad_norm": 0.259765625, - "learning_rate": 1.1915789473684211e-05, - "loss": 1.9298, + "grad_norm": 0.296875, + "learning_rate": 3.376623376623377e-05, + "loss": 1.7143, "step": 868 }, { "epoch": 1.11, - "grad_norm": 0.2138671875, - "learning_rate": 1.1905263157894737e-05, - "loss": 1.8877, + "grad_norm": 0.34765625, + "learning_rate": 3.3549783549783555e-05, + "loss": 1.7195, "step": 869 }, { "epoch": 1.11, - "grad_norm": 0.2275390625, - "learning_rate": 1.1894736842105264e-05, - "loss": 1.9596, + "grad_norm": 0.3984375, + "learning_rate": 3.3333333333333335e-05, + "loss": 1.7606, "step": 870 }, { "epoch": 1.11, - "grad_norm": 0.22265625, - "learning_rate": 1.1884210526315792e-05, - "loss": 1.892, + "grad_norm": 0.3515625, + "learning_rate": 3.311688311688312e-05, + "loss": 1.7038, "step": 871 }, { "epoch": 1.12, - "grad_norm": 0.216796875, - "learning_rate": 1.1873684210526318e-05, - "loss": 1.9588, + "grad_norm": 0.33984375, + "learning_rate": 3.29004329004329e-05, + "loss": 1.7545, "step": 872 }, { "epoch": 1.12, - "grad_norm": 0.224609375, - "learning_rate": 1.1863157894736843e-05, - "loss": 1.9004, + "grad_norm": 0.306640625, + "learning_rate": 3.268398268398268e-05, + "loss": 1.6866, "step": 873 }, { "epoch": 1.12, - "grad_norm": 0.29296875, - "learning_rate": 1.1852631578947369e-05, - "loss": 1.8668, + "grad_norm": 0.404296875, + "learning_rate": 3.246753246753247e-05, + "loss": 1.683, "step": 874 }, { "epoch": 1.12, - "grad_norm": 0.2353515625, - "learning_rate": 1.1842105263157895e-05, - "loss": 1.9236, + "grad_norm": 0.345703125, + "learning_rate": 3.2251082251082254e-05, + "loss": 1.7329, "step": 875 }, { "epoch": 1.12, - "grad_norm": 0.2158203125, - "learning_rate": 1.1831578947368423e-05, - "loss": 1.9828, + "grad_norm": 0.341796875, + "learning_rate": 3.2034632034632034e-05, + "loss": 1.7675, "step": 876 }, { "epoch": 1.12, - "grad_norm": 0.2294921875, - "learning_rate": 1.1821052631578948e-05, - "loss": 1.8475, + "grad_norm": 0.333984375, + "learning_rate": 3.181818181818182e-05, + "loss": 1.6585, "step": 877 }, { "epoch": 1.12, - "grad_norm": 0.2333984375, - "learning_rate": 1.1810526315789474e-05, - "loss": 1.9532, + "grad_norm": 0.34375, + "learning_rate": 3.160173160173161e-05, + "loss": 1.7628, "step": 878 }, { "epoch": 1.12, - "grad_norm": 0.244140625, - "learning_rate": 1.18e-05, - "loss": 1.8683, + "grad_norm": 0.384765625, + "learning_rate": 3.1385281385281387e-05, + "loss": 1.6784, "step": 879 }, { "epoch": 1.13, - "grad_norm": 0.2080078125, - "learning_rate": 1.1789473684210527e-05, - "loss": 1.9361, + "grad_norm": 0.345703125, + "learning_rate": 3.1168831168831166e-05, + "loss": 1.7177, "step": 880 }, { "epoch": 1.13, - "grad_norm": 0.2236328125, - "learning_rate": 1.1778947368421055e-05, - "loss": 1.8961, + "grad_norm": 0.34375, + "learning_rate": 3.095238095238095e-05, + "loss": 1.6945, "step": 881 }, { "epoch": 1.13, - "grad_norm": 0.2275390625, - "learning_rate": 1.176842105263158e-05, - "loss": 1.906, + "grad_norm": 0.380859375, + "learning_rate": 3.073593073593073e-05, + "loss": 1.7096, "step": 882 }, { "epoch": 1.13, - "grad_norm": 0.228515625, - "learning_rate": 1.1757894736842106e-05, - "loss": 1.9503, + "grad_norm": 0.361328125, + "learning_rate": 3.051948051948052e-05, + "loss": 1.7545, "step": 883 }, { "epoch": 1.13, - "grad_norm": 0.2177734375, - "learning_rate": 1.1747368421052632e-05, - "loss": 1.8866, + "grad_norm": 0.369140625, + "learning_rate": 3.0303030303030306e-05, + "loss": 1.7122, "step": 884 }, { "epoch": 1.13, - "grad_norm": 0.2021484375, - "learning_rate": 1.173684210526316e-05, - "loss": 1.8286, + "grad_norm": 0.36328125, + "learning_rate": 3.0086580086580092e-05, + "loss": 1.6433, "step": 885 }, { "epoch": 1.13, - "grad_norm": 0.26953125, - "learning_rate": 1.1726315789473685e-05, - "loss": 1.8873, + "grad_norm": 0.37109375, + "learning_rate": 2.9870129870129872e-05, + "loss": 1.6908, "step": 886 }, { "epoch": 1.13, - "grad_norm": 0.216796875, - "learning_rate": 1.1715789473684211e-05, - "loss": 1.9954, + "grad_norm": 0.326171875, + "learning_rate": 2.9653679653679655e-05, + "loss": 1.8096, "step": 887 }, { "epoch": 1.14, - "grad_norm": 0.275390625, - "learning_rate": 1.1705263157894737e-05, - "loss": 1.7937, + "grad_norm": 0.375, + "learning_rate": 2.943722943722944e-05, + "loss": 1.5972, "step": 888 }, { "epoch": 1.14, - "grad_norm": 0.2314453125, - "learning_rate": 1.1694736842105264e-05, - "loss": 1.9872, + "grad_norm": 0.345703125, + "learning_rate": 2.922077922077922e-05, + "loss": 1.7858, "step": 889 }, { "epoch": 1.14, - "grad_norm": 0.259765625, - "learning_rate": 1.1684210526315792e-05, - "loss": 1.9047, + "grad_norm": 0.326171875, + "learning_rate": 2.9004329004329005e-05, + "loss": 1.7353, "step": 890 }, { "epoch": 1.14, - "grad_norm": 0.2109375, - "learning_rate": 1.1673684210526316e-05, - "loss": 1.9496, + "grad_norm": 0.35546875, + "learning_rate": 2.878787878787879e-05, + "loss": 1.7572, "step": 891 }, { "epoch": 1.14, - "grad_norm": 0.2255859375, - "learning_rate": 1.1663157894736843e-05, - "loss": 1.925, + "grad_norm": 0.361328125, + "learning_rate": 2.857142857142857e-05, + "loss": 1.7268, "step": 892 }, { "epoch": 1.14, - "grad_norm": 0.2412109375, - "learning_rate": 1.1652631578947369e-05, - "loss": 1.9966, + "grad_norm": 0.37109375, + "learning_rate": 2.8354978354978357e-05, + "loss": 1.7919, "step": 893 }, { "epoch": 1.14, - "grad_norm": 0.2197265625, - "learning_rate": 1.1642105263157897e-05, - "loss": 1.9153, + "grad_norm": 0.37109375, + "learning_rate": 2.813852813852814e-05, + "loss": 1.735, "step": 894 }, { "epoch": 1.15, - "grad_norm": 0.248046875, - "learning_rate": 1.1631578947368423e-05, - "loss": 1.9138, + "grad_norm": 0.32421875, + "learning_rate": 2.792207792207792e-05, + "loss": 1.7174, "step": 895 }, { "epoch": 1.15, - "grad_norm": 0.259765625, - "learning_rate": 1.1621052631578948e-05, - "loss": 1.9022, + "grad_norm": 0.37890625, + "learning_rate": 2.7705627705627707e-05, + "loss": 1.6896, "step": 896 }, { "epoch": 1.15, - "grad_norm": 0.23046875, - "learning_rate": 1.1610526315789474e-05, - "loss": 1.9798, + "grad_norm": 0.318359375, + "learning_rate": 2.7489177489177493e-05, + "loss": 1.7908, "step": 897 }, { "epoch": 1.15, - "grad_norm": 0.2177734375, - "learning_rate": 1.16e-05, - "loss": 1.9533, + "grad_norm": 0.33984375, + "learning_rate": 2.7272727272727273e-05, + "loss": 1.7684, "step": 898 }, { "epoch": 1.15, - "grad_norm": 0.2080078125, - "learning_rate": 1.1589473684210529e-05, - "loss": 1.9027, + "grad_norm": 0.33203125, + "learning_rate": 2.7056277056277056e-05, + "loss": 1.7138, "step": 899 }, { "epoch": 1.15, - "grad_norm": 0.2451171875, - "learning_rate": 1.1578947368421053e-05, - "loss": 1.9008, + "grad_norm": 0.34375, + "learning_rate": 2.6839826839826843e-05, + "loss": 1.7012, "step": 900 }, { "epoch": 1.15, - "grad_norm": 0.236328125, - "learning_rate": 1.156842105263158e-05, - "loss": 1.9567, + "eval_loss": 1.725927710533142, + "eval_runtime": 125.737, + "eval_samples_per_second": 39.766, + "eval_steps_per_second": 1.249, + "step": 900 + }, + { + "epoch": 1.15, + "grad_norm": 0.35546875, + "learning_rate": 2.6623376623376623e-05, + "loss": 1.7415, "step": 901 }, { "epoch": 1.15, - "grad_norm": 0.2373046875, - "learning_rate": 1.1557894736842106e-05, - "loss": 1.9152, + "grad_norm": 0.365234375, + "learning_rate": 2.640692640692641e-05, + "loss": 1.7195, "step": 902 }, { "epoch": 1.16, - "grad_norm": 0.2392578125, - "learning_rate": 1.1547368421052632e-05, - "loss": 1.9616, + "grad_norm": 0.337890625, + "learning_rate": 2.6190476190476192e-05, + "loss": 1.7539, "step": 903 }, { "epoch": 1.16, - "grad_norm": 0.2490234375, - "learning_rate": 1.153684210526316e-05, - "loss": 1.8135, + "grad_norm": 0.369140625, + "learning_rate": 2.5974025974025972e-05, + "loss": 1.6218, "step": 904 }, { "epoch": 1.16, - "grad_norm": 0.2177734375, - "learning_rate": 1.1526315789473685e-05, - "loss": 1.9143, + "grad_norm": 0.3125, + "learning_rate": 2.575757575757576e-05, + "loss": 1.6949, "step": 905 }, { "epoch": 1.16, - "grad_norm": 0.2470703125, - "learning_rate": 1.1515789473684211e-05, - "loss": 1.9554, + "grad_norm": 0.361328125, + "learning_rate": 2.5541125541125545e-05, + "loss": 1.7539, "step": 906 }, { "epoch": 1.16, - "grad_norm": 0.2412109375, - "learning_rate": 1.1505263157894738e-05, - "loss": 1.9495, + "grad_norm": 0.345703125, + "learning_rate": 2.5324675324675325e-05, + "loss": 1.7398, "step": 907 }, { "epoch": 1.16, - "grad_norm": 0.2734375, - "learning_rate": 1.1494736842105266e-05, - "loss": 2.0138, + "grad_norm": 0.3515625, + "learning_rate": 2.5108225108225108e-05, + "loss": 1.7924, "step": 908 }, { "epoch": 1.16, - "grad_norm": 0.23046875, - "learning_rate": 1.148421052631579e-05, - "loss": 1.9318, + "grad_norm": 0.34765625, + "learning_rate": 2.4891774891774894e-05, + "loss": 1.7182, "step": 909 }, { "epoch": 1.16, - "grad_norm": 0.27734375, - "learning_rate": 1.1473684210526317e-05, - "loss": 1.919, + "grad_norm": 0.412109375, + "learning_rate": 2.4675324675324678e-05, + "loss": 1.7197, "step": 910 }, { "epoch": 1.17, - "grad_norm": 0.2353515625, - "learning_rate": 1.1463157894736843e-05, - "loss": 1.8965, + "grad_norm": 0.310546875, + "learning_rate": 2.4458874458874457e-05, + "loss": 1.6976, "step": 911 }, { "epoch": 1.17, - "grad_norm": 0.2353515625, - "learning_rate": 1.145263157894737e-05, - "loss": 1.9294, + "grad_norm": 0.369140625, + "learning_rate": 2.4242424242424244e-05, + "loss": 1.7478, "step": 912 }, { "epoch": 1.17, - "grad_norm": 0.208984375, - "learning_rate": 1.1442105263157897e-05, - "loss": 1.8079, + "grad_norm": 0.333984375, + "learning_rate": 2.4025974025974027e-05, + "loss": 1.6187, "step": 913 }, { "epoch": 1.17, - "grad_norm": 0.24609375, - "learning_rate": 1.1431578947368422e-05, - "loss": 1.9185, + "grad_norm": 0.345703125, + "learning_rate": 2.380952380952381e-05, + "loss": 1.7191, "step": 914 }, { "epoch": 1.17, - "grad_norm": 0.291015625, - "learning_rate": 1.1421052631578948e-05, - "loss": 1.8557, + "grad_norm": 0.439453125, + "learning_rate": 2.3593073593073593e-05, + "loss": 1.6616, "step": 915 }, { "epoch": 1.17, - "grad_norm": 0.212890625, - "learning_rate": 1.1410526315789475e-05, - "loss": 1.8313, + "grad_norm": 0.328125, + "learning_rate": 2.3376623376623376e-05, + "loss": 1.6345, "step": 916 }, { "epoch": 1.17, - "grad_norm": 0.291015625, - "learning_rate": 1.14e-05, - "loss": 1.9439, + "grad_norm": 0.447265625, + "learning_rate": 2.3160173160173163e-05, + "loss": 1.7367, "step": 917 }, { "epoch": 1.17, - "grad_norm": 0.3203125, - "learning_rate": 1.1389473684210527e-05, - "loss": 1.921, + "grad_norm": 0.40625, + "learning_rate": 2.2943722943722946e-05, + "loss": 1.7257, "step": 918 }, { "epoch": 1.18, - "grad_norm": 0.259765625, - "learning_rate": 1.1378947368421054e-05, - "loss": 1.9136, + "grad_norm": 0.35546875, + "learning_rate": 2.272727272727273e-05, + "loss": 1.6999, "step": 919 }, { "epoch": 1.18, - "grad_norm": 0.310546875, - "learning_rate": 1.136842105263158e-05, - "loss": 1.8291, + "grad_norm": 0.37109375, + "learning_rate": 2.2510822510822512e-05, + "loss": 1.6337, "step": 920 }, { "epoch": 1.18, - "grad_norm": 0.251953125, - "learning_rate": 1.1357894736842106e-05, - "loss": 1.9956, + "grad_norm": 0.349609375, + "learning_rate": 2.2294372294372296e-05, + "loss": 1.7632, "step": 921 }, { "epoch": 1.18, - "grad_norm": 0.220703125, - "learning_rate": 1.1347368421052634e-05, - "loss": 1.9689, + "grad_norm": 0.341796875, + "learning_rate": 2.207792207792208e-05, + "loss": 1.7686, "step": 922 }, { "epoch": 1.18, - "grad_norm": 0.24609375, - "learning_rate": 1.1336842105263159e-05, - "loss": 1.8752, + "grad_norm": 0.3671875, + "learning_rate": 2.1861471861471862e-05, + "loss": 1.6897, "step": 923 }, { "epoch": 1.18, - "grad_norm": 0.220703125, - "learning_rate": 1.1326315789473685e-05, - "loss": 2.0572, + "grad_norm": 0.365234375, + "learning_rate": 2.1645021645021645e-05, + "loss": 1.8555, "step": 924 }, { "epoch": 1.18, - "grad_norm": 0.232421875, - "learning_rate": 1.1315789473684212e-05, - "loss": 1.9353, + "grad_norm": 0.408203125, + "learning_rate": 2.1428571428571428e-05, + "loss": 1.7255, "step": 925 }, { "epoch": 1.18, - "grad_norm": 0.2373046875, - "learning_rate": 1.1305263157894736e-05, - "loss": 1.8876, + "grad_norm": 0.3359375, + "learning_rate": 2.1212121212121215e-05, + "loss": 1.6774, "step": 926 }, { "epoch": 1.19, - "grad_norm": 0.201171875, - "learning_rate": 1.1294736842105264e-05, - "loss": 1.9436, + "grad_norm": 0.39453125, + "learning_rate": 2.0995670995670998e-05, + "loss": 1.7524, "step": 927 }, { "epoch": 1.19, - "grad_norm": 0.234375, - "learning_rate": 1.128421052631579e-05, - "loss": 1.8469, + "grad_norm": 0.328125, + "learning_rate": 2.077922077922078e-05, + "loss": 1.6569, "step": 928 }, { "epoch": 1.19, - "grad_norm": 0.27734375, - "learning_rate": 1.1273684210526317e-05, - "loss": 1.8954, + "grad_norm": 0.5078125, + "learning_rate": 2.0562770562770564e-05, + "loss": 1.7028, "step": 929 }, { "epoch": 1.19, - "grad_norm": 0.2294921875, - "learning_rate": 1.1263157894736843e-05, - "loss": 1.9251, + "grad_norm": 0.349609375, + "learning_rate": 2.0346320346320347e-05, + "loss": 1.7377, "step": 930 }, { "epoch": 1.19, - "grad_norm": 0.24609375, - "learning_rate": 1.1252631578947368e-05, - "loss": 1.9006, + "grad_norm": 0.341796875, + "learning_rate": 2.012987012987013e-05, + "loss": 1.6855, "step": 931 }, { "epoch": 1.19, - "grad_norm": 0.23828125, - "learning_rate": 1.1242105263157896e-05, - "loss": 1.8943, + "grad_norm": 0.326171875, + "learning_rate": 1.9913419913419914e-05, + "loss": 1.6917, "step": 932 }, { "epoch": 1.19, - "grad_norm": 0.2236328125, - "learning_rate": 1.1231578947368422e-05, - "loss": 1.9219, + "grad_norm": 0.341796875, + "learning_rate": 1.9696969696969697e-05, + "loss": 1.714, "step": 933 }, { "epoch": 1.2, - "grad_norm": 0.2080078125, - "learning_rate": 1.1221052631578949e-05, - "loss": 1.9485, + "grad_norm": 0.333984375, + "learning_rate": 1.9480519480519483e-05, + "loss": 1.7653, "step": 934 }, { "epoch": 1.2, - "grad_norm": 0.2294921875, - "learning_rate": 1.1210526315789473e-05, - "loss": 1.9052, + "grad_norm": 0.337890625, + "learning_rate": 1.9264069264069266e-05, + "loss": 1.6973, "step": 935 }, { "epoch": 1.2, - "grad_norm": 0.2265625, - "learning_rate": 1.1200000000000001e-05, - "loss": 1.8826, + "grad_norm": 0.34765625, + "learning_rate": 1.9047619047619046e-05, + "loss": 1.7066, "step": 936 }, { "epoch": 1.2, - "grad_norm": 0.2255859375, - "learning_rate": 1.1189473684210528e-05, - "loss": 1.889, + "grad_norm": 0.34375, + "learning_rate": 1.8831168831168833e-05, + "loss": 1.6812, "step": 937 }, { "epoch": 1.2, - "grad_norm": 0.2109375, - "learning_rate": 1.1178947368421054e-05, - "loss": 1.9161, + "grad_norm": 0.341796875, + "learning_rate": 1.8614718614718616e-05, + "loss": 1.7114, "step": 938 }, { "epoch": 1.2, - "grad_norm": 0.3515625, - "learning_rate": 1.116842105263158e-05, - "loss": 1.9225, + "grad_norm": 0.3984375, + "learning_rate": 1.83982683982684e-05, + "loss": 1.7109, "step": 939 }, { "epoch": 1.2, - "grad_norm": 0.2421875, - "learning_rate": 1.1157894736842105e-05, - "loss": 1.8309, + "grad_norm": 0.375, + "learning_rate": 1.8181818181818182e-05, + "loss": 1.6037, "step": 940 }, { "epoch": 1.2, - "grad_norm": 0.28125, - "learning_rate": 1.1147368421052633e-05, - "loss": 1.9685, + "grad_norm": 0.34765625, + "learning_rate": 1.7965367965367965e-05, + "loss": 1.7588, "step": 941 }, { "epoch": 1.21, - "grad_norm": 0.25390625, - "learning_rate": 1.1136842105263159e-05, - "loss": 1.9035, + "grad_norm": 0.412109375, + "learning_rate": 1.7748917748917752e-05, + "loss": 1.7093, "step": 942 }, { "epoch": 1.21, - "grad_norm": 0.236328125, - "learning_rate": 1.1126315789473685e-05, - "loss": 1.9429, + "grad_norm": 0.353515625, + "learning_rate": 1.7532467532467535e-05, + "loss": 1.7642, "step": 943 }, { "epoch": 1.21, - "grad_norm": 0.263671875, - "learning_rate": 1.111578947368421e-05, - "loss": 1.9281, + "grad_norm": 0.466796875, + "learning_rate": 1.7316017316017315e-05, + "loss": 1.724, "step": 944 }, { "epoch": 1.21, - "grad_norm": 0.24609375, - "learning_rate": 1.1105263157894736e-05, - "loss": 1.7952, + "grad_norm": 0.34375, + "learning_rate": 1.70995670995671e-05, + "loss": 1.6136, "step": 945 }, { "epoch": 1.21, - "grad_norm": 0.318359375, - "learning_rate": 1.1094736842105264e-05, - "loss": 1.8396, + "grad_norm": 0.431640625, + "learning_rate": 1.6883116883116884e-05, + "loss": 1.6256, "step": 946 }, { "epoch": 1.21, - "grad_norm": 0.26171875, - "learning_rate": 1.108421052631579e-05, - "loss": 1.8809, + "grad_norm": 0.400390625, + "learning_rate": 1.6666666666666667e-05, + "loss": 1.6878, "step": 947 }, { "epoch": 1.21, - "grad_norm": 0.2431640625, - "learning_rate": 1.1073684210526317e-05, - "loss": 1.9206, + "grad_norm": 0.388671875, + "learning_rate": 1.645021645021645e-05, + "loss": 1.7281, "step": 948 }, { "epoch": 1.21, - "grad_norm": 0.267578125, - "learning_rate": 1.1063157894736842e-05, - "loss": 1.9018, + "grad_norm": 0.3359375, + "learning_rate": 1.6233766233766234e-05, + "loss": 1.7021, "step": 949 }, { "epoch": 1.22, - "grad_norm": 0.25, - "learning_rate": 1.105263157894737e-05, - "loss": 2.0147, + "grad_norm": 0.375, + "learning_rate": 1.6017316017316017e-05, + "loss": 1.7982, "step": 950 }, { "epoch": 1.22, - "grad_norm": 0.302734375, - "learning_rate": 1.1042105263157896e-05, - "loss": 1.8462, + "eval_loss": 1.7258570194244385, + "eval_runtime": 124.9465, + "eval_samples_per_second": 40.017, + "eval_steps_per_second": 1.257, + "step": 950 + }, + { + "epoch": 1.22, + "grad_norm": 0.400390625, + "learning_rate": 1.5800865800865803e-05, + "loss": 1.641, "step": 951 }, { "epoch": 1.22, - "grad_norm": 0.2255859375, - "learning_rate": 1.1031578947368422e-05, - "loss": 1.8618, + "grad_norm": 0.357421875, + "learning_rate": 1.5584415584415583e-05, + "loss": 1.6773, "step": 952 }, { "epoch": 1.22, - "grad_norm": 0.2216796875, - "learning_rate": 1.1021052631578947e-05, - "loss": 1.8524, + "grad_norm": 0.341796875, + "learning_rate": 1.5367965367965366e-05, + "loss": 1.654, "step": 953 }, { "epoch": 1.22, - "grad_norm": 0.2470703125, - "learning_rate": 1.1010526315789473e-05, - "loss": 1.9275, + "grad_norm": 0.380859375, + "learning_rate": 1.5151515151515153e-05, + "loss": 1.7277, "step": 954 }, { "epoch": 1.22, - "grad_norm": 0.2255859375, - "learning_rate": 1.1000000000000001e-05, - "loss": 1.8913, + "grad_norm": 0.359375, + "learning_rate": 1.4935064935064936e-05, + "loss": 1.6995, "step": 955 }, { "epoch": 1.22, - "grad_norm": 0.23828125, - "learning_rate": 1.0989473684210528e-05, - "loss": 1.9005, + "grad_norm": 0.359375, + "learning_rate": 1.471861471861472e-05, + "loss": 1.7112, "step": 956 }, { "epoch": 1.22, - "grad_norm": 0.2197265625, - "learning_rate": 1.0978947368421054e-05, - "loss": 1.962, + "grad_norm": 0.35546875, + "learning_rate": 1.4502164502164502e-05, + "loss": 1.7486, "step": 957 }, { "epoch": 1.23, - "grad_norm": 0.2451171875, - "learning_rate": 1.0968421052631579e-05, - "loss": 1.9427, + "grad_norm": 0.40234375, + "learning_rate": 1.4285714285714285e-05, + "loss": 1.7627, "step": 958 }, { "epoch": 1.23, - "grad_norm": 0.228515625, - "learning_rate": 1.0957894736842105e-05, - "loss": 1.8673, + "grad_norm": 0.34375, + "learning_rate": 1.406926406926407e-05, + "loss": 1.6709, "step": 959 }, { "epoch": 1.23, - "grad_norm": 0.2265625, - "learning_rate": 1.0947368421052633e-05, - "loss": 1.9166, + "grad_norm": 0.345703125, + "learning_rate": 1.3852813852813853e-05, + "loss": 1.7112, "step": 960 }, { "epoch": 1.23, - "grad_norm": 0.279296875, - "learning_rate": 1.093684210526316e-05, - "loss": 1.9077, + "grad_norm": 0.375, + "learning_rate": 1.3636363636363637e-05, + "loss": 1.7299, "step": 961 }, { "epoch": 1.23, - "grad_norm": 0.244140625, - "learning_rate": 1.0926315789473686e-05, - "loss": 1.8829, + "grad_norm": 0.390625, + "learning_rate": 1.3419913419913421e-05, + "loss": 1.6656, "step": 962 }, { "epoch": 1.23, - "grad_norm": 0.296875, - "learning_rate": 1.091578947368421e-05, - "loss": 1.8466, + "grad_norm": 0.37890625, + "learning_rate": 1.3203463203463205e-05, + "loss": 1.6621, "step": 963 }, { "epoch": 1.23, - "grad_norm": 0.259765625, - "learning_rate": 1.0905263157894738e-05, - "loss": 1.8662, + "grad_norm": 0.33203125, + "learning_rate": 1.2987012987012986e-05, + "loss": 1.6751, "step": 964 }, { "epoch": 1.23, - "grad_norm": 0.298828125, - "learning_rate": 1.0894736842105265e-05, - "loss": 1.8827, + "grad_norm": 0.48828125, + "learning_rate": 1.2770562770562773e-05, + "loss": 1.7, "step": 965 }, { "epoch": 1.24, - "grad_norm": 0.2197265625, - "learning_rate": 1.0884210526315791e-05, - "loss": 1.942, + "grad_norm": 0.345703125, + "learning_rate": 1.2554112554112554e-05, + "loss": 1.7547, "step": 966 }, { "epoch": 1.24, - "grad_norm": 0.236328125, - "learning_rate": 1.0873684210526316e-05, - "loss": 1.9131, + "grad_norm": 0.357421875, + "learning_rate": 1.2337662337662339e-05, + "loss": 1.6858, "step": 967 }, { "epoch": 1.24, - "grad_norm": 0.224609375, - "learning_rate": 1.0863157894736842e-05, - "loss": 1.7553, + "grad_norm": 0.33984375, + "learning_rate": 1.2121212121212122e-05, + "loss": 1.5868, "step": 968 }, { "epoch": 1.24, - "grad_norm": 0.2255859375, - "learning_rate": 1.085263157894737e-05, - "loss": 1.9672, + "grad_norm": 0.3359375, + "learning_rate": 1.1904761904761905e-05, + "loss": 1.7628, "step": 969 }, { "epoch": 1.24, - "grad_norm": 0.2236328125, - "learning_rate": 1.0842105263157896e-05, - "loss": 1.8225, + "grad_norm": 0.314453125, + "learning_rate": 1.1688311688311688e-05, + "loss": 1.6664, "step": 970 }, { "epoch": 1.24, - "grad_norm": 0.232421875, - "learning_rate": 1.0831578947368423e-05, - "loss": 1.9378, + "grad_norm": 0.33984375, + "learning_rate": 1.1471861471861473e-05, + "loss": 1.7387, "step": 971 }, { "epoch": 1.24, - "grad_norm": 0.25, - "learning_rate": 1.0821052631578947e-05, - "loss": 1.9634, + "grad_norm": 0.392578125, + "learning_rate": 1.1255411255411256e-05, + "loss": 1.7765, "step": 972 }, { "epoch": 1.25, - "grad_norm": 0.27734375, - "learning_rate": 1.0810526315789474e-05, - "loss": 2.0081, + "grad_norm": 0.341796875, + "learning_rate": 1.103896103896104e-05, + "loss": 1.7887, "step": 973 }, { "epoch": 1.25, - "grad_norm": 0.232421875, - "learning_rate": 1.0800000000000002e-05, - "loss": 1.9105, + "grad_norm": 0.341796875, + "learning_rate": 1.0822510822510823e-05, + "loss": 1.7143, "step": 974 }, { "epoch": 1.25, - "grad_norm": 0.2255859375, - "learning_rate": 1.0789473684210528e-05, - "loss": 1.8352, + "grad_norm": 0.33984375, + "learning_rate": 1.0606060606060607e-05, + "loss": 1.6573, "step": 975 }, { "epoch": 1.25, - "grad_norm": 0.23046875, - "learning_rate": 1.0778947368421053e-05, - "loss": 1.91, + "grad_norm": 0.33203125, + "learning_rate": 1.038961038961039e-05, + "loss": 1.7191, "step": 976 }, { "epoch": 1.25, - "grad_norm": 0.244140625, - "learning_rate": 1.0768421052631579e-05, - "loss": 1.9209, + "grad_norm": 0.337890625, + "learning_rate": 1.0173160173160174e-05, + "loss": 1.7437, "step": 977 }, { "epoch": 1.25, - "grad_norm": 0.291015625, - "learning_rate": 1.0757894736842107e-05, - "loss": 2.0162, + "grad_norm": 0.36328125, + "learning_rate": 9.956709956709957e-06, + "loss": 1.8012, "step": 978 }, { "epoch": 1.25, - "grad_norm": 0.2451171875, - "learning_rate": 1.0747368421052633e-05, - "loss": 1.8842, + "grad_norm": 0.3515625, + "learning_rate": 9.740259740259742e-06, + "loss": 1.6769, "step": 979 }, { "epoch": 1.25, - "grad_norm": 0.2392578125, - "learning_rate": 1.073684210526316e-05, - "loss": 1.978, + "grad_norm": 0.326171875, + "learning_rate": 9.523809523809523e-06, + "loss": 1.7836, "step": 980 }, { "epoch": 1.26, - "grad_norm": 0.2294921875, - "learning_rate": 1.0726315789473684e-05, - "loss": 1.8531, + "grad_norm": 0.375, + "learning_rate": 9.307359307359308e-06, + "loss": 1.6654, "step": 981 }, { "epoch": 1.26, - "grad_norm": 0.255859375, - "learning_rate": 1.071578947368421e-05, - "loss": 1.9201, + "grad_norm": 0.37109375, + "learning_rate": 9.090909090909091e-06, + "loss": 1.7367, "step": 982 }, { "epoch": 1.26, - "grad_norm": 0.2265625, - "learning_rate": 1.0705263157894739e-05, - "loss": 1.8885, + "grad_norm": 0.353515625, + "learning_rate": 8.874458874458876e-06, + "loss": 1.6735, "step": 983 }, { "epoch": 1.26, - "grad_norm": 0.244140625, - "learning_rate": 1.0694736842105265e-05, - "loss": 1.7738, + "grad_norm": 0.32421875, + "learning_rate": 8.658008658008657e-06, + "loss": 1.5795, "step": 984 }, { "epoch": 1.26, - "grad_norm": 0.2265625, - "learning_rate": 1.068421052631579e-05, - "loss": 1.87, + "grad_norm": 0.333984375, + "learning_rate": 8.441558441558442e-06, + "loss": 1.6784, "step": 985 }, { "epoch": 1.26, - "grad_norm": 0.23046875, - "learning_rate": 1.0673684210526316e-05, - "loss": 1.8681, + "grad_norm": 0.33203125, + "learning_rate": 8.225108225108225e-06, + "loss": 1.6743, "step": 986 }, { "epoch": 1.26, - "grad_norm": 0.2255859375, - "learning_rate": 1.0663157894736842e-05, - "loss": 1.8929, + "grad_norm": 0.3671875, + "learning_rate": 8.008658008658008e-06, + "loss": 1.6877, "step": 987 }, { "epoch": 1.26, - "grad_norm": 0.20703125, - "learning_rate": 1.065263157894737e-05, - "loss": 1.9815, + "grad_norm": 0.3125, + "learning_rate": 7.792207792207792e-06, + "loss": 1.7913, "step": 988 }, { "epoch": 1.27, - "grad_norm": 0.267578125, - "learning_rate": 1.0642105263157897e-05, - "loss": 1.8243, + "grad_norm": 0.38671875, + "learning_rate": 7.5757575757575764e-06, + "loss": 1.6147, "step": 989 }, { "epoch": 1.27, - "grad_norm": 0.22265625, - "learning_rate": 1.0631578947368421e-05, - "loss": 1.9526, + "grad_norm": 0.33203125, + "learning_rate": 7.35930735930736e-06, + "loss": 1.7615, "step": 990 }, { "epoch": 1.27, - "grad_norm": 0.2275390625, - "learning_rate": 1.0621052631578948e-05, - "loss": 1.9205, + "grad_norm": 0.3515625, + "learning_rate": 7.142857142857143e-06, + "loss": 1.7406, "step": 991 }, { "epoch": 1.27, - "grad_norm": 0.314453125, - "learning_rate": 1.0610526315789476e-05, - "loss": 1.84, + "grad_norm": 0.427734375, + "learning_rate": 6.926406926406927e-06, + "loss": 1.6287, "step": 992 }, { "epoch": 1.27, - "grad_norm": 0.25390625, - "learning_rate": 1.0600000000000002e-05, - "loss": 1.9487, + "grad_norm": 0.345703125, + "learning_rate": 6.709956709956711e-06, + "loss": 1.7655, "step": 993 }, { "epoch": 1.27, - "grad_norm": 0.2392578125, - "learning_rate": 1.0589473684210526e-05, - "loss": 1.8785, + "grad_norm": 0.330078125, + "learning_rate": 6.493506493506493e-06, + "loss": 1.7003, "step": 994 }, { "epoch": 1.27, - "grad_norm": 0.2333984375, - "learning_rate": 1.0578947368421053e-05, - "loss": 1.9572, + "grad_norm": 0.345703125, + "learning_rate": 6.277056277056277e-06, + "loss": 1.7531, "step": 995 }, { "epoch": 1.27, - "grad_norm": 0.244140625, - "learning_rate": 1.0568421052631579e-05, - "loss": 1.9612, + "grad_norm": 0.365234375, + "learning_rate": 6.060606060606061e-06, + "loss": 1.7475, "step": 996 }, { "epoch": 1.28, - "grad_norm": 0.2177734375, - "learning_rate": 1.0557894736842107e-05, - "loss": 1.8746, + "grad_norm": 0.361328125, + "learning_rate": 5.844155844155844e-06, + "loss": 1.6811, "step": 997 }, { "epoch": 1.28, - "grad_norm": 0.2216796875, - "learning_rate": 1.0547368421052633e-05, - "loss": 1.9251, + "grad_norm": 0.3203125, + "learning_rate": 5.627705627705628e-06, + "loss": 1.7256, "step": 998 }, { "epoch": 1.28, - "grad_norm": 0.2890625, - "learning_rate": 1.0536842105263158e-05, - "loss": 1.923, + "grad_norm": 0.349609375, + "learning_rate": 5.411255411255411e-06, + "loss": 1.7238, "step": 999 }, { "epoch": 1.28, - "grad_norm": 0.234375, - "learning_rate": 1.0526315789473684e-05, - "loss": 1.8895, + "grad_norm": 0.34375, + "learning_rate": 5.194805194805195e-06, + "loss": 1.6972, "step": 1000 }, { "epoch": 1.28, - "eval_loss": 1.8982402086257935, - "eval_runtime": 125.5691, - "eval_samples_per_second": 39.819, + "eval_loss": 1.7247449159622192, + "eval_runtime": 125.5655, + "eval_samples_per_second": 39.82, "eval_steps_per_second": 1.25, "step": 1000 }, { "epoch": 1.28, - "grad_norm": 0.26171875, - "learning_rate": 1.051578947368421e-05, - "loss": 1.9124, + "grad_norm": 0.359375, + "learning_rate": 4.978354978354978e-06, + "loss": 1.6889, "step": 1001 }, { "epoch": 1.28, - "grad_norm": 0.2392578125, - "learning_rate": 1.0505263157894739e-05, - "loss": 1.9535, + "grad_norm": 0.330078125, + "learning_rate": 4.7619047619047615e-06, + "loss": 1.7495, "step": 1002 }, { "epoch": 1.28, - "grad_norm": 0.259765625, - "learning_rate": 1.0494736842105263e-05, - "loss": 1.8472, + "grad_norm": 0.37890625, + "learning_rate": 4.5454545454545455e-06, + "loss": 1.6499, "step": 1003 }, { "epoch": 1.28, - "grad_norm": 0.26171875, - "learning_rate": 1.048421052631579e-05, - "loss": 1.8814, + "grad_norm": 0.353515625, + "learning_rate": 4.329004329004329e-06, + "loss": 1.6646, "step": 1004 }, { "epoch": 1.29, - "grad_norm": 0.23046875, - "learning_rate": 1.0473684210526316e-05, - "loss": 1.9454, + "grad_norm": 0.3671875, + "learning_rate": 4.112554112554113e-06, + "loss": 1.7523, "step": 1005 }, { "epoch": 1.29, - "grad_norm": 0.23828125, - "learning_rate": 1.0463157894736844e-05, - "loss": 1.9222, + "grad_norm": 0.333984375, + "learning_rate": 3.896103896103896e-06, + "loss": 1.7232, "step": 1006 }, { "epoch": 1.29, - "grad_norm": 0.298828125, - "learning_rate": 1.045263157894737e-05, - "loss": 1.9051, + "grad_norm": 0.400390625, + "learning_rate": 3.67965367965368e-06, + "loss": 1.7209, "step": 1007 }, { "epoch": 1.29, - "grad_norm": 0.21875, - "learning_rate": 1.0442105263157895e-05, - "loss": 1.8414, + "grad_norm": 0.333984375, + "learning_rate": 3.4632034632034634e-06, + "loss": 1.6627, "step": 1008 }, { "epoch": 1.29, - "grad_norm": 0.2158203125, - "learning_rate": 1.0431578947368421e-05, - "loss": 1.8454, + "grad_norm": 0.3203125, + "learning_rate": 3.2467532467532465e-06, + "loss": 1.6678, "step": 1009 }, { "epoch": 1.29, - "grad_norm": 0.2314453125, - "learning_rate": 1.0421052631578948e-05, - "loss": 1.856, + "grad_norm": 0.333984375, + "learning_rate": 3.0303030303030305e-06, + "loss": 1.6757, "step": 1010 }, { "epoch": 1.29, - "grad_norm": 0.244140625, - "learning_rate": 1.0410526315789476e-05, - "loss": 1.9136, + "grad_norm": 0.33203125, + "learning_rate": 2.813852813852814e-06, + "loss": 1.7232, "step": 1011 }, { "epoch": 1.29, - "grad_norm": 0.22265625, - "learning_rate": 1.04e-05, - "loss": 1.8602, + "grad_norm": 0.33203125, + "learning_rate": 2.5974025974025976e-06, + "loss": 1.6797, "step": 1012 }, { "epoch": 1.3, - "grad_norm": 0.2275390625, - "learning_rate": 1.0389473684210527e-05, - "loss": 2.015, + "grad_norm": 0.337890625, + "learning_rate": 2.3809523809523808e-06, + "loss": 1.825, "step": 1013 }, { "epoch": 1.3, - "grad_norm": 0.216796875, - "learning_rate": 1.0378947368421053e-05, - "loss": 1.883, + "grad_norm": 0.337890625, + "learning_rate": 2.1645021645021643e-06, + "loss": 1.6841, "step": 1014 }, { "epoch": 1.3, - "grad_norm": 0.21875, - "learning_rate": 1.036842105263158e-05, - "loss": 1.9039, + "grad_norm": 0.328125, + "learning_rate": 1.948051948051948e-06, + "loss": 1.726, "step": 1015 }, { "epoch": 1.3, - "grad_norm": 0.322265625, - "learning_rate": 1.0357894736842107e-05, - "loss": 2.0001, + "grad_norm": 0.392578125, + "learning_rate": 1.7316017316017317e-06, + "loss": 1.7702, "step": 1016 }, { "epoch": 1.3, - "grad_norm": 0.2470703125, - "learning_rate": 1.0347368421052632e-05, - "loss": 1.9709, + "grad_norm": 0.3359375, + "learning_rate": 1.5151515151515152e-06, + "loss": 1.7689, "step": 1017 }, { "epoch": 1.3, - "grad_norm": 0.236328125, - "learning_rate": 1.0336842105263158e-05, - "loss": 1.8884, + "grad_norm": 0.37109375, + "learning_rate": 1.2987012987012988e-06, + "loss": 1.7003, "step": 1018 }, { "epoch": 1.3, - "grad_norm": 0.265625, - "learning_rate": 1.0326315789473685e-05, - "loss": 1.8962, + "grad_norm": 0.40234375, + "learning_rate": 1.0822510822510822e-06, + "loss": 1.6841, "step": 1019 }, { "epoch": 1.31, - "grad_norm": 0.287109375, - "learning_rate": 1.0315789473684213e-05, - "loss": 1.9054, + "grad_norm": 0.380859375, + "learning_rate": 8.658008658008658e-07, + "loss": 1.7318, "step": 1020 }, { "epoch": 1.31, - "grad_norm": 0.31640625, - "learning_rate": 1.0305263157894739e-05, - "loss": 1.9776, + "grad_norm": 0.345703125, + "learning_rate": 6.493506493506494e-07, + "loss": 1.7868, "step": 1021 }, { "epoch": 1.31, - "grad_norm": 0.2197265625, - "learning_rate": 1.0294736842105264e-05, - "loss": 1.8947, + "grad_norm": 0.34765625, + "learning_rate": 4.329004329004329e-07, + "loss": 1.7199, "step": 1022 }, { "epoch": 1.31, - "grad_norm": 0.2373046875, - "learning_rate": 1.028421052631579e-05, - "loss": 1.9671, + "grad_norm": 0.359375, + "learning_rate": 2.1645021645021646e-07, + "loss": 1.7633, "step": 1023 }, { "epoch": 1.31, - "grad_norm": 0.2060546875, - "learning_rate": 1.0273684210526316e-05, - "loss": 1.9824, + "grad_norm": 0.328125, + "learning_rate": 0.0, + "loss": 1.7878, "step": 1024 }, { "epoch": 1.31, - "grad_norm": 0.2236328125, - "learning_rate": 1.0263157894736844e-05, - "loss": 1.9456, - "step": 1025 - }, - { - "epoch": 1.31, - "grad_norm": 0.203125, - "learning_rate": 1.0252631578947369e-05, - "loss": 1.923, - "step": 1026 + "step": 1024, + "total_flos": 7.46638513614422e+17, + "train_loss": 1.8673716291086748, + "train_runtime": 7714.204, + "train_samples_per_second": 8.495, + "train_steps_per_second": 0.133 }, { "epoch": 1.31, - "grad_norm": 0.2490234375, - "learning_rate": 1.0242105263157895e-05, - "loss": 1.9494, - "step": 1027 - }, - { - "epoch": 1.32, - "grad_norm": 0.2294921875, - "learning_rate": 1.0231578947368422e-05, - "loss": 2.008, - "step": 1028 - }, - { - "epoch": 1.32, - "grad_norm": 0.26171875, - "learning_rate": 1.0221052631578948e-05, - "loss": 1.9639, - "step": 1029 - }, - { - "epoch": 1.32, - "grad_norm": 0.2236328125, - "learning_rate": 1.0210526315789476e-05, - "loss": 1.8448, - "step": 1030 - }, - { - "epoch": 1.32, - "grad_norm": 0.23828125, - "learning_rate": 1.02e-05, - "loss": 1.9404, - "step": 1031 - }, - { - "epoch": 1.32, - "grad_norm": 0.26953125, - "learning_rate": 1.0189473684210527e-05, - "loss": 1.9008, - "step": 1032 - }, - { - "epoch": 1.32, - "grad_norm": 0.25390625, - "learning_rate": 1.0178947368421053e-05, - "loss": 1.9309, - "step": 1033 - }, - { - "epoch": 1.32, - "grad_norm": 0.2451171875, - "learning_rate": 1.0168421052631581e-05, - "loss": 1.8554, - "step": 1034 - }, - { - "epoch": 1.32, - "grad_norm": 0.20703125, - "learning_rate": 1.0157894736842106e-05, - "loss": 1.9283, - "step": 1035 - }, - { - "epoch": 1.33, - "grad_norm": 0.24609375, - "learning_rate": 1.0147368421052632e-05, - "loss": 1.9251, - "step": 1036 - }, - { - "epoch": 1.33, - "grad_norm": 0.2236328125, - "learning_rate": 1.0136842105263159e-05, - "loss": 1.8915, - "step": 1037 - }, - { - "epoch": 1.33, - "grad_norm": 0.2275390625, - "learning_rate": 1.0126315789473685e-05, - "loss": 1.842, - "step": 1038 - }, - { - "epoch": 1.33, - "grad_norm": 0.240234375, - "learning_rate": 1.0115789473684213e-05, - "loss": 1.9776, - "step": 1039 - }, - { - "epoch": 1.33, - "grad_norm": 0.275390625, - "learning_rate": 1.0105263157894738e-05, - "loss": 1.9869, - "step": 1040 - }, - { - "epoch": 1.33, - "grad_norm": 0.279296875, - "learning_rate": 1.0094736842105264e-05, - "loss": 1.8346, - "step": 1041 - }, - { - "epoch": 1.33, - "grad_norm": 0.2490234375, - "learning_rate": 1.008421052631579e-05, - "loss": 1.8519, - "step": 1042 - }, - { - "epoch": 1.33, - "grad_norm": 0.3125, - "learning_rate": 1.0073684210526315e-05, - "loss": 1.8381, - "step": 1043 - }, - { - "epoch": 1.34, - "grad_norm": 0.220703125, - "learning_rate": 1.0063157894736843e-05, - "loss": 1.9232, - "step": 1044 - }, - { - "epoch": 1.34, - "grad_norm": 0.2294921875, - "learning_rate": 1.005263157894737e-05, - "loss": 1.8917, - "step": 1045 - }, - { - "epoch": 1.34, - "grad_norm": 0.2216796875, - "learning_rate": 1.0042105263157896e-05, - "loss": 1.8798, - "step": 1046 - }, - { - "epoch": 1.34, - "grad_norm": 0.271484375, - "learning_rate": 1.0031578947368422e-05, - "loss": 1.9875, - "step": 1047 - }, - { - "epoch": 1.34, - "grad_norm": 0.23046875, - "learning_rate": 1.002105263157895e-05, - "loss": 1.8191, - "step": 1048 - }, - { - "epoch": 1.34, - "grad_norm": 0.25390625, - "learning_rate": 1.0010526315789474e-05, - "loss": 1.8813, - "step": 1049 - }, - { - "epoch": 1.34, - "grad_norm": 0.2578125, - "learning_rate": 1e-05, - "loss": 1.8996, - "step": 1050 - }, - { - "epoch": 1.34, - "grad_norm": 0.232421875, - "learning_rate": 9.989473684210527e-06, - "loss": 1.9899, - "step": 1051 - }, - { - "epoch": 1.35, - "grad_norm": 0.25, - "learning_rate": 9.978947368421053e-06, - "loss": 1.9122, - "step": 1052 - }, - { - "epoch": 1.35, - "grad_norm": 0.2216796875, - "learning_rate": 9.96842105263158e-06, - "loss": 1.8423, - "step": 1053 - }, - { - "epoch": 1.35, - "grad_norm": 0.2451171875, - "learning_rate": 9.957894736842106e-06, - "loss": 1.9176, - "step": 1054 - }, - { - "epoch": 1.35, - "grad_norm": 0.2177734375, - "learning_rate": 9.947368421052632e-06, - "loss": 1.9154, - "step": 1055 - }, - { - "epoch": 1.35, - "grad_norm": 0.279296875, - "learning_rate": 9.936842105263159e-06, - "loss": 1.9441, - "step": 1056 - }, - { - "epoch": 1.35, - "grad_norm": 0.212890625, - "learning_rate": 9.926315789473685e-06, - "loss": 1.9348, - "step": 1057 - }, - { - "epoch": 1.35, - "grad_norm": 0.25, - "learning_rate": 9.915789473684211e-06, - "loss": 1.8764, - "step": 1058 - }, - { - "epoch": 1.36, - "grad_norm": 0.27734375, - "learning_rate": 9.905263157894738e-06, - "loss": 2.0069, - "step": 1059 - }, - { - "epoch": 1.36, - "grad_norm": 0.2158203125, - "learning_rate": 9.894736842105264e-06, - "loss": 1.8697, - "step": 1060 - }, - { - "epoch": 1.36, - "grad_norm": 0.263671875, - "learning_rate": 9.88421052631579e-06, - "loss": 1.883, - "step": 1061 - }, - { - "epoch": 1.36, - "grad_norm": 0.3046875, - "learning_rate": 9.873684210526317e-06, - "loss": 1.9454, - "step": 1062 - }, - { - "epoch": 1.36, - "grad_norm": 0.341796875, - "learning_rate": 9.863157894736843e-06, - "loss": 1.9243, - "step": 1063 - }, - { - "epoch": 1.36, - "grad_norm": 0.2431640625, - "learning_rate": 9.85263157894737e-06, - "loss": 1.9216, - "step": 1064 - }, - { - "epoch": 1.36, - "grad_norm": 0.2138671875, - "learning_rate": 9.842105263157896e-06, - "loss": 1.9205, - "step": 1065 - }, - { - "epoch": 1.36, - "grad_norm": 0.251953125, - "learning_rate": 9.831578947368422e-06, - "loss": 1.8346, - "step": 1066 - }, - { - "epoch": 1.37, - "grad_norm": 0.267578125, - "learning_rate": 9.821052631578948e-06, - "loss": 1.964, - "step": 1067 - }, - { - "epoch": 1.37, - "grad_norm": 0.2490234375, - "learning_rate": 9.810526315789475e-06, - "loss": 1.8823, - "step": 1068 - }, - { - "epoch": 1.37, - "grad_norm": 0.25, - "learning_rate": 9.800000000000001e-06, - "loss": 1.9385, - "step": 1069 - }, - { - "epoch": 1.37, - "grad_norm": 0.2314453125, - "learning_rate": 9.789473684210527e-06, - "loss": 1.9, - "step": 1070 - }, - { - "epoch": 1.37, - "grad_norm": 0.2314453125, - "learning_rate": 9.778947368421054e-06, - "loss": 1.8596, - "step": 1071 - }, - { - "epoch": 1.37, - "grad_norm": 0.2236328125, - "learning_rate": 9.76842105263158e-06, - "loss": 1.9197, - "step": 1072 - }, - { - "epoch": 1.37, - "grad_norm": 0.2373046875, - "learning_rate": 9.757894736842106e-06, - "loss": 1.8574, - "step": 1073 - }, - { - "epoch": 1.37, - "grad_norm": 0.22265625, - "learning_rate": 9.747368421052633e-06, - "loss": 1.8386, - "step": 1074 - }, - { - "epoch": 1.38, - "grad_norm": 0.267578125, - "learning_rate": 9.736842105263159e-06, - "loss": 1.93, - "step": 1075 - }, - { - "epoch": 1.38, - "grad_norm": 0.244140625, - "learning_rate": 9.726315789473685e-06, - "loss": 1.9072, - "step": 1076 - }, - { - "epoch": 1.38, - "grad_norm": 0.248046875, - "learning_rate": 9.715789473684212e-06, - "loss": 1.9346, - "step": 1077 - }, - { - "epoch": 1.38, - "grad_norm": 0.2333984375, - "learning_rate": 9.705263157894738e-06, - "loss": 1.9914, - "step": 1078 - }, - { - "epoch": 1.38, - "grad_norm": 0.248046875, - "learning_rate": 9.694736842105263e-06, - "loss": 1.9472, - "step": 1079 - }, - { - "epoch": 1.38, - "grad_norm": 0.2158203125, - "learning_rate": 9.68421052631579e-06, - "loss": 1.8756, - "step": 1080 - }, - { - "epoch": 1.38, - "grad_norm": 0.228515625, - "learning_rate": 9.673684210526317e-06, - "loss": 1.8667, - "step": 1081 - }, - { - "epoch": 1.38, - "grad_norm": 0.2373046875, - "learning_rate": 9.663157894736843e-06, - "loss": 1.9025, - "step": 1082 - }, - { - "epoch": 1.39, - "grad_norm": 0.25390625, - "learning_rate": 9.65263157894737e-06, - "loss": 1.8888, - "step": 1083 - }, - { - "epoch": 1.39, - "grad_norm": 0.3125, - "learning_rate": 9.642105263157896e-06, - "loss": 1.8776, - "step": 1084 - }, - { - "epoch": 1.39, - "grad_norm": 0.197265625, - "learning_rate": 9.631578947368422e-06, - "loss": 1.8577, - "step": 1085 - }, - { - "epoch": 1.39, - "grad_norm": 0.21875, - "learning_rate": 9.621052631578947e-06, - "loss": 1.9356, - "step": 1086 - }, - { - "epoch": 1.39, - "grad_norm": 0.27734375, - "learning_rate": 9.610526315789475e-06, - "loss": 2.0179, - "step": 1087 - }, - { - "epoch": 1.39, - "grad_norm": 0.27734375, - "learning_rate": 9.600000000000001e-06, - "loss": 1.9079, - "step": 1088 - }, - { - "epoch": 1.39, - "grad_norm": 0.244140625, - "learning_rate": 9.589473684210528e-06, - "loss": 1.9096, - "step": 1089 - }, - { - "epoch": 1.39, - "grad_norm": 0.22265625, - "learning_rate": 9.578947368421054e-06, - "loss": 1.9515, - "step": 1090 - }, - { - "epoch": 1.4, - "grad_norm": 0.3046875, - "learning_rate": 9.56842105263158e-06, - "loss": 1.985, - "step": 1091 - }, - { - "epoch": 1.4, - "grad_norm": 0.220703125, - "learning_rate": 9.557894736842107e-06, - "loss": 1.9059, - "step": 1092 - }, - { - "epoch": 1.4, - "grad_norm": 0.220703125, - "learning_rate": 9.547368421052631e-06, - "loss": 1.8877, - "step": 1093 - }, - { - "epoch": 1.4, - "grad_norm": 0.2109375, - "learning_rate": 9.53684210526316e-06, - "loss": 1.8944, - "step": 1094 - }, - { - "epoch": 1.4, - "grad_norm": 0.29296875, - "learning_rate": 9.526315789473684e-06, - "loss": 1.899, - "step": 1095 - }, - { - "epoch": 1.4, - "grad_norm": 0.2421875, - "learning_rate": 9.515789473684212e-06, - "loss": 1.9102, - "step": 1096 - }, - { - "epoch": 1.4, - "grad_norm": 0.259765625, - "learning_rate": 9.505263157894738e-06, - "loss": 1.8217, - "step": 1097 - }, - { - "epoch": 1.4, - "grad_norm": 0.2275390625, - "learning_rate": 9.494736842105265e-06, - "loss": 1.9465, - "step": 1098 - }, - { - "epoch": 1.41, - "grad_norm": 0.298828125, - "learning_rate": 9.484210526315791e-06, - "loss": 1.8662, - "step": 1099 - }, - { - "epoch": 1.41, - "grad_norm": 0.2470703125, - "learning_rate": 9.473684210526315e-06, - "loss": 2.0122, - "step": 1100 - }, - { - "epoch": 1.41, - "grad_norm": 0.2431640625, - "learning_rate": 9.463157894736844e-06, - "loss": 1.8356, - "step": 1101 - }, - { - "epoch": 1.41, - "grad_norm": 0.2109375, - "learning_rate": 9.452631578947368e-06, - "loss": 1.83, - "step": 1102 - }, - { - "epoch": 1.41, - "grad_norm": 0.251953125, - "learning_rate": 9.442105263157896e-06, - "loss": 1.9262, - "step": 1103 - }, - { - "epoch": 1.41, - "grad_norm": 0.2294921875, - "learning_rate": 9.43157894736842e-06, - "loss": 1.9248, - "step": 1104 - }, - { - "epoch": 1.41, - "grad_norm": 0.2353515625, - "learning_rate": 9.421052631578949e-06, - "loss": 1.9209, - "step": 1105 - }, - { - "epoch": 1.42, - "grad_norm": 0.302734375, - "learning_rate": 9.410526315789475e-06, - "loss": 1.9139, - "step": 1106 - }, - { - "epoch": 1.42, - "grad_norm": 0.2177734375, - "learning_rate": 9.4e-06, - "loss": 1.9184, - "step": 1107 - }, - { - "epoch": 1.42, - "grad_norm": 0.291015625, - "learning_rate": 9.389473684210528e-06, - "loss": 1.9164, - "step": 1108 - }, - { - "epoch": 1.42, - "grad_norm": 0.248046875, - "learning_rate": 9.378947368421052e-06, - "loss": 1.9127, - "step": 1109 - }, - { - "epoch": 1.42, - "grad_norm": 0.2265625, - "learning_rate": 9.36842105263158e-06, - "loss": 1.9426, - "step": 1110 - }, - { - "epoch": 1.42, - "grad_norm": 0.263671875, - "learning_rate": 9.357894736842105e-06, - "loss": 1.9142, - "step": 1111 - }, - { - "epoch": 1.42, - "grad_norm": 0.2412109375, - "learning_rate": 9.347368421052633e-06, - "loss": 2.0506, - "step": 1112 - }, - { - "epoch": 1.42, - "grad_norm": 0.2255859375, - "learning_rate": 9.336842105263158e-06, - "loss": 1.9869, - "step": 1113 - }, - { - "epoch": 1.43, - "grad_norm": 0.21875, - "learning_rate": 9.326315789473684e-06, - "loss": 1.8889, - "step": 1114 - }, - { - "epoch": 1.43, - "grad_norm": 0.2333984375, - "learning_rate": 9.315789473684212e-06, - "loss": 1.9789, - "step": 1115 - }, - { - "epoch": 1.43, - "grad_norm": 0.2333984375, - "learning_rate": 9.305263157894737e-06, - "loss": 1.9307, - "step": 1116 - }, - { - "epoch": 1.43, - "grad_norm": 0.240234375, - "learning_rate": 9.294736842105265e-06, - "loss": 1.9016, - "step": 1117 - }, - { - "epoch": 1.43, - "grad_norm": 0.21484375, - "learning_rate": 9.28421052631579e-06, - "loss": 1.8972, - "step": 1118 - }, - { - "epoch": 1.43, - "grad_norm": 0.2431640625, - "learning_rate": 9.273684210526317e-06, - "loss": 1.9231, - "step": 1119 - }, - { - "epoch": 1.43, - "grad_norm": 0.240234375, - "learning_rate": 9.263157894736842e-06, - "loss": 1.9178, - "step": 1120 - }, - { - "epoch": 1.43, - "grad_norm": 0.2216796875, - "learning_rate": 9.252631578947368e-06, - "loss": 1.8257, - "step": 1121 - }, - { - "epoch": 1.44, - "grad_norm": 0.21875, - "learning_rate": 9.242105263157896e-06, - "loss": 1.7957, - "step": 1122 - }, - { - "epoch": 1.44, - "grad_norm": 0.2109375, - "learning_rate": 9.231578947368421e-06, - "loss": 1.9333, - "step": 1123 - }, - { - "epoch": 1.44, - "grad_norm": 0.24609375, - "learning_rate": 9.221052631578949e-06, - "loss": 1.877, - "step": 1124 - }, - { - "epoch": 1.44, - "grad_norm": 0.2177734375, - "learning_rate": 9.210526315789474e-06, - "loss": 1.8739, - "step": 1125 - }, - { - "epoch": 1.44, - "grad_norm": 0.2275390625, - "learning_rate": 9.200000000000002e-06, - "loss": 1.9471, - "step": 1126 - }, - { - "epoch": 1.44, - "grad_norm": 0.228515625, - "learning_rate": 9.189473684210526e-06, - "loss": 1.8775, - "step": 1127 - }, - { - "epoch": 1.44, - "grad_norm": 0.2353515625, - "learning_rate": 9.178947368421053e-06, - "loss": 1.9019, - "step": 1128 - }, - { - "epoch": 1.44, - "grad_norm": 0.21875, - "learning_rate": 9.168421052631579e-06, - "loss": 1.9452, - "step": 1129 - }, - { - "epoch": 1.45, - "grad_norm": 0.314453125, - "learning_rate": 9.157894736842105e-06, - "loss": 1.8572, - "step": 1130 - }, - { - "epoch": 1.45, - "grad_norm": 0.265625, - "learning_rate": 9.147368421052633e-06, - "loss": 1.8832, - "step": 1131 - }, - { - "epoch": 1.45, - "grad_norm": 0.279296875, - "learning_rate": 9.136842105263158e-06, - "loss": 1.9411, - "step": 1132 - }, - { - "epoch": 1.45, - "grad_norm": 0.2177734375, - "learning_rate": 9.126315789473686e-06, - "loss": 1.8916, - "step": 1133 - }, - { - "epoch": 1.45, - "grad_norm": 0.20703125, - "learning_rate": 9.11578947368421e-06, - "loss": 1.9433, - "step": 1134 - }, - { - "epoch": 1.45, - "grad_norm": 0.236328125, - "learning_rate": 9.105263157894739e-06, - "loss": 1.8442, - "step": 1135 - }, - { - "epoch": 1.45, - "grad_norm": 0.240234375, - "learning_rate": 9.094736842105263e-06, - "loss": 1.9152, - "step": 1136 - }, - { - "epoch": 1.45, - "grad_norm": 0.259765625, - "learning_rate": 9.08421052631579e-06, - "loss": 1.83, - "step": 1137 - }, - { - "epoch": 1.46, - "grad_norm": 0.287109375, - "learning_rate": 9.073684210526316e-06, - "loss": 1.8937, - "step": 1138 - }, - { - "epoch": 1.46, - "grad_norm": 0.236328125, - "learning_rate": 9.063157894736842e-06, - "loss": 1.9221, - "step": 1139 - }, - { - "epoch": 1.46, - "grad_norm": 0.232421875, - "learning_rate": 9.05263157894737e-06, - "loss": 1.9185, - "step": 1140 - }, - { - "epoch": 1.46, - "grad_norm": 0.248046875, - "learning_rate": 9.042105263157895e-06, - "loss": 1.8642, - "step": 1141 - }, - { - "epoch": 1.46, - "grad_norm": 0.267578125, - "learning_rate": 9.031578947368423e-06, - "loss": 1.8963, - "step": 1142 - }, - { - "epoch": 1.46, - "grad_norm": 0.2734375, - "learning_rate": 9.021052631578948e-06, - "loss": 1.8007, - "step": 1143 - }, - { - "epoch": 1.46, - "grad_norm": 0.2421875, - "learning_rate": 9.010526315789474e-06, - "loss": 1.921, - "step": 1144 - }, - { - "epoch": 1.47, - "grad_norm": 0.263671875, - "learning_rate": 9e-06, - "loss": 1.9043, - "step": 1145 - }, - { - "epoch": 1.47, - "grad_norm": 0.2431640625, - "learning_rate": 8.989473684210527e-06, - "loss": 1.8945, - "step": 1146 - }, - { - "epoch": 1.47, - "grad_norm": 0.21875, - "learning_rate": 8.978947368421055e-06, - "loss": 1.8491, - "step": 1147 - }, - { - "epoch": 1.47, - "grad_norm": 0.216796875, - "learning_rate": 8.96842105263158e-06, - "loss": 1.998, - "step": 1148 - }, - { - "epoch": 1.47, - "grad_norm": 0.251953125, - "learning_rate": 8.957894736842107e-06, - "loss": 1.8651, - "step": 1149 - }, - { - "epoch": 1.47, - "grad_norm": 0.265625, - "learning_rate": 8.947368421052632e-06, - "loss": 1.9223, - "step": 1150 - }, - { - "epoch": 1.47, - "grad_norm": 0.234375, - "learning_rate": 8.936842105263158e-06, - "loss": 1.8898, - "step": 1151 - }, - { - "epoch": 1.47, - "grad_norm": 0.291015625, - "learning_rate": 8.926315789473685e-06, - "loss": 1.8814, - "step": 1152 - }, - { - "epoch": 1.48, - "grad_norm": 0.2275390625, - "learning_rate": 8.915789473684211e-06, - "loss": 1.8262, - "step": 1153 - }, - { - "epoch": 1.48, - "grad_norm": 0.2265625, - "learning_rate": 8.905263157894737e-06, - "loss": 1.8841, - "step": 1154 - }, - { - "epoch": 1.48, - "grad_norm": 0.216796875, - "learning_rate": 8.894736842105264e-06, - "loss": 1.9063, - "step": 1155 - }, - { - "epoch": 1.48, - "grad_norm": 0.236328125, - "learning_rate": 8.884210526315792e-06, - "loss": 1.9586, - "step": 1156 - }, - { - "epoch": 1.48, - "grad_norm": 0.2158203125, - "learning_rate": 8.873684210526316e-06, - "loss": 1.8472, - "step": 1157 - }, - { - "epoch": 1.48, - "grad_norm": 0.248046875, - "learning_rate": 8.863157894736842e-06, - "loss": 1.8196, - "step": 1158 - }, - { - "epoch": 1.48, - "grad_norm": 0.2421875, - "learning_rate": 8.852631578947369e-06, - "loss": 1.9705, - "step": 1159 - }, - { - "epoch": 1.48, - "grad_norm": 0.23046875, - "learning_rate": 8.842105263157895e-06, - "loss": 1.8738, - "step": 1160 - }, - { - "epoch": 1.49, - "grad_norm": 0.224609375, - "learning_rate": 8.831578947368421e-06, - "loss": 1.8666, - "step": 1161 - }, - { - "epoch": 1.49, - "grad_norm": 0.330078125, - "learning_rate": 8.821052631578948e-06, - "loss": 1.8591, - "step": 1162 - }, - { - "epoch": 1.49, - "grad_norm": 0.2333984375, - "learning_rate": 8.810526315789474e-06, - "loss": 1.8924, - "step": 1163 - }, - { - "epoch": 1.49, - "grad_norm": 0.30859375, - "learning_rate": 8.8e-06, - "loss": 1.8854, - "step": 1164 - }, - { - "epoch": 1.49, - "grad_norm": 0.2392578125, - "learning_rate": 8.789473684210527e-06, - "loss": 1.9526, - "step": 1165 - }, - { - "epoch": 1.49, - "grad_norm": 0.267578125, - "learning_rate": 8.778947368421053e-06, - "loss": 1.9703, - "step": 1166 - }, - { - "epoch": 1.49, - "grad_norm": 0.2236328125, - "learning_rate": 8.76842105263158e-06, - "loss": 1.9315, - "step": 1167 - }, - { - "epoch": 1.49, - "grad_norm": 0.2421875, - "learning_rate": 8.757894736842106e-06, - "loss": 1.9262, - "step": 1168 - }, - { - "epoch": 1.5, - "grad_norm": 0.271484375, - "learning_rate": 8.747368421052632e-06, - "loss": 1.9432, - "step": 1169 - }, - { - "epoch": 1.5, - "grad_norm": 0.2373046875, - "learning_rate": 8.736842105263158e-06, - "loss": 1.7965, - "step": 1170 - }, - { - "epoch": 1.5, - "grad_norm": 0.2431640625, - "learning_rate": 8.726315789473685e-06, - "loss": 1.921, - "step": 1171 - }, - { - "epoch": 1.5, - "grad_norm": 0.2333984375, - "learning_rate": 8.715789473684211e-06, - "loss": 1.9527, - "step": 1172 - }, - { - "epoch": 1.5, - "grad_norm": 0.287109375, - "learning_rate": 8.705263157894737e-06, - "loss": 2.0361, - "step": 1173 - }, - { - "epoch": 1.5, - "grad_norm": 0.220703125, - "learning_rate": 8.694736842105264e-06, - "loss": 1.9555, - "step": 1174 - }, - { - "epoch": 1.5, - "grad_norm": 0.283203125, - "learning_rate": 8.68421052631579e-06, - "loss": 1.9218, - "step": 1175 - }, - { - "epoch": 1.5, - "grad_norm": 0.21875, - "learning_rate": 8.673684210526316e-06, - "loss": 1.9396, - "step": 1176 - }, - { - "epoch": 1.51, - "grad_norm": 0.2255859375, - "learning_rate": 8.663157894736843e-06, - "loss": 1.8431, - "step": 1177 - }, - { - "epoch": 1.51, - "grad_norm": 0.2353515625, - "learning_rate": 8.652631578947369e-06, - "loss": 1.8895, - "step": 1178 - }, - { - "epoch": 1.51, - "grad_norm": 0.244140625, - "learning_rate": 8.642105263157895e-06, - "loss": 1.8098, - "step": 1179 - }, - { - "epoch": 1.51, - "grad_norm": 0.306640625, - "learning_rate": 8.631578947368422e-06, - "loss": 1.8705, - "step": 1180 - }, - { - "epoch": 1.51, - "grad_norm": 0.25390625, - "learning_rate": 8.621052631578948e-06, - "loss": 1.8234, - "step": 1181 - }, - { - "epoch": 1.51, - "grad_norm": 0.314453125, - "learning_rate": 8.610526315789474e-06, - "loss": 1.8878, - "step": 1182 - }, - { - "epoch": 1.51, - "grad_norm": 0.216796875, - "learning_rate": 8.6e-06, - "loss": 1.8896, - "step": 1183 - }, - { - "epoch": 1.52, - "grad_norm": 0.2265625, - "learning_rate": 8.589473684210527e-06, - "loss": 1.8976, - "step": 1184 - }, - { - "epoch": 1.52, - "grad_norm": 0.341796875, - "learning_rate": 8.578947368421053e-06, - "loss": 1.9351, - "step": 1185 - }, - { - "epoch": 1.52, - "grad_norm": 0.25390625, - "learning_rate": 8.56842105263158e-06, - "loss": 1.9329, - "step": 1186 - }, - { - "epoch": 1.52, - "grad_norm": 0.228515625, - "learning_rate": 8.557894736842106e-06, - "loss": 1.8807, - "step": 1187 - }, - { - "epoch": 1.52, - "grad_norm": 0.2294921875, - "learning_rate": 8.547368421052632e-06, - "loss": 1.9033, - "step": 1188 - }, - { - "epoch": 1.52, - "grad_norm": 0.2275390625, - "learning_rate": 8.536842105263159e-06, - "loss": 1.9005, - "step": 1189 - }, - { - "epoch": 1.52, - "grad_norm": 0.291015625, - "learning_rate": 8.526315789473685e-06, - "loss": 1.9445, - "step": 1190 - }, - { - "epoch": 1.52, - "grad_norm": 0.240234375, - "learning_rate": 8.515789473684211e-06, - "loss": 1.9367, - "step": 1191 - }, - { - "epoch": 1.53, - "grad_norm": 0.279296875, - "learning_rate": 8.505263157894738e-06, - "loss": 1.9401, - "step": 1192 - }, - { - "epoch": 1.53, - "grad_norm": 0.294921875, - "learning_rate": 8.494736842105264e-06, - "loss": 1.8584, - "step": 1193 - }, - { - "epoch": 1.53, - "grad_norm": 0.244140625, - "learning_rate": 8.48421052631579e-06, - "loss": 1.939, - "step": 1194 - }, - { - "epoch": 1.53, - "grad_norm": 0.21875, - "learning_rate": 8.473684210526317e-06, - "loss": 1.7782, - "step": 1195 - }, - { - "epoch": 1.53, - "grad_norm": 0.265625, - "learning_rate": 8.463157894736843e-06, - "loss": 1.8567, - "step": 1196 - }, - { - "epoch": 1.53, - "grad_norm": 0.2294921875, - "learning_rate": 8.45263157894737e-06, - "loss": 1.9567, - "step": 1197 - }, - { - "epoch": 1.53, - "grad_norm": 0.291015625, - "learning_rate": 8.442105263157896e-06, - "loss": 1.9318, - "step": 1198 - }, - { - "epoch": 1.53, - "grad_norm": 0.248046875, - "learning_rate": 8.431578947368422e-06, - "loss": 1.8268, - "step": 1199 - }, - { - "epoch": 1.54, - "grad_norm": 0.220703125, - "learning_rate": 8.421052631578948e-06, - "loss": 1.8818, - "step": 1200 - }, - { - "epoch": 1.54, - "eval_loss": 1.8961261510849, - "eval_runtime": 125.6412, - "eval_samples_per_second": 39.796, - "eval_steps_per_second": 1.25, - "step": 1200 - }, - { - "epoch": 1.54, - "grad_norm": 0.2255859375, - "learning_rate": 8.410526315789475e-06, - "loss": 1.9245, - "step": 1201 - }, - { - "epoch": 1.54, - "grad_norm": 0.2353515625, - "learning_rate": 8.400000000000001e-06, - "loss": 1.9879, - "step": 1202 - }, - { - "epoch": 1.54, - "grad_norm": 0.2236328125, - "learning_rate": 8.389473684210527e-06, - "loss": 1.8884, - "step": 1203 - }, - { - "epoch": 1.54, - "grad_norm": 0.24609375, - "learning_rate": 8.378947368421054e-06, - "loss": 1.8985, - "step": 1204 - }, - { - "epoch": 1.54, - "grad_norm": 0.2275390625, - "learning_rate": 8.36842105263158e-06, - "loss": 1.9992, - "step": 1205 - }, - { - "epoch": 1.54, - "grad_norm": 0.2099609375, - "learning_rate": 8.357894736842106e-06, - "loss": 1.8394, - "step": 1206 - }, - { - "epoch": 1.54, - "grad_norm": 0.2099609375, - "learning_rate": 8.347368421052633e-06, - "loss": 1.9532, - "step": 1207 - }, - { - "epoch": 1.55, - "grad_norm": 0.2451171875, - "learning_rate": 8.336842105263159e-06, - "loss": 1.9159, - "step": 1208 - }, - { - "epoch": 1.55, - "grad_norm": 0.2138671875, - "learning_rate": 8.326315789473685e-06, - "loss": 1.9269, - "step": 1209 - }, - { - "epoch": 1.55, - "grad_norm": 0.2099609375, - "learning_rate": 8.315789473684212e-06, - "loss": 1.8584, - "step": 1210 - }, - { - "epoch": 1.55, - "grad_norm": 0.27734375, - "learning_rate": 8.305263157894738e-06, - "loss": 2.0144, - "step": 1211 - }, - { - "epoch": 1.55, - "grad_norm": 0.2451171875, - "learning_rate": 8.294736842105264e-06, - "loss": 1.9491, - "step": 1212 - }, - { - "epoch": 1.55, - "grad_norm": 0.2578125, - "learning_rate": 8.28421052631579e-06, - "loss": 1.9526, - "step": 1213 - }, - { - "epoch": 1.55, - "grad_norm": 0.267578125, - "learning_rate": 8.273684210526317e-06, - "loss": 1.928, - "step": 1214 - }, - { - "epoch": 1.55, - "grad_norm": 0.333984375, - "learning_rate": 8.263157894736843e-06, - "loss": 1.9027, - "step": 1215 - }, - { - "epoch": 1.56, - "grad_norm": 0.2255859375, - "learning_rate": 8.25263157894737e-06, - "loss": 1.8995, - "step": 1216 - }, - { - "epoch": 1.56, - "grad_norm": 0.25, - "learning_rate": 8.242105263157896e-06, - "loss": 1.8516, - "step": 1217 - }, - { - "epoch": 1.56, - "grad_norm": 0.216796875, - "learning_rate": 8.231578947368422e-06, - "loss": 1.8992, - "step": 1218 - }, - { - "epoch": 1.56, - "grad_norm": 0.2265625, - "learning_rate": 8.221052631578948e-06, - "loss": 1.9762, - "step": 1219 - }, - { - "epoch": 1.56, - "grad_norm": 0.251953125, - "learning_rate": 8.210526315789475e-06, - "loss": 1.8863, - "step": 1220 - }, - { - "epoch": 1.56, - "grad_norm": 0.21484375, - "learning_rate": 8.2e-06, - "loss": 1.9601, - "step": 1221 - }, - { - "epoch": 1.56, - "grad_norm": 0.21484375, - "learning_rate": 8.189473684210527e-06, - "loss": 1.8831, - "step": 1222 - }, - { - "epoch": 1.56, - "grad_norm": 0.263671875, - "learning_rate": 8.178947368421054e-06, - "loss": 1.937, - "step": 1223 - }, - { - "epoch": 1.57, - "grad_norm": 0.240234375, - "learning_rate": 8.16842105263158e-06, - "loss": 1.871, - "step": 1224 - }, - { - "epoch": 1.57, - "grad_norm": 0.22265625, - "learning_rate": 8.157894736842106e-06, - "loss": 1.8558, - "step": 1225 - }, - { - "epoch": 1.57, - "grad_norm": 0.2333984375, - "learning_rate": 8.147368421052633e-06, - "loss": 1.8865, - "step": 1226 - }, - { - "epoch": 1.57, - "grad_norm": 0.2158203125, - "learning_rate": 8.136842105263159e-06, - "loss": 1.8499, - "step": 1227 - }, - { - "epoch": 1.57, - "grad_norm": 0.26953125, - "learning_rate": 8.126315789473684e-06, - "loss": 1.8911, - "step": 1228 - }, - { - "epoch": 1.57, - "grad_norm": 0.40625, - "learning_rate": 8.115789473684212e-06, - "loss": 1.8465, - "step": 1229 - }, - { - "epoch": 1.57, - "grad_norm": 0.2490234375, - "learning_rate": 8.105263157894736e-06, - "loss": 1.9695, - "step": 1230 - }, - { - "epoch": 1.58, - "grad_norm": 0.220703125, - "learning_rate": 8.094736842105264e-06, - "loss": 1.9287, - "step": 1231 - }, - { - "epoch": 1.58, - "grad_norm": 0.322265625, - "learning_rate": 8.08421052631579e-06, - "loss": 1.8354, - "step": 1232 - }, - { - "epoch": 1.58, - "grad_norm": 0.2080078125, - "learning_rate": 8.073684210526317e-06, - "loss": 1.8258, - "step": 1233 - }, - { - "epoch": 1.58, - "grad_norm": 0.232421875, - "learning_rate": 8.063157894736843e-06, - "loss": 1.8711, - "step": 1234 - }, - { - "epoch": 1.58, - "grad_norm": 0.236328125, - "learning_rate": 8.052631578947368e-06, - "loss": 1.9256, - "step": 1235 - }, - { - "epoch": 1.58, - "grad_norm": 0.2451171875, - "learning_rate": 8.042105263157896e-06, - "loss": 1.856, - "step": 1236 - }, - { - "epoch": 1.58, - "grad_norm": 0.322265625, - "learning_rate": 8.03157894736842e-06, - "loss": 1.9228, - "step": 1237 - }, - { - "epoch": 1.58, - "grad_norm": 0.25390625, - "learning_rate": 8.021052631578949e-06, - "loss": 1.8469, - "step": 1238 - }, - { - "epoch": 1.59, - "grad_norm": 0.2490234375, - "learning_rate": 8.010526315789473e-06, - "loss": 2.0113, - "step": 1239 - }, - { - "epoch": 1.59, - "grad_norm": 0.255859375, - "learning_rate": 8.000000000000001e-06, - "loss": 1.9373, - "step": 1240 - }, - { - "epoch": 1.59, - "grad_norm": 0.2451171875, - "learning_rate": 7.989473684210528e-06, - "loss": 1.9944, - "step": 1241 - }, - { - "epoch": 1.59, - "grad_norm": 0.2734375, - "learning_rate": 7.978947368421052e-06, - "loss": 1.9254, - "step": 1242 - }, - { - "epoch": 1.59, - "grad_norm": 0.2314453125, - "learning_rate": 7.96842105263158e-06, - "loss": 1.8991, - "step": 1243 - }, - { - "epoch": 1.59, - "grad_norm": 0.275390625, - "learning_rate": 7.957894736842105e-06, - "loss": 1.8851, - "step": 1244 - }, - { - "epoch": 1.59, - "grad_norm": 0.2451171875, - "learning_rate": 7.947368421052633e-06, - "loss": 1.8784, - "step": 1245 - }, - { - "epoch": 1.59, - "grad_norm": 0.240234375, - "learning_rate": 7.936842105263158e-06, - "loss": 1.8263, - "step": 1246 - }, - { - "epoch": 1.6, - "grad_norm": 0.265625, - "learning_rate": 7.926315789473686e-06, - "loss": 1.9526, - "step": 1247 - }, - { - "epoch": 1.6, - "grad_norm": 0.251953125, - "learning_rate": 7.915789473684212e-06, - "loss": 1.8443, - "step": 1248 - }, - { - "epoch": 1.6, - "grad_norm": 0.228515625, - "learning_rate": 7.905263157894737e-06, - "loss": 1.9485, - "step": 1249 - }, - { - "epoch": 1.6, - "grad_norm": 0.2392578125, - "learning_rate": 7.894736842105265e-06, - "loss": 1.9466, - "step": 1250 - }, - { - "epoch": 1.6, - "grad_norm": 0.25390625, - "learning_rate": 7.88421052631579e-06, - "loss": 1.9032, - "step": 1251 - }, - { - "epoch": 1.6, - "grad_norm": 0.2353515625, - "learning_rate": 7.873684210526317e-06, - "loss": 1.8934, - "step": 1252 - }, - { - "epoch": 1.6, - "grad_norm": 0.23046875, - "learning_rate": 7.863157894736842e-06, - "loss": 1.8177, - "step": 1253 - }, - { - "epoch": 1.6, - "grad_norm": 0.2236328125, - "learning_rate": 7.85263157894737e-06, - "loss": 1.8544, - "step": 1254 - }, - { - "epoch": 1.61, - "grad_norm": 0.267578125, - "learning_rate": 7.842105263157895e-06, - "loss": 1.9238, - "step": 1255 - }, - { - "epoch": 1.61, - "grad_norm": 0.212890625, - "learning_rate": 7.831578947368421e-06, - "loss": 1.9009, - "step": 1256 - }, - { - "epoch": 1.61, - "grad_norm": 0.302734375, - "learning_rate": 7.821052631578949e-06, - "loss": 1.9066, - "step": 1257 - }, - { - "epoch": 1.61, - "grad_norm": 0.291015625, - "learning_rate": 7.810526315789474e-06, - "loss": 1.9204, - "step": 1258 - }, - { - "epoch": 1.61, - "grad_norm": 0.271484375, - "learning_rate": 7.800000000000002e-06, - "loss": 1.9595, - "step": 1259 - }, - { - "epoch": 1.61, - "grad_norm": 0.251953125, - "learning_rate": 7.789473684210526e-06, - "loss": 1.8198, - "step": 1260 - }, - { - "epoch": 1.61, - "grad_norm": 0.2431640625, - "learning_rate": 7.778947368421054e-06, - "loss": 1.9372, - "step": 1261 - }, - { - "epoch": 1.61, - "grad_norm": 0.2294921875, - "learning_rate": 7.768421052631579e-06, - "loss": 1.9665, - "step": 1262 - }, - { - "epoch": 1.62, - "grad_norm": 0.24609375, - "learning_rate": 7.757894736842105e-06, - "loss": 1.9578, - "step": 1263 - }, - { - "epoch": 1.62, - "grad_norm": 0.2470703125, - "learning_rate": 7.747368421052631e-06, - "loss": 1.876, - "step": 1264 - }, - { - "epoch": 1.62, - "grad_norm": 0.248046875, - "learning_rate": 7.736842105263158e-06, - "loss": 1.9408, - "step": 1265 - }, - { - "epoch": 1.62, - "grad_norm": 0.259765625, - "learning_rate": 7.726315789473686e-06, - "loss": 1.9078, - "step": 1266 - }, - { - "epoch": 1.62, - "grad_norm": 0.205078125, - "learning_rate": 7.71578947368421e-06, - "loss": 1.8501, - "step": 1267 - }, - { - "epoch": 1.62, - "grad_norm": 0.20703125, - "learning_rate": 7.705263157894738e-06, - "loss": 1.9317, - "step": 1268 - }, - { - "epoch": 1.62, - "grad_norm": 0.2265625, - "learning_rate": 7.694736842105263e-06, - "loss": 1.8919, - "step": 1269 - }, - { - "epoch": 1.63, - "grad_norm": 0.26953125, - "learning_rate": 7.68421052631579e-06, - "loss": 1.8957, - "step": 1270 - }, - { - "epoch": 1.63, - "grad_norm": 0.2373046875, - "learning_rate": 7.673684210526316e-06, - "loss": 1.8003, - "step": 1271 - }, - { - "epoch": 1.63, - "grad_norm": 0.2578125, - "learning_rate": 7.663157894736842e-06, - "loss": 1.8929, - "step": 1272 - }, - { - "epoch": 1.63, - "grad_norm": 0.24609375, - "learning_rate": 7.65263157894737e-06, - "loss": 1.9176, - "step": 1273 - }, - { - "epoch": 1.63, - "grad_norm": 0.2275390625, - "learning_rate": 7.642105263157895e-06, - "loss": 1.9481, - "step": 1274 - }, - { - "epoch": 1.63, - "grad_norm": 0.2333984375, - "learning_rate": 7.631578947368423e-06, - "loss": 1.9182, - "step": 1275 - }, - { - "epoch": 1.63, - "grad_norm": 0.302734375, - "learning_rate": 7.621052631578948e-06, - "loss": 1.833, - "step": 1276 - }, - { - "epoch": 1.63, - "grad_norm": 0.25390625, - "learning_rate": 7.610526315789474e-06, - "loss": 1.9053, - "step": 1277 - }, - { - "epoch": 1.64, - "grad_norm": 0.2451171875, - "learning_rate": 7.600000000000001e-06, - "loss": 1.9602, - "step": 1278 - }, - { - "epoch": 1.64, - "grad_norm": 0.208984375, - "learning_rate": 7.589473684210526e-06, - "loss": 1.8983, - "step": 1279 - }, - { - "epoch": 1.64, - "grad_norm": 0.2236328125, - "learning_rate": 7.578947368421054e-06, - "loss": 1.9267, - "step": 1280 - }, - { - "epoch": 1.64, - "grad_norm": 0.2216796875, - "learning_rate": 7.568421052631579e-06, - "loss": 1.8895, - "step": 1281 - }, - { - "epoch": 1.64, - "grad_norm": 0.279296875, - "learning_rate": 7.557894736842106e-06, - "loss": 1.9893, - "step": 1282 - }, - { - "epoch": 1.64, - "grad_norm": 0.2392578125, - "learning_rate": 7.547368421052632e-06, - "loss": 1.8945, - "step": 1283 - }, - { - "epoch": 1.64, - "grad_norm": 0.23828125, - "learning_rate": 7.536842105263158e-06, - "loss": 1.913, - "step": 1284 - }, - { - "epoch": 1.64, - "grad_norm": 0.2412109375, - "learning_rate": 7.526315789473685e-06, - "loss": 1.9063, - "step": 1285 - }, - { - "epoch": 1.65, - "grad_norm": 0.2578125, - "learning_rate": 7.515789473684211e-06, - "loss": 1.8506, - "step": 1286 - }, - { - "epoch": 1.65, - "grad_norm": 0.271484375, - "learning_rate": 7.505263157894738e-06, - "loss": 1.9556, - "step": 1287 - }, - { - "epoch": 1.65, - "grad_norm": 0.220703125, - "learning_rate": 7.494736842105263e-06, - "loss": 1.8959, - "step": 1288 - }, - { - "epoch": 1.65, - "grad_norm": 0.2578125, - "learning_rate": 7.4842105263157905e-06, - "loss": 1.9555, - "step": 1289 - }, - { - "epoch": 1.65, - "grad_norm": 0.2236328125, - "learning_rate": 7.473684210526316e-06, - "loss": 1.9396, - "step": 1290 - }, - { - "epoch": 1.65, - "grad_norm": 0.2294921875, - "learning_rate": 7.463157894736843e-06, - "loss": 1.911, - "step": 1291 - }, - { - "epoch": 1.65, - "grad_norm": 0.2138671875, - "learning_rate": 7.4526315789473695e-06, - "loss": 1.9081, - "step": 1292 - }, - { - "epoch": 1.65, - "grad_norm": 0.25390625, - "learning_rate": 7.442105263157895e-06, - "loss": 1.9168, - "step": 1293 - }, - { - "epoch": 1.66, - "grad_norm": 0.2158203125, - "learning_rate": 7.431578947368422e-06, - "loss": 1.8913, - "step": 1294 - }, - { - "epoch": 1.66, - "grad_norm": 0.2392578125, - "learning_rate": 7.421052631578948e-06, - "loss": 1.9043, - "step": 1295 - }, - { - "epoch": 1.66, - "grad_norm": 0.240234375, - "learning_rate": 7.410526315789475e-06, - "loss": 1.9416, - "step": 1296 - }, - { - "epoch": 1.66, - "grad_norm": 0.267578125, - "learning_rate": 7.4e-06, - "loss": 1.8398, - "step": 1297 - }, - { - "epoch": 1.66, - "grad_norm": 0.255859375, - "learning_rate": 7.3894736842105275e-06, - "loss": 1.8704, - "step": 1298 - }, - { - "epoch": 1.66, - "grad_norm": 0.2265625, - "learning_rate": 7.378947368421053e-06, - "loss": 1.8735, - "step": 1299 - }, - { - "epoch": 1.66, - "grad_norm": 0.2734375, - "learning_rate": 7.368421052631579e-06, - "loss": 1.9462, - "step": 1300 - }, - { - "epoch": 1.66, - "grad_norm": 0.291015625, - "learning_rate": 7.3578947368421065e-06, - "loss": 1.8968, - "step": 1301 - }, - { - "epoch": 1.67, - "grad_norm": 0.240234375, - "learning_rate": 7.347368421052632e-06, - "loss": 1.927, - "step": 1302 - }, - { - "epoch": 1.67, - "grad_norm": 0.306640625, - "learning_rate": 7.336842105263159e-06, - "loss": 1.8462, - "step": 1303 - }, - { - "epoch": 1.67, - "grad_norm": 0.2333984375, - "learning_rate": 7.326315789473685e-06, - "loss": 1.9024, - "step": 1304 - }, - { - "epoch": 1.67, - "grad_norm": 0.2080078125, - "learning_rate": 7.315789473684212e-06, - "loss": 1.8456, - "step": 1305 - }, - { - "epoch": 1.67, - "grad_norm": 0.275390625, - "learning_rate": 7.305263157894737e-06, - "loss": 1.8485, - "step": 1306 - }, - { - "epoch": 1.67, - "grad_norm": 0.291015625, - "learning_rate": 7.2947368421052636e-06, - "loss": 1.8647, - "step": 1307 - }, - { - "epoch": 1.67, - "grad_norm": 0.2158203125, - "learning_rate": 7.28421052631579e-06, - "loss": 1.8696, - "step": 1308 - }, - { - "epoch": 1.67, - "grad_norm": 0.205078125, - "learning_rate": 7.273684210526316e-06, - "loss": 1.917, - "step": 1309 - }, - { - "epoch": 1.68, - "grad_norm": 0.2412109375, - "learning_rate": 7.263157894736843e-06, - "loss": 1.8909, - "step": 1310 - }, - { - "epoch": 1.68, - "grad_norm": 0.21484375, - "learning_rate": 7.252631578947369e-06, - "loss": 1.8995, - "step": 1311 - }, - { - "epoch": 1.68, - "grad_norm": 0.2353515625, - "learning_rate": 7.242105263157896e-06, - "loss": 1.8922, - "step": 1312 - }, - { - "epoch": 1.68, - "grad_norm": 0.2421875, - "learning_rate": 7.2315789473684215e-06, - "loss": 1.9468, - "step": 1313 - }, - { - "epoch": 1.68, - "grad_norm": 0.2451171875, - "learning_rate": 7.221052631578948e-06, - "loss": 1.8185, - "step": 1314 - }, - { - "epoch": 1.68, - "grad_norm": 0.234375, - "learning_rate": 7.210526315789474e-06, - "loss": 1.8363, - "step": 1315 - }, - { - "epoch": 1.68, - "grad_norm": 0.33984375, - "learning_rate": 7.2000000000000005e-06, - "loss": 1.9007, - "step": 1316 - }, - { - "epoch": 1.69, - "grad_norm": 0.2333984375, - "learning_rate": 7.189473684210527e-06, - "loss": 1.9356, - "step": 1317 - }, - { - "epoch": 1.69, - "grad_norm": 0.234375, - "learning_rate": 7.178947368421053e-06, - "loss": 1.963, - "step": 1318 - }, - { - "epoch": 1.69, - "grad_norm": 0.25, - "learning_rate": 7.16842105263158e-06, - "loss": 1.8566, - "step": 1319 - }, - { - "epoch": 1.69, - "grad_norm": 0.314453125, - "learning_rate": 7.157894736842106e-06, - "loss": 1.8616, - "step": 1320 - }, - { - "epoch": 1.69, - "grad_norm": 0.2333984375, - "learning_rate": 7.147368421052631e-06, - "loss": 2.0066, - "step": 1321 - }, - { - "epoch": 1.69, - "grad_norm": 0.26171875, - "learning_rate": 7.1368421052631585e-06, - "loss": 1.9054, - "step": 1322 - }, - { - "epoch": 1.69, - "grad_norm": 0.232421875, - "learning_rate": 7.126315789473685e-06, - "loss": 1.829, - "step": 1323 - }, - { - "epoch": 1.69, - "grad_norm": 0.2255859375, - "learning_rate": 7.115789473684211e-06, - "loss": 1.9107, - "step": 1324 - }, - { - "epoch": 1.7, - "grad_norm": 0.251953125, - "learning_rate": 7.1052631578947375e-06, - "loss": 1.8422, - "step": 1325 - }, - { - "epoch": 1.7, - "grad_norm": 0.26171875, - "learning_rate": 7.094736842105265e-06, - "loss": 1.9214, - "step": 1326 - }, - { - "epoch": 1.7, - "grad_norm": 0.2490234375, - "learning_rate": 7.08421052631579e-06, - "loss": 1.9058, - "step": 1327 - }, - { - "epoch": 1.7, - "grad_norm": 0.298828125, - "learning_rate": 7.073684210526316e-06, - "loss": 1.8804, - "step": 1328 - }, - { - "epoch": 1.7, - "grad_norm": 0.2392578125, - "learning_rate": 7.063157894736843e-06, - "loss": 1.9205, - "step": 1329 - }, - { - "epoch": 1.7, - "grad_norm": 0.31640625, - "learning_rate": 7.052631578947369e-06, - "loss": 1.9135, - "step": 1330 - }, - { - "epoch": 1.7, - "grad_norm": 0.2490234375, - "learning_rate": 7.0421052631578954e-06, - "loss": 1.9461, - "step": 1331 - }, - { - "epoch": 1.7, - "grad_norm": 0.23046875, - "learning_rate": 7.031578947368422e-06, - "loss": 1.9539, - "step": 1332 - }, - { - "epoch": 1.71, - "grad_norm": 0.2119140625, - "learning_rate": 7.021052631578948e-06, - "loss": 1.9158, - "step": 1333 - }, - { - "epoch": 1.71, - "grad_norm": 0.24609375, - "learning_rate": 7.010526315789474e-06, - "loss": 2.013, - "step": 1334 - }, - { - "epoch": 1.71, - "grad_norm": 0.2470703125, - "learning_rate": 7e-06, - "loss": 1.9366, - "step": 1335 - }, - { - "epoch": 1.71, - "grad_norm": 0.322265625, - "learning_rate": 6.989473684210527e-06, - "loss": 1.8742, - "step": 1336 - }, - { - "epoch": 1.71, - "grad_norm": 0.255859375, - "learning_rate": 6.9789473684210525e-06, - "loss": 2.0126, - "step": 1337 - }, - { - "epoch": 1.71, - "grad_norm": 0.2119140625, - "learning_rate": 6.96842105263158e-06, - "loss": 1.9438, - "step": 1338 - }, - { - "epoch": 1.71, - "grad_norm": 0.2265625, - "learning_rate": 6.957894736842106e-06, - "loss": 1.9372, - "step": 1339 - }, - { - "epoch": 1.71, - "grad_norm": 0.267578125, - "learning_rate": 6.947368421052632e-06, - "loss": 1.9576, - "step": 1340 - }, - { - "epoch": 1.72, - "grad_norm": 0.259765625, - "learning_rate": 6.936842105263159e-06, - "loss": 2.0031, - "step": 1341 - }, - { - "epoch": 1.72, - "grad_norm": 0.255859375, - "learning_rate": 6.926315789473684e-06, - "loss": 1.8873, - "step": 1342 - }, - { - "epoch": 1.72, - "grad_norm": 0.2294921875, - "learning_rate": 6.915789473684211e-06, - "loss": 1.8633, - "step": 1343 - }, - { - "epoch": 1.72, - "grad_norm": 0.2392578125, - "learning_rate": 6.905263157894737e-06, - "loss": 1.8821, - "step": 1344 - }, - { - "epoch": 1.72, - "grad_norm": 0.2451171875, - "learning_rate": 6.894736842105264e-06, - "loss": 1.8588, - "step": 1345 - }, - { - "epoch": 1.72, - "grad_norm": 0.271484375, - "learning_rate": 6.8842105263157895e-06, - "loss": 1.9406, - "step": 1346 - }, - { - "epoch": 1.72, - "grad_norm": 0.310546875, - "learning_rate": 6.873684210526317e-06, - "loss": 1.9178, - "step": 1347 - }, - { - "epoch": 1.72, - "grad_norm": 0.318359375, - "learning_rate": 6.863157894736843e-06, - "loss": 1.8435, - "step": 1348 - }, - { - "epoch": 1.73, - "grad_norm": 0.2275390625, - "learning_rate": 6.8526315789473685e-06, - "loss": 2.0113, - "step": 1349 - }, - { - "epoch": 1.73, - "grad_norm": 0.23046875, - "learning_rate": 6.842105263157896e-06, - "loss": 1.9184, - "step": 1350 - }, - { - "epoch": 1.73, - "grad_norm": 0.2373046875, - "learning_rate": 6.831578947368421e-06, - "loss": 1.9393, - "step": 1351 - }, - { - "epoch": 1.73, - "grad_norm": 0.251953125, - "learning_rate": 6.821052631578948e-06, - "loss": 1.9022, - "step": 1352 - }, - { - "epoch": 1.73, - "grad_norm": 0.263671875, - "learning_rate": 6.810526315789474e-06, - "loss": 1.8978, - "step": 1353 - }, - { - "epoch": 1.73, - "grad_norm": 0.2412109375, - "learning_rate": 6.800000000000001e-06, - "loss": 1.9155, - "step": 1354 - }, - { - "epoch": 1.73, - "grad_norm": 0.2294921875, - "learning_rate": 6.789473684210527e-06, - "loss": 1.9096, - "step": 1355 - }, - { - "epoch": 1.74, - "grad_norm": 0.236328125, - "learning_rate": 6.778947368421053e-06, - "loss": 2.0138, - "step": 1356 - }, - { - "epoch": 1.74, - "grad_norm": 0.2265625, - "learning_rate": 6.76842105263158e-06, - "loss": 1.8323, - "step": 1357 - }, - { - "epoch": 1.74, - "grad_norm": 0.2099609375, - "learning_rate": 6.7578947368421054e-06, - "loss": 1.9182, - "step": 1358 - }, - { - "epoch": 1.74, - "grad_norm": 0.2255859375, - "learning_rate": 6.747368421052633e-06, - "loss": 1.9222, - "step": 1359 - }, - { - "epoch": 1.74, - "grad_norm": 0.236328125, - "learning_rate": 6.736842105263158e-06, - "loss": 1.9302, - "step": 1360 - }, - { - "epoch": 1.74, - "grad_norm": 0.2197265625, - "learning_rate": 6.726315789473685e-06, - "loss": 1.9111, - "step": 1361 - }, - { - "epoch": 1.74, - "grad_norm": 0.26953125, - "learning_rate": 6.715789473684211e-06, - "loss": 1.945, - "step": 1362 - }, - { - "epoch": 1.74, - "grad_norm": 0.2197265625, - "learning_rate": 6.705263157894737e-06, - "loss": 1.8843, - "step": 1363 - }, - { - "epoch": 1.75, - "grad_norm": 0.30078125, - "learning_rate": 6.694736842105264e-06, - "loss": 1.909, - "step": 1364 - }, - { - "epoch": 1.75, - "grad_norm": 0.26171875, - "learning_rate": 6.68421052631579e-06, - "loss": 1.9715, - "step": 1365 - }, - { - "epoch": 1.75, - "grad_norm": 0.283203125, - "learning_rate": 6.673684210526317e-06, - "loss": 1.9393, - "step": 1366 - }, - { - "epoch": 1.75, - "grad_norm": 0.279296875, - "learning_rate": 6.663157894736842e-06, - "loss": 1.8699, - "step": 1367 - }, - { - "epoch": 1.75, - "grad_norm": 0.21875, - "learning_rate": 6.6526315789473695e-06, - "loss": 1.9398, - "step": 1368 - }, - { - "epoch": 1.75, - "grad_norm": 0.24609375, - "learning_rate": 6.642105263157895e-06, - "loss": 1.9197, - "step": 1369 - }, - { - "epoch": 1.75, - "grad_norm": 0.2177734375, - "learning_rate": 6.631578947368421e-06, - "loss": 1.9059, - "step": 1370 - }, - { - "epoch": 1.75, - "grad_norm": 0.271484375, - "learning_rate": 6.621052631578948e-06, - "loss": 1.9082, - "step": 1371 - }, - { - "epoch": 1.76, - "grad_norm": 0.306640625, - "learning_rate": 6.610526315789474e-06, - "loss": 1.9589, - "step": 1372 - }, - { - "epoch": 1.76, - "grad_norm": 0.302734375, - "learning_rate": 6.600000000000001e-06, - "loss": 1.9758, - "step": 1373 - }, - { - "epoch": 1.76, - "grad_norm": 0.28515625, - "learning_rate": 6.589473684210527e-06, - "loss": 1.9041, - "step": 1374 - }, - { - "epoch": 1.76, - "grad_norm": 0.251953125, - "learning_rate": 6.578947368421054e-06, - "loss": 1.9452, - "step": 1375 - }, - { - "epoch": 1.76, - "grad_norm": 0.2294921875, - "learning_rate": 6.568421052631579e-06, - "loss": 1.8756, - "step": 1376 - }, - { - "epoch": 1.76, - "grad_norm": 0.21484375, - "learning_rate": 6.557894736842106e-06, - "loss": 1.8808, - "step": 1377 - }, - { - "epoch": 1.76, - "grad_norm": 0.2421875, - "learning_rate": 6.547368421052632e-06, - "loss": 1.9059, - "step": 1378 - }, - { - "epoch": 1.76, - "grad_norm": 0.212890625, - "learning_rate": 6.536842105263158e-06, - "loss": 1.9956, - "step": 1379 - }, - { - "epoch": 1.77, - "grad_norm": 0.236328125, - "learning_rate": 6.526315789473685e-06, - "loss": 1.9977, - "step": 1380 - }, - { - "epoch": 1.77, - "grad_norm": 0.2421875, - "learning_rate": 6.515789473684211e-06, - "loss": 1.8735, - "step": 1381 - }, - { - "epoch": 1.77, - "grad_norm": 0.2177734375, - "learning_rate": 6.505263157894738e-06, - "loss": 1.8738, - "step": 1382 - }, - { - "epoch": 1.77, - "grad_norm": 0.23046875, - "learning_rate": 6.494736842105264e-06, - "loss": 1.9186, - "step": 1383 - }, - { - "epoch": 1.77, - "grad_norm": 0.26171875, - "learning_rate": 6.484210526315789e-06, - "loss": 1.8935, - "step": 1384 - }, - { - "epoch": 1.77, - "grad_norm": 0.267578125, - "learning_rate": 6.473684210526316e-06, - "loss": 1.9347, - "step": 1385 - }, - { - "epoch": 1.77, - "grad_norm": 0.31640625, - "learning_rate": 6.463157894736843e-06, - "loss": 1.9152, - "step": 1386 - }, - { - "epoch": 1.77, - "grad_norm": 0.255859375, - "learning_rate": 6.452631578947369e-06, - "loss": 2.0466, - "step": 1387 - }, - { - "epoch": 1.78, - "grad_norm": 0.208984375, - "learning_rate": 6.442105263157895e-06, - "loss": 1.8459, - "step": 1388 - }, - { - "epoch": 1.78, - "grad_norm": 0.2265625, - "learning_rate": 6.431578947368422e-06, - "loss": 1.8968, - "step": 1389 - }, - { - "epoch": 1.78, - "grad_norm": 0.2890625, - "learning_rate": 6.421052631578948e-06, - "loss": 1.906, - "step": 1390 - }, - { - "epoch": 1.78, - "grad_norm": 0.228515625, - "learning_rate": 6.410526315789473e-06, - "loss": 1.8801, - "step": 1391 - }, - { - "epoch": 1.78, - "grad_norm": 0.216796875, - "learning_rate": 6.4000000000000006e-06, - "loss": 1.8669, - "step": 1392 - }, - { - "epoch": 1.78, - "grad_norm": 0.25390625, - "learning_rate": 6.389473684210527e-06, - "loss": 1.9109, - "step": 1393 - }, - { - "epoch": 1.78, - "grad_norm": 0.255859375, - "learning_rate": 6.378947368421053e-06, - "loss": 1.8578, - "step": 1394 - }, - { - "epoch": 1.79, - "grad_norm": 0.2373046875, - "learning_rate": 6.3684210526315795e-06, - "loss": 1.9041, - "step": 1395 - }, - { - "epoch": 1.79, - "grad_norm": 0.2421875, - "learning_rate": 6.357894736842106e-06, - "loss": 1.9224, - "step": 1396 - }, - { - "epoch": 1.79, - "grad_norm": 0.240234375, - "learning_rate": 6.347368421052632e-06, - "loss": 1.916, - "step": 1397 - }, - { - "epoch": 1.79, - "grad_norm": 0.25, - "learning_rate": 6.336842105263158e-06, - "loss": 1.8718, - "step": 1398 - }, - { - "epoch": 1.79, - "grad_norm": 0.263671875, - "learning_rate": 6.326315789473685e-06, - "loss": 1.8821, - "step": 1399 - }, - { - "epoch": 1.79, - "grad_norm": 0.240234375, - "learning_rate": 6.31578947368421e-06, - "loss": 1.8528, - "step": 1400 - }, - { - "epoch": 1.79, - "eval_loss": 1.895195722579956, - "eval_runtime": 125.5929, - "eval_samples_per_second": 39.811, - "eval_steps_per_second": 1.25, - "step": 1400 - }, - { - "epoch": 1.79, - "grad_norm": 0.26171875, - "learning_rate": 6.3052631578947375e-06, - "loss": 1.9258, - "step": 1401 - }, - { - "epoch": 1.79, - "grad_norm": 0.25, - "learning_rate": 6.294736842105264e-06, - "loss": 1.8574, - "step": 1402 - }, - { - "epoch": 1.8, - "grad_norm": 0.24609375, - "learning_rate": 6.28421052631579e-06, - "loss": 1.9348, - "step": 1403 - }, - { - "epoch": 1.8, - "grad_norm": 0.2177734375, - "learning_rate": 6.2736842105263165e-06, - "loss": 1.8845, - "step": 1404 - }, - { - "epoch": 1.8, - "grad_norm": 0.251953125, - "learning_rate": 6.263157894736842e-06, - "loss": 1.968, - "step": 1405 - }, - { - "epoch": 1.8, - "grad_norm": 0.216796875, - "learning_rate": 6.252631578947369e-06, - "loss": 1.8991, - "step": 1406 - }, - { - "epoch": 1.8, - "grad_norm": 0.240234375, - "learning_rate": 6.242105263157895e-06, - "loss": 1.9025, - "step": 1407 - }, - { - "epoch": 1.8, - "grad_norm": 0.25, - "learning_rate": 6.231578947368422e-06, - "loss": 1.9865, - "step": 1408 - }, - { - "epoch": 1.8, - "grad_norm": 0.30859375, - "learning_rate": 6.221052631578947e-06, - "loss": 1.8611, - "step": 1409 - }, - { - "epoch": 1.8, - "grad_norm": 0.29296875, - "learning_rate": 6.2105263157894745e-06, - "loss": 1.8929, - "step": 1410 - }, - { - "epoch": 1.81, - "grad_norm": 0.2578125, - "learning_rate": 6.200000000000001e-06, - "loss": 1.8693, - "step": 1411 - }, - { - "epoch": 1.81, - "grad_norm": 0.294921875, - "learning_rate": 6.189473684210526e-06, - "loss": 1.9122, - "step": 1412 - }, - { - "epoch": 1.81, - "grad_norm": 0.2412109375, - "learning_rate": 6.1789473684210534e-06, - "loss": 1.9358, - "step": 1413 - }, - { - "epoch": 1.81, - "grad_norm": 0.25, - "learning_rate": 6.168421052631579e-06, - "loss": 1.941, - "step": 1414 - }, - { - "epoch": 1.81, - "grad_norm": 0.2060546875, - "learning_rate": 6.157894736842106e-06, - "loss": 1.8947, - "step": 1415 - }, - { - "epoch": 1.81, - "grad_norm": 0.248046875, - "learning_rate": 6.1473684210526316e-06, - "loss": 1.8962, - "step": 1416 - }, - { - "epoch": 1.81, - "grad_norm": 0.265625, - "learning_rate": 6.136842105263159e-06, - "loss": 1.9131, - "step": 1417 - }, - { - "epoch": 1.81, - "grad_norm": 0.2451171875, - "learning_rate": 6.126315789473685e-06, - "loss": 1.867, - "step": 1418 - }, - { - "epoch": 1.82, - "grad_norm": 0.2236328125, - "learning_rate": 6.1157894736842106e-06, - "loss": 1.9122, - "step": 1419 - }, - { - "epoch": 1.82, - "grad_norm": 0.2451171875, - "learning_rate": 6.105263157894738e-06, - "loss": 1.9009, - "step": 1420 - }, - { - "epoch": 1.82, - "grad_norm": 0.255859375, - "learning_rate": 6.094736842105263e-06, - "loss": 1.9027, - "step": 1421 - }, - { - "epoch": 1.82, - "grad_norm": 0.2197265625, - "learning_rate": 6.08421052631579e-06, - "loss": 1.8239, - "step": 1422 - }, - { - "epoch": 1.82, - "grad_norm": 0.212890625, - "learning_rate": 6.073684210526316e-06, - "loss": 1.875, - "step": 1423 - }, - { - "epoch": 1.82, - "grad_norm": 0.306640625, - "learning_rate": 6.063157894736843e-06, - "loss": 1.8322, - "step": 1424 - }, - { - "epoch": 1.82, - "grad_norm": 0.220703125, - "learning_rate": 6.0526315789473685e-06, - "loss": 1.8125, - "step": 1425 - }, - { - "epoch": 1.82, - "grad_norm": 0.2890625, - "learning_rate": 6.042105263157895e-06, - "loss": 1.8788, - "step": 1426 - }, - { - "epoch": 1.83, - "grad_norm": 0.2197265625, - "learning_rate": 6.031578947368422e-06, - "loss": 1.8411, - "step": 1427 - }, - { - "epoch": 1.83, - "grad_norm": 0.28515625, - "learning_rate": 6.0210526315789475e-06, - "loss": 1.906, - "step": 1428 - }, - { - "epoch": 1.83, - "grad_norm": 0.25, - "learning_rate": 6.010526315789475e-06, - "loss": 1.9611, - "step": 1429 - }, - { - "epoch": 1.83, - "grad_norm": 0.212890625, - "learning_rate": 6e-06, - "loss": 1.9268, - "step": 1430 - }, - { - "epoch": 1.83, - "grad_norm": 0.216796875, - "learning_rate": 5.989473684210527e-06, - "loss": 1.9049, - "step": 1431 - }, - { - "epoch": 1.83, - "grad_norm": 0.25, - "learning_rate": 5.978947368421053e-06, - "loss": 1.7729, - "step": 1432 - }, - { - "epoch": 1.83, - "grad_norm": 0.26953125, - "learning_rate": 5.968421052631579e-06, - "loss": 2.0196, - "step": 1433 - }, - { - "epoch": 1.83, - "grad_norm": 0.232421875, - "learning_rate": 5.9578947368421055e-06, - "loss": 1.8925, - "step": 1434 - }, - { - "epoch": 1.84, - "grad_norm": 0.20703125, - "learning_rate": 5.947368421052632e-06, - "loss": 1.9573, - "step": 1435 - }, - { - "epoch": 1.84, - "grad_norm": 0.271484375, - "learning_rate": 5.936842105263159e-06, - "loss": 1.9993, - "step": 1436 - }, - { - "epoch": 1.84, - "grad_norm": 0.2255859375, - "learning_rate": 5.9263157894736844e-06, - "loss": 1.9408, - "step": 1437 - }, - { - "epoch": 1.84, - "grad_norm": 0.244140625, - "learning_rate": 5.915789473684212e-06, - "loss": 1.8908, - "step": 1438 - }, - { - "epoch": 1.84, - "grad_norm": 0.2412109375, - "learning_rate": 5.905263157894737e-06, - "loss": 1.943, - "step": 1439 - }, - { - "epoch": 1.84, - "grad_norm": 0.2314453125, - "learning_rate": 5.8947368421052634e-06, - "loss": 1.9589, - "step": 1440 - }, - { - "epoch": 1.84, - "grad_norm": 0.265625, - "learning_rate": 5.88421052631579e-06, - "loss": 1.8824, - "step": 1441 - }, - { - "epoch": 1.85, - "grad_norm": 0.251953125, - "learning_rate": 5.873684210526316e-06, - "loss": 1.896, - "step": 1442 - }, - { - "epoch": 1.85, - "grad_norm": 0.2470703125, - "learning_rate": 5.863157894736842e-06, - "loss": 1.8914, - "step": 1443 - }, - { - "epoch": 1.85, - "grad_norm": 0.2275390625, - "learning_rate": 5.852631578947369e-06, - "loss": 1.9748, - "step": 1444 - }, - { - "epoch": 1.85, - "grad_norm": 0.234375, - "learning_rate": 5.842105263157896e-06, - "loss": 1.9042, - "step": 1445 - }, - { - "epoch": 1.85, - "grad_norm": 0.25390625, - "learning_rate": 5.831578947368421e-06, - "loss": 1.9478, - "step": 1446 - }, - { - "epoch": 1.85, - "grad_norm": 0.25390625, - "learning_rate": 5.8210526315789486e-06, - "loss": 1.913, - "step": 1447 - }, - { - "epoch": 1.85, - "grad_norm": 0.265625, - "learning_rate": 5.810526315789474e-06, - "loss": 1.9093, - "step": 1448 - }, - { - "epoch": 1.85, - "grad_norm": 0.240234375, - "learning_rate": 5.8e-06, - "loss": 1.9195, - "step": 1449 - }, - { - "epoch": 1.86, - "grad_norm": 0.2412109375, - "learning_rate": 5.789473684210527e-06, - "loss": 1.8839, - "step": 1450 - }, - { - "epoch": 1.86, - "grad_norm": 0.2578125, - "learning_rate": 5.778947368421053e-06, - "loss": 1.9514, - "step": 1451 - }, - { - "epoch": 1.86, - "grad_norm": 0.259765625, - "learning_rate": 5.76842105263158e-06, - "loss": 1.9451, - "step": 1452 - }, - { - "epoch": 1.86, - "grad_norm": 0.2275390625, - "learning_rate": 5.757894736842106e-06, - "loss": 1.9538, - "step": 1453 - }, - { - "epoch": 1.86, - "grad_norm": 0.2578125, - "learning_rate": 5.747368421052633e-06, - "loss": 1.8807, - "step": 1454 - }, - { - "epoch": 1.86, - "grad_norm": 0.248046875, - "learning_rate": 5.736842105263158e-06, - "loss": 1.8817, - "step": 1455 - }, - { - "epoch": 1.86, - "grad_norm": 0.328125, - "learning_rate": 5.726315789473685e-06, - "loss": 1.8869, - "step": 1456 - }, - { - "epoch": 1.86, - "grad_norm": 0.205078125, - "learning_rate": 5.715789473684211e-06, - "loss": 1.9214, - "step": 1457 - }, - { - "epoch": 1.87, - "grad_norm": 0.2216796875, - "learning_rate": 5.705263157894737e-06, - "loss": 1.8007, - "step": 1458 - }, - { - "epoch": 1.87, - "grad_norm": 0.23828125, - "learning_rate": 5.694736842105264e-06, - "loss": 1.8808, - "step": 1459 - }, - { - "epoch": 1.87, - "grad_norm": 0.263671875, - "learning_rate": 5.68421052631579e-06, - "loss": 1.9734, - "step": 1460 - }, - { - "epoch": 1.87, - "grad_norm": 0.2294921875, - "learning_rate": 5.673684210526317e-06, - "loss": 1.9555, - "step": 1461 - }, - { - "epoch": 1.87, - "grad_norm": 0.2412109375, - "learning_rate": 5.663157894736843e-06, - "loss": 1.8947, - "step": 1462 - }, - { - "epoch": 1.87, - "grad_norm": 0.228515625, - "learning_rate": 5.652631578947368e-06, - "loss": 1.8661, - "step": 1463 - }, - { - "epoch": 1.87, - "grad_norm": 0.2353515625, - "learning_rate": 5.642105263157895e-06, - "loss": 1.8801, - "step": 1464 - }, - { - "epoch": 1.87, - "grad_norm": 0.21484375, - "learning_rate": 5.631578947368422e-06, - "loss": 1.9487, - "step": 1465 - }, - { - "epoch": 1.88, - "grad_norm": 0.2431640625, - "learning_rate": 5.621052631578948e-06, - "loss": 1.8386, - "step": 1466 - }, - { - "epoch": 1.88, - "grad_norm": 0.2333984375, - "learning_rate": 5.610526315789474e-06, - "loss": 1.9301, - "step": 1467 - }, - { - "epoch": 1.88, - "grad_norm": 0.212890625, - "learning_rate": 5.600000000000001e-06, - "loss": 1.8864, - "step": 1468 - }, - { - "epoch": 1.88, - "grad_norm": 0.224609375, - "learning_rate": 5.589473684210527e-06, - "loss": 1.8681, - "step": 1469 - }, - { - "epoch": 1.88, - "grad_norm": 0.271484375, - "learning_rate": 5.578947368421052e-06, - "loss": 1.8573, - "step": 1470 - }, - { - "epoch": 1.88, - "grad_norm": 0.2177734375, - "learning_rate": 5.5684210526315796e-06, - "loss": 1.8511, - "step": 1471 - }, - { - "epoch": 1.88, - "grad_norm": 0.2392578125, - "learning_rate": 5.557894736842105e-06, - "loss": 1.9348, - "step": 1472 - }, - { - "epoch": 1.88, - "grad_norm": 0.2451171875, - "learning_rate": 5.547368421052632e-06, - "loss": 1.9307, - "step": 1473 - }, - { - "epoch": 1.89, - "grad_norm": 0.25390625, - "learning_rate": 5.5368421052631586e-06, - "loss": 1.8966, - "step": 1474 - }, - { - "epoch": 1.89, - "grad_norm": 0.275390625, - "learning_rate": 5.526315789473685e-06, - "loss": 1.8971, - "step": 1475 - }, - { - "epoch": 1.89, - "grad_norm": 0.240234375, - "learning_rate": 5.515789473684211e-06, - "loss": 1.8926, - "step": 1476 - }, - { - "epoch": 1.89, - "grad_norm": 0.3203125, - "learning_rate": 5.505263157894737e-06, - "loss": 1.8474, - "step": 1477 - }, - { - "epoch": 1.89, - "grad_norm": 0.212890625, - "learning_rate": 5.494736842105264e-06, - "loss": 1.8972, - "step": 1478 - }, - { - "epoch": 1.89, - "grad_norm": 0.251953125, - "learning_rate": 5.484210526315789e-06, - "loss": 1.9631, - "step": 1479 - }, - { - "epoch": 1.89, - "grad_norm": 0.234375, - "learning_rate": 5.4736842105263165e-06, - "loss": 1.8804, - "step": 1480 - }, - { - "epoch": 1.9, - "grad_norm": 0.2275390625, - "learning_rate": 5.463157894736843e-06, - "loss": 1.9367, - "step": 1481 - }, - { - "epoch": 1.9, - "grad_norm": 0.2314453125, - "learning_rate": 5.452631578947369e-06, - "loss": 1.8806, - "step": 1482 - }, - { - "epoch": 1.9, - "grad_norm": 0.23828125, - "learning_rate": 5.4421052631578955e-06, - "loss": 1.8513, - "step": 1483 - }, - { - "epoch": 1.9, - "grad_norm": 0.27734375, - "learning_rate": 5.431578947368421e-06, - "loss": 1.9457, - "step": 1484 - }, - { - "epoch": 1.9, - "grad_norm": 0.2353515625, - "learning_rate": 5.421052631578948e-06, - "loss": 1.9683, - "step": 1485 - }, - { - "epoch": 1.9, - "grad_norm": 0.2373046875, - "learning_rate": 5.410526315789474e-06, - "loss": 1.8526, - "step": 1486 - }, - { - "epoch": 1.9, - "grad_norm": 0.28125, - "learning_rate": 5.400000000000001e-06, - "loss": 1.9982, - "step": 1487 - }, - { - "epoch": 1.9, - "grad_norm": 0.2412109375, - "learning_rate": 5.389473684210526e-06, - "loss": 1.9584, - "step": 1488 - }, - { - "epoch": 1.91, - "grad_norm": 0.2353515625, - "learning_rate": 5.3789473684210535e-06, - "loss": 1.8936, - "step": 1489 - }, - { - "epoch": 1.91, - "grad_norm": 0.31640625, - "learning_rate": 5.36842105263158e-06, - "loss": 1.9046, - "step": 1490 - }, - { - "epoch": 1.91, - "grad_norm": 0.21875, - "learning_rate": 5.357894736842105e-06, - "loss": 1.9039, - "step": 1491 - }, - { - "epoch": 1.91, - "grad_norm": 0.267578125, - "learning_rate": 5.3473684210526325e-06, - "loss": 1.9029, - "step": 1492 - }, - { - "epoch": 1.91, - "grad_norm": 0.2109375, - "learning_rate": 5.336842105263158e-06, - "loss": 1.8936, - "step": 1493 - }, - { - "epoch": 1.91, - "grad_norm": 0.2041015625, - "learning_rate": 5.326315789473685e-06, - "loss": 1.8142, - "step": 1494 - }, - { - "epoch": 1.91, - "grad_norm": 0.2412109375, - "learning_rate": 5.315789473684211e-06, - "loss": 1.9549, - "step": 1495 - }, - { - "epoch": 1.91, - "grad_norm": 0.330078125, - "learning_rate": 5.305263157894738e-06, - "loss": 1.9496, - "step": 1496 - }, - { - "epoch": 1.92, - "grad_norm": 0.23046875, - "learning_rate": 5.294736842105263e-06, - "loss": 1.865, - "step": 1497 - }, - { - "epoch": 1.92, - "grad_norm": 0.21875, - "learning_rate": 5.2842105263157896e-06, - "loss": 1.7992, - "step": 1498 - }, - { - "epoch": 1.92, - "grad_norm": 0.26953125, - "learning_rate": 5.273684210526317e-06, - "loss": 1.974, - "step": 1499 - }, - { - "epoch": 1.92, - "grad_norm": 0.2373046875, - "learning_rate": 5.263157894736842e-06, - "loss": 1.9513, - "step": 1500 - }, - { - "epoch": 1.92, - "grad_norm": 0.2412109375, - "learning_rate": 5.252631578947369e-06, - "loss": 1.9872, - "step": 1501 - }, - { - "epoch": 1.92, - "grad_norm": 0.23828125, - "learning_rate": 5.242105263157895e-06, - "loss": 1.9384, - "step": 1502 - }, - { - "epoch": 1.92, - "grad_norm": 0.240234375, - "learning_rate": 5.231578947368422e-06, - "loss": 1.877, - "step": 1503 - }, - { - "epoch": 1.92, - "grad_norm": 0.265625, - "learning_rate": 5.2210526315789475e-06, - "loss": 1.92, - "step": 1504 - }, - { - "epoch": 1.93, - "grad_norm": 0.2197265625, - "learning_rate": 5.210526315789474e-06, - "loss": 1.9279, - "step": 1505 - }, - { - "epoch": 1.93, - "grad_norm": 0.2421875, - "learning_rate": 5.2e-06, - "loss": 1.8567, - "step": 1506 - }, - { - "epoch": 1.93, - "grad_norm": 0.2041015625, - "learning_rate": 5.1894736842105265e-06, - "loss": 1.8203, - "step": 1507 - }, - { - "epoch": 1.93, - "grad_norm": 0.26171875, - "learning_rate": 5.178947368421054e-06, - "loss": 1.9934, - "step": 1508 - }, - { - "epoch": 1.93, - "grad_norm": 0.2734375, - "learning_rate": 5.168421052631579e-06, - "loss": 1.9312, - "step": 1509 - }, - { - "epoch": 1.93, - "grad_norm": 0.236328125, - "learning_rate": 5.157894736842106e-06, - "loss": 1.9267, - "step": 1510 - }, - { - "epoch": 1.93, - "grad_norm": 0.26953125, - "learning_rate": 5.147368421052632e-06, - "loss": 1.9068, - "step": 1511 - }, - { - "epoch": 1.93, - "grad_norm": 0.23828125, - "learning_rate": 5.136842105263158e-06, - "loss": 1.9, - "step": 1512 - }, - { - "epoch": 1.94, - "grad_norm": 0.25390625, - "learning_rate": 5.1263157894736845e-06, - "loss": 1.9352, - "step": 1513 - }, - { - "epoch": 1.94, - "grad_norm": 0.248046875, - "learning_rate": 5.115789473684211e-06, - "loss": 1.9252, - "step": 1514 - }, - { - "epoch": 1.94, - "grad_norm": 0.240234375, - "learning_rate": 5.105263157894738e-06, - "loss": 1.8704, - "step": 1515 - }, - { - "epoch": 1.94, - "grad_norm": 0.2158203125, - "learning_rate": 5.0947368421052635e-06, - "loss": 1.8296, - "step": 1516 - }, - { - "epoch": 1.94, - "grad_norm": 0.2333984375, - "learning_rate": 5.084210526315791e-06, - "loss": 1.8916, - "step": 1517 - }, - { - "epoch": 1.94, - "grad_norm": 0.2275390625, - "learning_rate": 5.073684210526316e-06, - "loss": 1.9229, - "step": 1518 - }, - { - "epoch": 1.94, - "grad_norm": 0.25, - "learning_rate": 5.0631578947368424e-06, - "loss": 1.9261, - "step": 1519 - }, - { - "epoch": 1.94, - "grad_norm": 0.23046875, - "learning_rate": 5.052631578947369e-06, - "loss": 1.9422, - "step": 1520 - }, - { - "epoch": 1.95, - "grad_norm": 0.2734375, - "learning_rate": 5.042105263157895e-06, - "loss": 1.8628, - "step": 1521 - }, - { - "epoch": 1.95, - "grad_norm": 0.236328125, - "learning_rate": 5.0315789473684214e-06, - "loss": 1.881, - "step": 1522 - }, - { - "epoch": 1.95, - "grad_norm": 0.208984375, - "learning_rate": 5.021052631578948e-06, - "loss": 1.8279, - "step": 1523 - }, - { - "epoch": 1.95, - "grad_norm": 0.271484375, - "learning_rate": 5.010526315789475e-06, - "loss": 1.9354, - "step": 1524 - }, - { - "epoch": 1.95, - "grad_norm": 0.224609375, - "learning_rate": 5e-06, - "loss": 1.8678, - "step": 1525 - }, - { - "epoch": 1.95, - "grad_norm": 0.23046875, - "learning_rate": 4.989473684210527e-06, - "loss": 1.913, - "step": 1526 - }, - { - "epoch": 1.95, - "grad_norm": 0.2890625, - "learning_rate": 4.978947368421053e-06, - "loss": 1.8205, - "step": 1527 - }, - { - "epoch": 1.96, - "grad_norm": 0.267578125, - "learning_rate": 4.968421052631579e-06, - "loss": 1.8916, - "step": 1528 - }, - { - "epoch": 1.96, - "grad_norm": 0.216796875, - "learning_rate": 4.957894736842106e-06, - "loss": 1.9832, - "step": 1529 - }, - { - "epoch": 1.96, - "grad_norm": 0.26171875, - "learning_rate": 4.947368421052632e-06, - "loss": 1.8796, - "step": 1530 - }, - { - "epoch": 1.96, - "grad_norm": 0.251953125, - "learning_rate": 4.936842105263158e-06, - "loss": 1.905, - "step": 1531 - }, - { - "epoch": 1.96, - "grad_norm": 0.22265625, - "learning_rate": 4.926315789473685e-06, - "loss": 1.8748, - "step": 1532 - }, - { - "epoch": 1.96, - "grad_norm": 0.2314453125, - "learning_rate": 4.915789473684211e-06, - "loss": 1.8688, - "step": 1533 - }, - { - "epoch": 1.96, - "grad_norm": 0.2314453125, - "learning_rate": 4.905263157894737e-06, - "loss": 1.8741, - "step": 1534 - }, - { - "epoch": 1.96, - "grad_norm": 0.23046875, - "learning_rate": 4.894736842105264e-06, - "loss": 1.8761, - "step": 1535 - }, - { - "epoch": 1.97, - "grad_norm": 0.259765625, - "learning_rate": 4.88421052631579e-06, - "loss": 1.9244, - "step": 1536 - }, - { - "epoch": 1.97, - "grad_norm": 0.23046875, - "learning_rate": 4.873684210526316e-06, - "loss": 1.9502, - "step": 1537 - }, - { - "epoch": 1.97, - "grad_norm": 0.232421875, - "learning_rate": 4.863157894736843e-06, - "loss": 1.8882, - "step": 1538 - }, - { - "epoch": 1.97, - "grad_norm": 0.236328125, - "learning_rate": 4.852631578947369e-06, - "loss": 1.91, - "step": 1539 - }, - { - "epoch": 1.97, - "grad_norm": 0.37890625, - "learning_rate": 4.842105263157895e-06, - "loss": 1.8705, - "step": 1540 - }, - { - "epoch": 1.97, - "grad_norm": 0.22265625, - "learning_rate": 4.831578947368422e-06, - "loss": 1.9757, - "step": 1541 - }, - { - "epoch": 1.97, - "grad_norm": 0.25390625, - "learning_rate": 4.821052631578948e-06, - "loss": 1.7779, - "step": 1542 - }, - { - "epoch": 1.97, - "grad_norm": 0.25, - "learning_rate": 4.8105263157894735e-06, - "loss": 1.9656, - "step": 1543 - }, - { - "epoch": 1.98, - "grad_norm": 0.234375, - "learning_rate": 4.800000000000001e-06, - "loss": 1.7459, - "step": 1544 - }, - { - "epoch": 1.98, - "grad_norm": 0.2138671875, - "learning_rate": 4.789473684210527e-06, - "loss": 1.9498, - "step": 1545 - }, - { - "epoch": 1.98, - "grad_norm": 0.2392578125, - "learning_rate": 4.778947368421053e-06, - "loss": 1.9647, - "step": 1546 - }, - { - "epoch": 1.98, - "grad_norm": 0.22265625, - "learning_rate": 4.76842105263158e-06, - "loss": 1.9162, - "step": 1547 - }, - { - "epoch": 1.98, - "grad_norm": 0.3046875, - "learning_rate": 4.757894736842106e-06, - "loss": 1.8916, - "step": 1548 - }, - { - "epoch": 1.98, - "grad_norm": 0.25, - "learning_rate": 4.747368421052632e-06, - "loss": 1.9362, - "step": 1549 - }, - { - "epoch": 1.98, - "grad_norm": 0.2451171875, - "learning_rate": 4.736842105263158e-06, - "loss": 1.8838, - "step": 1550 - }, - { - "epoch": 1.98, - "grad_norm": 0.259765625, - "learning_rate": 4.726315789473684e-06, - "loss": 1.917, - "step": 1551 - }, - { - "epoch": 1.99, - "grad_norm": 0.228515625, - "learning_rate": 4.71578947368421e-06, - "loss": 1.9427, - "step": 1552 - }, - { - "epoch": 1.99, - "grad_norm": 0.2412109375, - "learning_rate": 4.705263157894738e-06, - "loss": 1.8813, - "step": 1553 - }, - { - "epoch": 1.99, - "grad_norm": 0.2431640625, - "learning_rate": 4.694736842105264e-06, - "loss": 1.942, - "step": 1554 - }, - { - "epoch": 1.99, - "grad_norm": 0.2265625, - "learning_rate": 4.68421052631579e-06, - "loss": 1.9783, - "step": 1555 - }, - { - "epoch": 1.99, - "grad_norm": 0.2333984375, - "learning_rate": 4.6736842105263166e-06, - "loss": 1.828, - "step": 1556 - }, - { - "epoch": 1.99, - "grad_norm": 0.2138671875, - "learning_rate": 4.663157894736842e-06, - "loss": 1.934, - "step": 1557 - }, - { - "epoch": 1.99, - "grad_norm": 0.23828125, - "learning_rate": 4.652631578947368e-06, - "loss": 1.8107, - "step": 1558 - }, - { - "epoch": 1.99, - "grad_norm": 0.2197265625, - "learning_rate": 4.642105263157895e-06, - "loss": 1.956, - "step": 1559 - }, - { - "epoch": 2.0, - "grad_norm": 0.400390625, - "learning_rate": 4.631578947368421e-06, - "loss": 1.8634, - "step": 1560 - }, - { - "epoch": 2.0, - "grad_norm": 0.23828125, - "learning_rate": 4.621052631578948e-06, - "loss": 1.9632, - "step": 1561 - }, - { - "epoch": 2.0, - "grad_norm": 0.267578125, - "learning_rate": 4.6105263157894745e-06, - "loss": 1.8474, - "step": 1562 - }, - { - "epoch": 2.0, - "grad_norm": 0.28125, - "learning_rate": 4.600000000000001e-06, - "loss": 1.9971, - "step": 1563 - }, - { - "epoch": 2.0, - "grad_norm": 0.2333984375, - "learning_rate": 4.589473684210526e-06, - "loss": 1.8644, - "step": 1564 - }, - { - "epoch": 2.0, - "grad_norm": 0.2392578125, - "learning_rate": 4.578947368421053e-06, - "loss": 1.912, - "step": 1565 - }, - { - "epoch": 2.0, - "grad_norm": 0.251953125, - "learning_rate": 4.568421052631579e-06, - "loss": 1.9676, - "step": 1566 - }, - { - "epoch": 2.01, - "grad_norm": 0.294921875, - "learning_rate": 4.557894736842105e-06, - "loss": 1.9009, - "step": 1567 - }, - { - "epoch": 2.01, - "grad_norm": 0.2314453125, - "learning_rate": 4.547368421052632e-06, - "loss": 1.9497, - "step": 1568 - }, - { - "epoch": 2.01, - "grad_norm": 0.21484375, - "learning_rate": 4.536842105263158e-06, - "loss": 1.8834, - "step": 1569 - }, - { - "epoch": 2.01, - "grad_norm": 0.248046875, - "learning_rate": 4.526315789473685e-06, - "loss": 2.0334, - "step": 1570 - }, - { - "epoch": 2.01, - "grad_norm": 0.244140625, - "learning_rate": 4.5157894736842115e-06, - "loss": 1.9218, - "step": 1571 - }, - { - "epoch": 2.01, - "grad_norm": 0.263671875, - "learning_rate": 4.505263157894737e-06, - "loss": 1.8624, - "step": 1572 - }, - { - "epoch": 2.01, - "grad_norm": 0.251953125, - "learning_rate": 4.494736842105263e-06, - "loss": 1.9691, - "step": 1573 - }, - { - "epoch": 2.01, - "grad_norm": 0.2275390625, - "learning_rate": 4.48421052631579e-06, - "loss": 1.9471, - "step": 1574 - }, - { - "epoch": 2.02, - "grad_norm": 0.255859375, - "learning_rate": 4.473684210526316e-06, - "loss": 1.8862, - "step": 1575 - }, - { - "epoch": 2.02, - "grad_norm": 0.2265625, - "learning_rate": 4.463157894736842e-06, - "loss": 1.8107, - "step": 1576 - }, - { - "epoch": 2.02, - "grad_norm": 0.259765625, - "learning_rate": 4.452631578947369e-06, - "loss": 1.9416, - "step": 1577 - }, - { - "epoch": 2.02, - "grad_norm": 0.2255859375, - "learning_rate": 4.442105263157896e-06, - "loss": 1.932, - "step": 1578 - }, - { - "epoch": 2.02, - "grad_norm": 0.298828125, - "learning_rate": 4.431578947368421e-06, - "loss": 1.9169, - "step": 1579 - }, - { - "epoch": 2.02, - "grad_norm": 0.3046875, - "learning_rate": 4.4210526315789476e-06, - "loss": 2.0032, - "step": 1580 - }, - { - "epoch": 2.02, - "grad_norm": 0.2412109375, - "learning_rate": 4.410526315789474e-06, - "loss": 2.0358, - "step": 1581 - }, - { - "epoch": 2.02, - "grad_norm": 0.248046875, - "learning_rate": 4.4e-06, - "loss": 1.7811, - "step": 1582 - }, - { - "epoch": 2.03, - "grad_norm": 0.2314453125, - "learning_rate": 4.3894736842105266e-06, - "loss": 1.953, - "step": 1583 - }, - { - "epoch": 2.03, - "grad_norm": 0.2333984375, - "learning_rate": 4.378947368421053e-06, - "loss": 1.9149, - "step": 1584 - }, - { - "epoch": 2.03, - "grad_norm": 0.2119140625, - "learning_rate": 4.368421052631579e-06, - "loss": 1.9241, - "step": 1585 - }, - { - "epoch": 2.03, - "grad_norm": 0.251953125, - "learning_rate": 4.3578947368421055e-06, - "loss": 1.9467, - "step": 1586 - }, - { - "epoch": 2.03, - "grad_norm": 0.263671875, - "learning_rate": 4.347368421052632e-06, - "loss": 1.9337, - "step": 1587 - }, - { - "epoch": 2.03, - "grad_norm": 0.267578125, - "learning_rate": 4.336842105263158e-06, - "loss": 1.8597, - "step": 1588 - }, - { - "epoch": 2.03, - "grad_norm": 0.255859375, - "learning_rate": 4.3263157894736845e-06, - "loss": 1.8732, - "step": 1589 - }, - { - "epoch": 2.03, - "grad_norm": 0.28515625, - "learning_rate": 4.315789473684211e-06, - "loss": 1.8016, - "step": 1590 - }, - { - "epoch": 2.04, - "grad_norm": 0.240234375, - "learning_rate": 4.305263157894737e-06, - "loss": 1.8166, - "step": 1591 - }, - { - "epoch": 2.04, - "grad_norm": 0.2275390625, - "learning_rate": 4.2947368421052635e-06, - "loss": 1.9341, - "step": 1592 - }, - { - "epoch": 2.04, - "grad_norm": 0.2255859375, - "learning_rate": 4.28421052631579e-06, - "loss": 1.8598, - "step": 1593 - }, - { - "epoch": 2.04, - "grad_norm": 0.208984375, - "learning_rate": 4.273684210526316e-06, - "loss": 1.8522, - "step": 1594 - }, - { - "epoch": 2.04, - "grad_norm": 0.2333984375, - "learning_rate": 4.2631578947368425e-06, - "loss": 1.8971, - "step": 1595 - }, - { - "epoch": 2.04, - "grad_norm": 0.3359375, - "learning_rate": 4.252631578947369e-06, - "loss": 1.8506, - "step": 1596 - }, - { - "epoch": 2.04, - "grad_norm": 0.25390625, - "learning_rate": 4.242105263157895e-06, - "loss": 1.961, - "step": 1597 - }, - { - "epoch": 2.04, - "grad_norm": 0.232421875, - "learning_rate": 4.2315789473684215e-06, - "loss": 1.929, - "step": 1598 - }, - { - "epoch": 2.05, - "grad_norm": 0.25, - "learning_rate": 4.221052631578948e-06, - "loss": 1.9705, - "step": 1599 - }, - { - "epoch": 2.05, - "grad_norm": 0.25390625, - "learning_rate": 4.210526315789474e-06, - "loss": 1.8839, - "step": 1600 - }, - { - "epoch": 2.05, - "eval_loss": 1.8947383165359497, - "eval_runtime": 125.9111, - "eval_samples_per_second": 39.711, - "eval_steps_per_second": 1.247, - "step": 1600 - }, - { - "epoch": 2.05, - "grad_norm": 0.244140625, - "learning_rate": 4.2000000000000004e-06, - "loss": 1.9274, - "step": 1601 - }, - { - "epoch": 2.05, - "grad_norm": 0.265625, - "learning_rate": 4.189473684210527e-06, - "loss": 1.9032, - "step": 1602 - }, - { - "epoch": 2.05, - "grad_norm": 0.26171875, - "learning_rate": 4.178947368421053e-06, - "loss": 1.9215, - "step": 1603 - }, - { - "epoch": 2.05, - "grad_norm": 0.2265625, - "learning_rate": 4.1684210526315794e-06, - "loss": 1.9147, - "step": 1604 - }, - { - "epoch": 2.05, - "grad_norm": 0.2373046875, - "learning_rate": 4.157894736842106e-06, - "loss": 1.9413, - "step": 1605 - }, - { - "epoch": 2.06, - "grad_norm": 0.2431640625, - "learning_rate": 4.147368421052632e-06, - "loss": 1.8985, - "step": 1606 - }, - { - "epoch": 2.06, - "grad_norm": 0.26171875, - "learning_rate": 4.136842105263158e-06, - "loss": 1.8152, - "step": 1607 - }, - { - "epoch": 2.06, - "grad_norm": 0.2333984375, - "learning_rate": 4.126315789473685e-06, - "loss": 1.8061, - "step": 1608 - }, - { - "epoch": 2.06, - "grad_norm": 0.232421875, - "learning_rate": 4.115789473684211e-06, - "loss": 1.956, - "step": 1609 - }, - { - "epoch": 2.06, - "grad_norm": 0.232421875, - "learning_rate": 4.105263157894737e-06, - "loss": 1.9495, - "step": 1610 - }, - { - "epoch": 2.06, - "grad_norm": 0.2158203125, - "learning_rate": 4.094736842105264e-06, - "loss": 1.8739, - "step": 1611 - }, - { - "epoch": 2.06, - "grad_norm": 0.2470703125, - "learning_rate": 4.08421052631579e-06, - "loss": 1.9841, - "step": 1612 - }, - { - "epoch": 2.06, - "grad_norm": 0.25390625, - "learning_rate": 4.073684210526316e-06, - "loss": 1.9026, - "step": 1613 - }, - { - "epoch": 2.07, - "grad_norm": 0.21875, - "learning_rate": 4.063157894736842e-06, - "loss": 1.879, - "step": 1614 - }, - { - "epoch": 2.07, - "grad_norm": 0.240234375, - "learning_rate": 4.052631578947368e-06, - "loss": 1.8633, - "step": 1615 - }, - { - "epoch": 2.07, - "grad_norm": 0.2177734375, - "learning_rate": 4.042105263157895e-06, - "loss": 1.8591, - "step": 1616 - }, - { - "epoch": 2.07, - "grad_norm": 0.2314453125, - "learning_rate": 4.031578947368422e-06, - "loss": 2.0179, - "step": 1617 - }, - { - "epoch": 2.07, - "grad_norm": 0.2451171875, - "learning_rate": 4.021052631578948e-06, - "loss": 1.8883, - "step": 1618 - }, - { - "epoch": 2.07, - "grad_norm": 0.2333984375, - "learning_rate": 4.010526315789474e-06, - "loss": 1.925, - "step": 1619 - }, - { - "epoch": 2.07, - "grad_norm": 0.2216796875, - "learning_rate": 4.000000000000001e-06, - "loss": 1.8596, - "step": 1620 - }, - { - "epoch": 2.07, - "grad_norm": 0.2265625, - "learning_rate": 3.989473684210526e-06, - "loss": 1.9106, - "step": 1621 - }, - { - "epoch": 2.08, - "grad_norm": 0.2431640625, - "learning_rate": 3.9789473684210525e-06, - "loss": 1.8703, - "step": 1622 - }, - { - "epoch": 2.08, - "grad_norm": 0.2490234375, - "learning_rate": 3.968421052631579e-06, - "loss": 1.8292, - "step": 1623 - }, - { - "epoch": 2.08, - "grad_norm": 0.26953125, - "learning_rate": 3.957894736842106e-06, - "loss": 1.9856, - "step": 1624 - }, - { - "epoch": 2.08, - "grad_norm": 0.267578125, - "learning_rate": 3.947368421052632e-06, - "loss": 1.8791, - "step": 1625 - }, - { - "epoch": 2.08, - "grad_norm": 0.24609375, - "learning_rate": 3.936842105263159e-06, - "loss": 2.0162, - "step": 1626 - }, - { - "epoch": 2.08, - "grad_norm": 0.22265625, - "learning_rate": 3.926315789473685e-06, - "loss": 1.8753, - "step": 1627 - }, - { - "epoch": 2.08, - "grad_norm": 0.248046875, - "learning_rate": 3.9157894736842104e-06, - "loss": 1.9565, - "step": 1628 - }, - { - "epoch": 2.08, - "grad_norm": 0.251953125, - "learning_rate": 3.905263157894737e-06, - "loss": 1.8649, - "step": 1629 - }, - { - "epoch": 2.09, - "grad_norm": 0.2333984375, - "learning_rate": 3.894736842105263e-06, - "loss": 1.8553, - "step": 1630 - }, - { - "epoch": 2.09, - "grad_norm": 0.228515625, - "learning_rate": 3.884210526315789e-06, - "loss": 1.9002, - "step": 1631 - }, - { - "epoch": 2.09, - "grad_norm": 0.25, - "learning_rate": 3.873684210526316e-06, - "loss": 1.8764, - "step": 1632 - }, - { - "epoch": 2.09, - "grad_norm": 0.23828125, - "learning_rate": 3.863157894736843e-06, - "loss": 1.908, - "step": 1633 - }, - { - "epoch": 2.09, - "grad_norm": 0.291015625, - "learning_rate": 3.852631578947369e-06, - "loss": 1.919, - "step": 1634 - }, - { - "epoch": 2.09, - "grad_norm": 0.267578125, - "learning_rate": 3.842105263157895e-06, - "loss": 1.8424, - "step": 1635 - }, - { - "epoch": 2.09, - "grad_norm": 0.291015625, - "learning_rate": 3.831578947368421e-06, - "loss": 1.904, - "step": 1636 - }, - { - "epoch": 2.09, - "grad_norm": 0.228515625, - "learning_rate": 3.821052631578947e-06, - "loss": 1.791, - "step": 1637 - }, - { - "epoch": 2.1, - "grad_norm": 0.23828125, - "learning_rate": 3.810526315789474e-06, - "loss": 1.8749, - "step": 1638 - }, - { - "epoch": 2.1, - "grad_norm": 0.296875, - "learning_rate": 3.8000000000000005e-06, - "loss": 1.8552, - "step": 1639 - }, - { - "epoch": 2.1, - "grad_norm": 0.2265625, - "learning_rate": 3.789473684210527e-06, - "loss": 1.814, - "step": 1640 - }, - { - "epoch": 2.1, - "grad_norm": 0.2109375, - "learning_rate": 3.778947368421053e-06, - "loss": 1.7941, - "step": 1641 - }, - { - "epoch": 2.1, - "grad_norm": 0.25, - "learning_rate": 3.768421052631579e-06, - "loss": 1.9317, - "step": 1642 - }, - { - "epoch": 2.1, - "grad_norm": 0.236328125, - "learning_rate": 3.7578947368421053e-06, - "loss": 1.8514, - "step": 1643 - }, - { - "epoch": 2.1, - "grad_norm": 0.28125, - "learning_rate": 3.7473684210526317e-06, - "loss": 1.9168, - "step": 1644 - }, - { - "epoch": 2.1, - "grad_norm": 0.25390625, - "learning_rate": 3.736842105263158e-06, - "loss": 1.91, - "step": 1645 - }, - { - "epoch": 2.11, - "grad_norm": 0.21875, - "learning_rate": 3.7263157894736848e-06, - "loss": 1.8779, - "step": 1646 - }, - { - "epoch": 2.11, - "grad_norm": 0.234375, - "learning_rate": 3.715789473684211e-06, - "loss": 1.9659, - "step": 1647 - }, - { - "epoch": 2.11, - "grad_norm": 0.228515625, - "learning_rate": 3.7052631578947374e-06, - "loss": 1.9977, - "step": 1648 - }, - { - "epoch": 2.11, - "grad_norm": 0.2578125, - "learning_rate": 3.6947368421052637e-06, - "loss": 1.9033, - "step": 1649 - }, - { - "epoch": 2.11, - "grad_norm": 0.25390625, - "learning_rate": 3.6842105263157896e-06, - "loss": 1.9265, - "step": 1650 - }, - { - "epoch": 2.11, - "grad_norm": 0.2177734375, - "learning_rate": 3.673684210526316e-06, - "loss": 1.8559, - "step": 1651 - }, - { - "epoch": 2.11, - "grad_norm": 0.240234375, - "learning_rate": 3.6631578947368423e-06, - "loss": 1.8673, - "step": 1652 - }, - { - "epoch": 2.12, - "grad_norm": 0.220703125, - "learning_rate": 3.6526315789473686e-06, - "loss": 1.9034, - "step": 1653 - }, - { - "epoch": 2.12, - "grad_norm": 0.244140625, - "learning_rate": 3.642105263157895e-06, - "loss": 1.9337, - "step": 1654 - }, - { - "epoch": 2.12, - "grad_norm": 0.275390625, - "learning_rate": 3.6315789473684217e-06, - "loss": 1.8757, - "step": 1655 - }, - { - "epoch": 2.12, - "grad_norm": 0.24609375, - "learning_rate": 3.621052631578948e-06, - "loss": 1.9441, - "step": 1656 - }, - { - "epoch": 2.12, - "grad_norm": 0.326171875, - "learning_rate": 3.610526315789474e-06, - "loss": 1.87, - "step": 1657 - }, - { - "epoch": 2.12, - "grad_norm": 0.2255859375, - "learning_rate": 3.6000000000000003e-06, - "loss": 1.8776, - "step": 1658 - }, - { - "epoch": 2.12, - "grad_norm": 0.216796875, - "learning_rate": 3.5894736842105266e-06, - "loss": 1.9709, - "step": 1659 - }, - { - "epoch": 2.12, - "grad_norm": 0.25390625, - "learning_rate": 3.578947368421053e-06, - "loss": 1.9192, - "step": 1660 - }, - { - "epoch": 2.13, - "grad_norm": 0.236328125, - "learning_rate": 3.5684210526315792e-06, - "loss": 1.931, - "step": 1661 - }, - { - "epoch": 2.13, - "grad_norm": 0.2314453125, - "learning_rate": 3.5578947368421056e-06, - "loss": 1.9552, - "step": 1662 - }, - { - "epoch": 2.13, - "grad_norm": 0.275390625, - "learning_rate": 3.5473684210526323e-06, - "loss": 1.7697, - "step": 1663 - }, - { - "epoch": 2.13, - "grad_norm": 0.2197265625, - "learning_rate": 3.536842105263158e-06, - "loss": 1.954, - "step": 1664 - }, - { - "epoch": 2.13, - "grad_norm": 0.228515625, - "learning_rate": 3.5263157894736846e-06, - "loss": 1.8643, - "step": 1665 - }, - { - "epoch": 2.13, - "grad_norm": 0.314453125, - "learning_rate": 3.515789473684211e-06, - "loss": 1.9029, - "step": 1666 - }, - { - "epoch": 2.13, - "grad_norm": 0.26953125, - "learning_rate": 3.505263157894737e-06, - "loss": 1.9261, - "step": 1667 - }, - { - "epoch": 2.13, - "grad_norm": 0.2392578125, - "learning_rate": 3.4947368421052635e-06, - "loss": 1.973, - "step": 1668 - }, - { - "epoch": 2.14, - "grad_norm": 0.2734375, - "learning_rate": 3.48421052631579e-06, - "loss": 1.8942, - "step": 1669 - }, - { - "epoch": 2.14, - "grad_norm": 0.2275390625, - "learning_rate": 3.473684210526316e-06, - "loss": 1.8996, - "step": 1670 - }, - { - "epoch": 2.14, - "grad_norm": 0.234375, - "learning_rate": 3.463157894736842e-06, - "loss": 1.9037, - "step": 1671 - }, - { - "epoch": 2.14, - "grad_norm": 0.2109375, - "learning_rate": 3.4526315789473684e-06, - "loss": 1.9525, - "step": 1672 - }, - { - "epoch": 2.14, - "grad_norm": 0.2265625, - "learning_rate": 3.4421052631578947e-06, - "loss": 1.854, - "step": 1673 - }, - { - "epoch": 2.14, - "grad_norm": 0.220703125, - "learning_rate": 3.4315789473684215e-06, - "loss": 1.9021, - "step": 1674 - }, - { - "epoch": 2.14, - "grad_norm": 0.271484375, - "learning_rate": 3.421052631578948e-06, - "loss": 1.9239, - "step": 1675 - }, - { - "epoch": 2.14, - "grad_norm": 0.216796875, - "learning_rate": 3.410526315789474e-06, - "loss": 1.8052, - "step": 1676 - }, - { - "epoch": 2.15, - "grad_norm": 0.2412109375, - "learning_rate": 3.4000000000000005e-06, - "loss": 1.8088, - "step": 1677 - }, - { - "epoch": 2.15, - "grad_norm": 0.23046875, - "learning_rate": 3.3894736842105264e-06, - "loss": 1.938, - "step": 1678 - }, - { - "epoch": 2.15, - "grad_norm": 0.2333984375, - "learning_rate": 3.3789473684210527e-06, - "loss": 1.8901, - "step": 1679 - }, - { - "epoch": 2.15, - "grad_norm": 0.236328125, - "learning_rate": 3.368421052631579e-06, - "loss": 1.7886, - "step": 1680 - }, - { - "epoch": 2.15, - "grad_norm": 0.2734375, - "learning_rate": 3.3578947368421054e-06, - "loss": 1.886, - "step": 1681 - }, - { - "epoch": 2.15, - "grad_norm": 0.2421875, - "learning_rate": 3.347368421052632e-06, - "loss": 1.8551, - "step": 1682 - }, - { - "epoch": 2.15, - "grad_norm": 0.2197265625, - "learning_rate": 3.3368421052631584e-06, - "loss": 1.8313, - "step": 1683 - }, - { - "epoch": 2.15, - "grad_norm": 0.298828125, - "learning_rate": 3.3263157894736848e-06, - "loss": 1.921, - "step": 1684 - }, - { - "epoch": 2.16, - "grad_norm": 0.234375, - "learning_rate": 3.3157894736842107e-06, - "loss": 1.8654, - "step": 1685 - }, - { - "epoch": 2.16, - "grad_norm": 0.271484375, - "learning_rate": 3.305263157894737e-06, - "loss": 1.8454, - "step": 1686 - }, - { - "epoch": 2.16, - "grad_norm": 0.2578125, - "learning_rate": 3.2947368421052633e-06, - "loss": 1.8604, - "step": 1687 - }, - { - "epoch": 2.16, - "grad_norm": 0.25390625, - "learning_rate": 3.2842105263157897e-06, - "loss": 1.8552, - "step": 1688 - }, - { - "epoch": 2.16, - "grad_norm": 0.22265625, - "learning_rate": 3.273684210526316e-06, - "loss": 1.9837, - "step": 1689 - }, - { - "epoch": 2.16, - "grad_norm": 0.240234375, - "learning_rate": 3.2631578947368423e-06, - "loss": 1.9915, - "step": 1690 - }, - { - "epoch": 2.16, - "grad_norm": 0.263671875, - "learning_rate": 3.252631578947369e-06, - "loss": 1.8385, - "step": 1691 - }, - { - "epoch": 2.17, - "grad_norm": 0.2392578125, - "learning_rate": 3.2421052631578945e-06, - "loss": 1.8499, - "step": 1692 - }, - { - "epoch": 2.17, - "grad_norm": 0.248046875, - "learning_rate": 3.2315789473684213e-06, - "loss": 1.9221, - "step": 1693 - }, - { - "epoch": 2.17, - "grad_norm": 0.291015625, - "learning_rate": 3.2210526315789476e-06, - "loss": 1.8883, - "step": 1694 - }, - { - "epoch": 2.17, - "grad_norm": 0.396484375, - "learning_rate": 3.210526315789474e-06, - "loss": 1.8816, - "step": 1695 - }, - { - "epoch": 2.17, - "grad_norm": 0.2392578125, - "learning_rate": 3.2000000000000003e-06, - "loss": 1.8813, - "step": 1696 - }, - { - "epoch": 2.17, - "grad_norm": 0.2421875, - "learning_rate": 3.1894736842105266e-06, - "loss": 1.9111, - "step": 1697 - }, - { - "epoch": 2.17, - "grad_norm": 0.24609375, - "learning_rate": 3.178947368421053e-06, - "loss": 1.8877, - "step": 1698 - }, - { - "epoch": 2.17, - "grad_norm": 0.22265625, - "learning_rate": 3.168421052631579e-06, - "loss": 1.8751, - "step": 1699 - }, - { - "epoch": 2.18, - "grad_norm": 0.232421875, - "learning_rate": 3.157894736842105e-06, - "loss": 1.9411, - "step": 1700 - }, - { - "epoch": 2.18, - "grad_norm": 0.2216796875, - "learning_rate": 3.147368421052632e-06, - "loss": 1.8908, - "step": 1701 - }, - { - "epoch": 2.18, - "grad_norm": 0.275390625, - "learning_rate": 3.1368421052631582e-06, - "loss": 1.9181, - "step": 1702 - }, - { - "epoch": 2.18, - "grad_norm": 0.27734375, - "learning_rate": 3.1263157894736846e-06, - "loss": 1.8834, - "step": 1703 - }, - { - "epoch": 2.18, - "grad_norm": 0.2412109375, - "learning_rate": 3.115789473684211e-06, - "loss": 1.9282, - "step": 1704 - }, - { - "epoch": 2.18, - "grad_norm": 0.24609375, - "learning_rate": 3.1052631578947372e-06, - "loss": 1.8771, - "step": 1705 - }, - { - "epoch": 2.18, - "grad_norm": 0.2431640625, - "learning_rate": 3.094736842105263e-06, - "loss": 1.8431, - "step": 1706 - }, - { - "epoch": 2.18, - "grad_norm": 0.23046875, - "learning_rate": 3.0842105263157895e-06, - "loss": 1.873, - "step": 1707 - }, - { - "epoch": 2.19, - "grad_norm": 0.2490234375, - "learning_rate": 3.0736842105263158e-06, - "loss": 1.9936, - "step": 1708 - }, - { - "epoch": 2.19, - "grad_norm": 0.263671875, - "learning_rate": 3.0631578947368425e-06, - "loss": 1.9106, - "step": 1709 - }, - { - "epoch": 2.19, - "grad_norm": 0.25, - "learning_rate": 3.052631578947369e-06, - "loss": 1.9725, - "step": 1710 - }, - { - "epoch": 2.19, - "grad_norm": 0.2119140625, - "learning_rate": 3.042105263157895e-06, - "loss": 1.9069, - "step": 1711 - }, - { - "epoch": 2.19, - "grad_norm": 0.2470703125, - "learning_rate": 3.0315789473684215e-06, - "loss": 1.866, - "step": 1712 - }, - { - "epoch": 2.19, - "grad_norm": 0.251953125, - "learning_rate": 3.0210526315789474e-06, - "loss": 1.8796, - "step": 1713 - }, - { - "epoch": 2.19, - "grad_norm": 0.228515625, - "learning_rate": 3.0105263157894737e-06, - "loss": 1.9112, - "step": 1714 - }, - { - "epoch": 2.19, - "grad_norm": 0.2451171875, - "learning_rate": 3e-06, - "loss": 1.9397, - "step": 1715 - }, - { - "epoch": 2.2, - "grad_norm": 0.2373046875, - "learning_rate": 2.9894736842105264e-06, - "loss": 1.8008, - "step": 1716 - }, - { - "epoch": 2.2, - "grad_norm": 0.234375, - "learning_rate": 2.9789473684210527e-06, - "loss": 1.9029, - "step": 1717 - }, - { - "epoch": 2.2, - "grad_norm": 0.248046875, - "learning_rate": 2.9684210526315795e-06, - "loss": 1.9538, - "step": 1718 - }, - { - "epoch": 2.2, - "grad_norm": 0.248046875, - "learning_rate": 2.957894736842106e-06, - "loss": 1.9964, - "step": 1719 - }, - { - "epoch": 2.2, - "grad_norm": 0.224609375, - "learning_rate": 2.9473684210526317e-06, - "loss": 1.8523, - "step": 1720 - }, - { - "epoch": 2.2, - "grad_norm": 0.271484375, - "learning_rate": 2.936842105263158e-06, - "loss": 1.842, - "step": 1721 - }, - { - "epoch": 2.2, - "grad_norm": 0.228515625, - "learning_rate": 2.9263157894736844e-06, - "loss": 1.8715, - "step": 1722 - }, - { - "epoch": 2.2, - "grad_norm": 0.32421875, - "learning_rate": 2.9157894736842107e-06, - "loss": 1.9104, - "step": 1723 - }, - { - "epoch": 2.21, - "grad_norm": 0.2392578125, - "learning_rate": 2.905263157894737e-06, - "loss": 1.9828, - "step": 1724 - }, - { - "epoch": 2.21, - "grad_norm": 0.275390625, - "learning_rate": 2.8947368421052634e-06, - "loss": 1.9064, - "step": 1725 - }, - { - "epoch": 2.21, - "grad_norm": 0.236328125, - "learning_rate": 2.88421052631579e-06, - "loss": 1.8761, - "step": 1726 - }, - { - "epoch": 2.21, - "grad_norm": 0.20703125, - "learning_rate": 2.8736842105263164e-06, - "loss": 1.9185, - "step": 1727 - }, - { - "epoch": 2.21, - "grad_norm": 0.255859375, - "learning_rate": 2.8631578947368423e-06, - "loss": 1.9253, - "step": 1728 - }, - { - "epoch": 2.21, - "grad_norm": 0.26171875, - "learning_rate": 2.8526315789473687e-06, - "loss": 1.9107, - "step": 1729 - }, - { - "epoch": 2.21, - "grad_norm": 0.263671875, - "learning_rate": 2.842105263157895e-06, - "loss": 1.8217, - "step": 1730 - }, - { - "epoch": 2.21, - "grad_norm": 0.251953125, - "learning_rate": 2.8315789473684213e-06, - "loss": 1.9444, - "step": 1731 - }, - { - "epoch": 2.22, - "grad_norm": 0.234375, - "learning_rate": 2.8210526315789476e-06, - "loss": 1.8555, - "step": 1732 - }, - { - "epoch": 2.22, - "grad_norm": 0.2099609375, - "learning_rate": 2.810526315789474e-06, - "loss": 1.9316, - "step": 1733 - }, - { - "epoch": 2.22, - "grad_norm": 0.240234375, - "learning_rate": 2.8000000000000003e-06, - "loss": 1.9639, - "step": 1734 - }, - { - "epoch": 2.22, - "grad_norm": 0.232421875, - "learning_rate": 2.789473684210526e-06, - "loss": 1.891, - "step": 1735 - }, - { - "epoch": 2.22, - "grad_norm": 0.279296875, - "learning_rate": 2.7789473684210525e-06, - "loss": 1.8641, - "step": 1736 - }, - { - "epoch": 2.22, - "grad_norm": 0.2294921875, - "learning_rate": 2.7684210526315793e-06, - "loss": 1.9369, - "step": 1737 - }, - { - "epoch": 2.22, - "grad_norm": 0.2451171875, - "learning_rate": 2.7578947368421056e-06, - "loss": 1.9439, - "step": 1738 - }, - { - "epoch": 2.23, - "grad_norm": 0.23828125, - "learning_rate": 2.747368421052632e-06, - "loss": 1.8613, - "step": 1739 - }, - { - "epoch": 2.23, - "grad_norm": 0.21875, - "learning_rate": 2.7368421052631583e-06, - "loss": 1.9376, - "step": 1740 - }, - { - "epoch": 2.23, - "grad_norm": 0.2236328125, - "learning_rate": 2.7263157894736846e-06, - "loss": 1.9027, - "step": 1741 - }, - { - "epoch": 2.23, - "grad_norm": 0.2890625, - "learning_rate": 2.7157894736842105e-06, - "loss": 1.8755, - "step": 1742 - }, - { - "epoch": 2.23, - "grad_norm": 0.265625, - "learning_rate": 2.705263157894737e-06, - "loss": 1.8704, - "step": 1743 - }, - { - "epoch": 2.23, - "grad_norm": 0.25390625, - "learning_rate": 2.694736842105263e-06, - "loss": 1.9364, - "step": 1744 - }, - { - "epoch": 2.23, - "grad_norm": 0.283203125, - "learning_rate": 2.68421052631579e-06, - "loss": 1.9381, - "step": 1745 - }, - { - "epoch": 2.23, - "grad_norm": 0.232421875, - "learning_rate": 2.6736842105263162e-06, - "loss": 1.8216, - "step": 1746 - }, - { - "epoch": 2.24, - "grad_norm": 0.2060546875, - "learning_rate": 2.6631578947368426e-06, - "loss": 1.8375, - "step": 1747 - }, - { - "epoch": 2.24, - "grad_norm": 0.271484375, - "learning_rate": 2.652631578947369e-06, - "loss": 1.9624, - "step": 1748 - }, - { - "epoch": 2.24, - "grad_norm": 0.240234375, - "learning_rate": 2.6421052631578948e-06, - "loss": 1.9972, - "step": 1749 - }, - { - "epoch": 2.24, - "grad_norm": 0.2294921875, - "learning_rate": 2.631578947368421e-06, - "loss": 1.8824, - "step": 1750 - }, - { - "epoch": 2.24, - "grad_norm": 0.267578125, - "learning_rate": 2.6210526315789474e-06, - "loss": 1.9784, - "step": 1751 - }, - { - "epoch": 2.24, - "grad_norm": 0.2578125, - "learning_rate": 2.6105263157894738e-06, - "loss": 1.8906, - "step": 1752 - }, - { - "epoch": 2.24, - "grad_norm": 0.25390625, - "learning_rate": 2.6e-06, - "loss": 1.9021, - "step": 1753 - }, - { - "epoch": 2.24, - "grad_norm": 0.2412109375, - "learning_rate": 2.589473684210527e-06, - "loss": 1.8927, - "step": 1754 - }, - { - "epoch": 2.25, - "grad_norm": 0.2451171875, - "learning_rate": 2.578947368421053e-06, - "loss": 1.9101, - "step": 1755 - }, - { - "epoch": 2.25, - "grad_norm": 0.2314453125, - "learning_rate": 2.568421052631579e-06, - "loss": 1.8883, - "step": 1756 - }, - { - "epoch": 2.25, - "grad_norm": 0.2158203125, - "learning_rate": 2.5578947368421054e-06, - "loss": 1.8871, - "step": 1757 - }, - { - "epoch": 2.25, - "grad_norm": 0.23046875, - "learning_rate": 2.5473684210526317e-06, - "loss": 1.95, - "step": 1758 - }, - { - "epoch": 2.25, - "grad_norm": 0.23828125, - "learning_rate": 2.536842105263158e-06, - "loss": 1.873, - "step": 1759 - }, - { - "epoch": 2.25, - "grad_norm": 0.25, - "learning_rate": 2.5263157894736844e-06, - "loss": 1.9696, - "step": 1760 - }, - { - "epoch": 2.25, - "grad_norm": 0.21484375, - "learning_rate": 2.5157894736842107e-06, - "loss": 1.8729, - "step": 1761 - }, - { - "epoch": 2.25, - "grad_norm": 0.2490234375, - "learning_rate": 2.5052631578947375e-06, - "loss": 1.802, - "step": 1762 - }, - { - "epoch": 2.26, - "grad_norm": 0.2392578125, - "learning_rate": 2.4947368421052634e-06, - "loss": 1.9309, - "step": 1763 - }, - { - "epoch": 2.26, - "grad_norm": 0.2392578125, - "learning_rate": 2.4842105263157897e-06, - "loss": 1.9498, - "step": 1764 - }, - { - "epoch": 2.26, - "grad_norm": 0.25, - "learning_rate": 2.473684210526316e-06, - "loss": 1.9004, - "step": 1765 - }, - { - "epoch": 2.26, - "grad_norm": 0.3359375, - "learning_rate": 2.4631578947368424e-06, - "loss": 1.8872, - "step": 1766 - }, - { - "epoch": 2.26, - "grad_norm": 0.271484375, - "learning_rate": 2.4526315789473687e-06, - "loss": 1.88, - "step": 1767 - }, - { - "epoch": 2.26, - "grad_norm": 0.267578125, - "learning_rate": 2.442105263157895e-06, - "loss": 1.8795, - "step": 1768 - }, - { - "epoch": 2.26, - "grad_norm": 0.236328125, - "learning_rate": 2.4315789473684213e-06, - "loss": 1.9139, - "step": 1769 - }, - { - "epoch": 2.26, - "grad_norm": 0.3046875, - "learning_rate": 2.4210526315789477e-06, - "loss": 1.9745, - "step": 1770 - }, - { - "epoch": 2.27, - "grad_norm": 0.26171875, - "learning_rate": 2.410526315789474e-06, - "loss": 1.8616, - "step": 1771 - }, - { - "epoch": 2.27, - "grad_norm": 0.224609375, - "learning_rate": 2.4000000000000003e-06, - "loss": 1.9393, - "step": 1772 - }, - { - "epoch": 2.27, - "grad_norm": 0.2470703125, - "learning_rate": 2.3894736842105266e-06, - "loss": 1.906, - "step": 1773 - }, - { - "epoch": 2.27, - "grad_norm": 0.224609375, - "learning_rate": 2.378947368421053e-06, - "loss": 1.9045, - "step": 1774 - }, - { - "epoch": 2.27, - "grad_norm": 0.27734375, - "learning_rate": 2.368421052631579e-06, - "loss": 1.8929, - "step": 1775 - }, - { - "epoch": 2.27, - "grad_norm": 0.2197265625, - "learning_rate": 2.357894736842105e-06, - "loss": 1.9025, - "step": 1776 - }, - { - "epoch": 2.27, - "grad_norm": 0.291015625, - "learning_rate": 2.347368421052632e-06, - "loss": 1.9627, - "step": 1777 - }, - { - "epoch": 2.28, - "grad_norm": 0.2451171875, - "learning_rate": 2.3368421052631583e-06, - "loss": 1.8225, - "step": 1778 - }, - { - "epoch": 2.28, - "grad_norm": 0.2490234375, - "learning_rate": 2.326315789473684e-06, - "loss": 1.8182, - "step": 1779 - }, - { - "epoch": 2.28, - "grad_norm": 0.24609375, - "learning_rate": 2.3157894736842105e-06, - "loss": 1.9834, - "step": 1780 - }, - { - "epoch": 2.28, - "grad_norm": 0.259765625, - "learning_rate": 2.3052631578947373e-06, - "loss": 2.0353, - "step": 1781 - }, - { - "epoch": 2.28, - "grad_norm": 0.255859375, - "learning_rate": 2.294736842105263e-06, - "loss": 1.7824, - "step": 1782 - }, - { - "epoch": 2.28, - "grad_norm": 0.2080078125, - "learning_rate": 2.2842105263157895e-06, - "loss": 1.8704, - "step": 1783 - }, - { - "epoch": 2.28, - "grad_norm": 0.3046875, - "learning_rate": 2.273684210526316e-06, - "loss": 1.9283, - "step": 1784 - }, - { - "epoch": 2.28, - "grad_norm": 0.2275390625, - "learning_rate": 2.2631578947368426e-06, - "loss": 1.9302, - "step": 1785 - }, - { - "epoch": 2.29, - "grad_norm": 0.2216796875, - "learning_rate": 2.2526315789473685e-06, - "loss": 1.9144, - "step": 1786 - }, - { - "epoch": 2.29, - "grad_norm": 0.244140625, - "learning_rate": 2.242105263157895e-06, - "loss": 1.9785, - "step": 1787 - }, - { - "epoch": 2.29, - "grad_norm": 0.2470703125, - "learning_rate": 2.231578947368421e-06, - "loss": 1.865, - "step": 1788 - }, - { - "epoch": 2.29, - "grad_norm": 0.2578125, - "learning_rate": 2.221052631578948e-06, - "loss": 1.809, - "step": 1789 - }, - { - "epoch": 2.29, - "grad_norm": 0.24609375, - "learning_rate": 2.2105263157894738e-06, - "loss": 1.9095, - "step": 1790 - }, - { - "epoch": 2.29, - "grad_norm": 0.2412109375, - "learning_rate": 2.2e-06, - "loss": 1.8343, - "step": 1791 - }, - { - "epoch": 2.29, - "grad_norm": 0.208984375, - "learning_rate": 2.1894736842105264e-06, - "loss": 1.8618, - "step": 1792 - }, - { - "epoch": 2.29, - "grad_norm": 0.2119140625, - "learning_rate": 2.1789473684210528e-06, - "loss": 1.9106, - "step": 1793 - }, - { - "epoch": 2.3, - "grad_norm": 0.2265625, - "learning_rate": 2.168421052631579e-06, - "loss": 1.9487, - "step": 1794 - }, - { - "epoch": 2.3, - "grad_norm": 0.2353515625, - "learning_rate": 2.1578947368421054e-06, - "loss": 1.9059, - "step": 1795 - }, - { - "epoch": 2.3, - "grad_norm": 0.2265625, - "learning_rate": 2.1473684210526317e-06, - "loss": 1.9161, - "step": 1796 - }, - { - "epoch": 2.3, - "grad_norm": 0.2373046875, - "learning_rate": 2.136842105263158e-06, - "loss": 1.9001, - "step": 1797 - }, - { - "epoch": 2.3, - "grad_norm": 0.251953125, - "learning_rate": 2.1263157894736844e-06, - "loss": 1.8745, - "step": 1798 - }, - { - "epoch": 2.3, - "grad_norm": 0.2119140625, - "learning_rate": 2.1157894736842107e-06, - "loss": 1.8809, - "step": 1799 - }, - { - "epoch": 2.3, - "grad_norm": 0.30859375, - "learning_rate": 2.105263157894737e-06, - "loss": 1.9147, - "step": 1800 - }, - { - "epoch": 2.3, - "eval_loss": 1.8946523666381836, - "eval_runtime": 124.7529, - "eval_samples_per_second": 40.079, - "eval_steps_per_second": 1.258, - "step": 1800 - }, - { - "epoch": 2.3, - "grad_norm": 0.2275390625, - "learning_rate": 2.0947368421052634e-06, - "loss": 1.8516, - "step": 1801 - }, - { - "epoch": 2.31, - "grad_norm": 0.251953125, - "learning_rate": 2.0842105263157897e-06, - "loss": 1.9213, - "step": 1802 - }, - { - "epoch": 2.31, - "grad_norm": 0.2333984375, - "learning_rate": 2.073684210526316e-06, - "loss": 1.9267, - "step": 1803 - }, - { - "epoch": 2.31, - "grad_norm": 0.251953125, - "learning_rate": 2.0631578947368424e-06, - "loss": 1.956, - "step": 1804 - }, - { - "epoch": 2.31, - "grad_norm": 0.2041015625, - "learning_rate": 2.0526315789473687e-06, - "loss": 1.899, - "step": 1805 - }, - { - "epoch": 2.31, - "grad_norm": 0.21875, - "learning_rate": 2.042105263157895e-06, - "loss": 1.9111, - "step": 1806 - }, - { - "epoch": 2.31, - "grad_norm": 0.2451171875, - "learning_rate": 2.031578947368421e-06, - "loss": 1.8342, - "step": 1807 - }, - { - "epoch": 2.31, - "grad_norm": 0.216796875, - "learning_rate": 2.0210526315789477e-06, - "loss": 1.9057, - "step": 1808 - }, - { - "epoch": 2.31, - "grad_norm": 0.2373046875, - "learning_rate": 2.010526315789474e-06, - "loss": 1.9889, - "step": 1809 - }, - { - "epoch": 2.32, - "grad_norm": 0.2294921875, - "learning_rate": 2.0000000000000003e-06, - "loss": 1.9708, - "step": 1810 - }, - { - "epoch": 2.32, - "grad_norm": 0.2060546875, - "learning_rate": 1.9894736842105262e-06, - "loss": 1.8917, - "step": 1811 - }, - { - "epoch": 2.32, - "grad_norm": 0.2265625, - "learning_rate": 1.978947368421053e-06, - "loss": 1.8532, - "step": 1812 - }, - { - "epoch": 2.32, - "grad_norm": 0.2333984375, - "learning_rate": 1.9684210526315793e-06, - "loss": 1.8964, - "step": 1813 - }, - { - "epoch": 2.32, - "grad_norm": 0.25390625, - "learning_rate": 1.9578947368421052e-06, - "loss": 1.8416, - "step": 1814 - }, - { - "epoch": 2.32, - "grad_norm": 0.220703125, - "learning_rate": 1.9473684210526315e-06, - "loss": 1.9684, - "step": 1815 - }, - { - "epoch": 2.32, - "grad_norm": 0.255859375, - "learning_rate": 1.936842105263158e-06, - "loss": 1.8919, - "step": 1816 - }, - { - "epoch": 2.33, - "grad_norm": 0.2197265625, - "learning_rate": 1.9263157894736846e-06, - "loss": 1.9509, - "step": 1817 - }, - { - "epoch": 2.33, - "grad_norm": 0.205078125, - "learning_rate": 1.9157894736842105e-06, - "loss": 1.9682, - "step": 1818 - }, - { - "epoch": 2.33, - "grad_norm": 0.23046875, - "learning_rate": 1.905263157894737e-06, - "loss": 1.9549, - "step": 1819 - }, - { - "epoch": 2.33, - "grad_norm": 0.23828125, - "learning_rate": 1.8947368421052634e-06, - "loss": 1.9546, - "step": 1820 - }, - { - "epoch": 2.33, - "grad_norm": 0.259765625, - "learning_rate": 1.8842105263157895e-06, - "loss": 1.9972, - "step": 1821 - }, - { - "epoch": 2.33, - "grad_norm": 0.31640625, - "learning_rate": 1.8736842105263158e-06, - "loss": 2.0368, - "step": 1822 - }, - { - "epoch": 2.33, - "grad_norm": 0.2490234375, - "learning_rate": 1.8631578947368424e-06, - "loss": 1.9936, - "step": 1823 - }, - { - "epoch": 2.33, - "grad_norm": 0.2373046875, - "learning_rate": 1.8526315789473687e-06, - "loss": 1.8637, - "step": 1824 - }, - { - "epoch": 2.34, - "grad_norm": 0.240234375, - "learning_rate": 1.8421052631578948e-06, - "loss": 1.9942, - "step": 1825 - }, - { - "epoch": 2.34, - "grad_norm": 0.2392578125, - "learning_rate": 1.8315789473684211e-06, - "loss": 1.8606, - "step": 1826 - }, - { - "epoch": 2.34, - "grad_norm": 0.294921875, - "learning_rate": 1.8210526315789475e-06, - "loss": 1.8774, - "step": 1827 - }, - { - "epoch": 2.34, - "grad_norm": 0.26953125, - "learning_rate": 1.810526315789474e-06, - "loss": 1.8735, - "step": 1828 - }, - { - "epoch": 2.34, - "grad_norm": 0.2255859375, - "learning_rate": 1.8000000000000001e-06, - "loss": 1.973, - "step": 1829 - }, - { - "epoch": 2.34, - "grad_norm": 0.234375, - "learning_rate": 1.7894736842105265e-06, - "loss": 1.8591, - "step": 1830 - }, - { - "epoch": 2.34, - "grad_norm": 0.240234375, - "learning_rate": 1.7789473684210528e-06, - "loss": 1.9256, - "step": 1831 - }, - { - "epoch": 2.34, - "grad_norm": 0.21484375, - "learning_rate": 1.768421052631579e-06, - "loss": 1.8957, - "step": 1832 - }, - { - "epoch": 2.35, - "grad_norm": 0.2109375, - "learning_rate": 1.7578947368421054e-06, - "loss": 1.8836, - "step": 1833 - }, - { - "epoch": 2.35, - "grad_norm": 0.2109375, - "learning_rate": 1.7473684210526318e-06, - "loss": 1.9167, - "step": 1834 - }, - { - "epoch": 2.35, - "grad_norm": 0.203125, - "learning_rate": 1.736842105263158e-06, - "loss": 1.9031, - "step": 1835 - }, - { - "epoch": 2.35, - "grad_norm": 0.283203125, - "learning_rate": 1.7263157894736842e-06, - "loss": 1.9428, - "step": 1836 - }, - { - "epoch": 2.35, - "grad_norm": 0.2255859375, - "learning_rate": 1.7157894736842107e-06, - "loss": 1.9509, - "step": 1837 - }, - { - "epoch": 2.35, - "grad_norm": 0.255859375, - "learning_rate": 1.705263157894737e-06, - "loss": 1.7711, - "step": 1838 - }, - { - "epoch": 2.35, - "grad_norm": 0.251953125, - "learning_rate": 1.6947368421052632e-06, - "loss": 1.9539, - "step": 1839 - }, - { - "epoch": 2.35, - "grad_norm": 0.22265625, - "learning_rate": 1.6842105263157895e-06, - "loss": 1.9614, - "step": 1840 - }, - { - "epoch": 2.36, - "grad_norm": 0.2080078125, - "learning_rate": 1.673684210526316e-06, - "loss": 1.8786, - "step": 1841 - }, - { - "epoch": 2.36, - "grad_norm": 0.2041015625, - "learning_rate": 1.6631578947368424e-06, - "loss": 1.9353, - "step": 1842 - }, - { - "epoch": 2.36, - "grad_norm": 0.2216796875, - "learning_rate": 1.6526315789473685e-06, - "loss": 1.8319, - "step": 1843 - }, - { - "epoch": 2.36, - "grad_norm": 0.255859375, - "learning_rate": 1.6421052631578948e-06, - "loss": 1.9028, - "step": 1844 - }, - { - "epoch": 2.36, - "grad_norm": 0.2109375, - "learning_rate": 1.6315789473684212e-06, - "loss": 1.9091, - "step": 1845 - }, - { - "epoch": 2.36, - "grad_norm": 0.240234375, - "learning_rate": 1.6210526315789473e-06, - "loss": 1.9672, - "step": 1846 - }, - { - "epoch": 2.36, - "grad_norm": 0.21484375, - "learning_rate": 1.6105263157894738e-06, - "loss": 1.9254, - "step": 1847 - }, - { - "epoch": 2.36, - "grad_norm": 0.2373046875, - "learning_rate": 1.6000000000000001e-06, - "loss": 1.8792, - "step": 1848 - }, - { - "epoch": 2.37, - "grad_norm": 0.216796875, - "learning_rate": 1.5894736842105265e-06, - "loss": 1.9297, - "step": 1849 - }, - { - "epoch": 2.37, - "grad_norm": 0.2294921875, - "learning_rate": 1.5789473684210526e-06, - "loss": 1.9312, - "step": 1850 - }, - { - "epoch": 2.37, - "grad_norm": 0.2275390625, - "learning_rate": 1.5684210526315791e-06, - "loss": 1.9754, - "step": 1851 - }, - { - "epoch": 2.37, - "grad_norm": 0.2392578125, - "learning_rate": 1.5578947368421054e-06, - "loss": 1.9173, - "step": 1852 - }, - { - "epoch": 2.37, - "grad_norm": 0.2734375, - "learning_rate": 1.5473684210526316e-06, - "loss": 1.893, - "step": 1853 - }, - { - "epoch": 2.37, - "grad_norm": 0.302734375, - "learning_rate": 1.5368421052631579e-06, - "loss": 1.9189, - "step": 1854 - }, - { - "epoch": 2.37, - "grad_norm": 0.2236328125, - "learning_rate": 1.5263157894736844e-06, - "loss": 1.8455, - "step": 1855 - }, - { - "epoch": 2.37, - "grad_norm": 0.2333984375, - "learning_rate": 1.5157894736842108e-06, - "loss": 1.9934, - "step": 1856 - }, - { - "epoch": 2.38, - "grad_norm": 0.2578125, - "learning_rate": 1.5052631578947369e-06, - "loss": 1.964, - "step": 1857 - }, - { - "epoch": 2.38, - "grad_norm": 0.2197265625, - "learning_rate": 1.4947368421052632e-06, - "loss": 1.8733, - "step": 1858 - }, - { - "epoch": 2.38, - "grad_norm": 0.2177734375, - "learning_rate": 1.4842105263157897e-06, - "loss": 1.8378, - "step": 1859 - }, - { - "epoch": 2.38, - "grad_norm": 0.21875, - "learning_rate": 1.4736842105263159e-06, - "loss": 1.9271, - "step": 1860 - }, - { - "epoch": 2.38, - "grad_norm": 0.2275390625, - "learning_rate": 1.4631578947368422e-06, - "loss": 1.9194, - "step": 1861 - }, - { - "epoch": 2.38, - "grad_norm": 0.2158203125, - "learning_rate": 1.4526315789473685e-06, - "loss": 1.9029, - "step": 1862 - }, - { - "epoch": 2.38, - "grad_norm": 0.216796875, - "learning_rate": 1.442105263157895e-06, - "loss": 1.905, - "step": 1863 - }, - { - "epoch": 2.39, - "grad_norm": 0.2275390625, - "learning_rate": 1.4315789473684212e-06, - "loss": 1.8611, - "step": 1864 - }, - { - "epoch": 2.39, - "grad_norm": 0.2197265625, - "learning_rate": 1.4210526315789475e-06, - "loss": 1.8994, - "step": 1865 - }, - { - "epoch": 2.39, - "grad_norm": 0.2021484375, - "learning_rate": 1.4105263157894738e-06, - "loss": 1.9729, - "step": 1866 - }, - { - "epoch": 2.39, - "grad_norm": 0.2216796875, - "learning_rate": 1.4000000000000001e-06, - "loss": 1.9565, - "step": 1867 - }, - { - "epoch": 2.39, - "grad_norm": 0.251953125, - "learning_rate": 1.3894736842105263e-06, - "loss": 1.8195, - "step": 1868 - }, - { - "epoch": 2.39, - "grad_norm": 0.251953125, - "learning_rate": 1.3789473684210528e-06, - "loss": 1.9657, - "step": 1869 - }, - { - "epoch": 2.39, - "grad_norm": 0.263671875, - "learning_rate": 1.3684210526315791e-06, - "loss": 1.9753, - "step": 1870 - }, - { - "epoch": 2.39, - "grad_norm": 0.271484375, - "learning_rate": 1.3578947368421052e-06, - "loss": 1.7857, - "step": 1871 - }, - { - "epoch": 2.4, - "grad_norm": 0.25390625, - "learning_rate": 1.3473684210526316e-06, - "loss": 1.8216, - "step": 1872 - }, - { - "epoch": 2.4, - "grad_norm": 0.2333984375, - "learning_rate": 1.3368421052631581e-06, - "loss": 1.9137, - "step": 1873 - }, - { - "epoch": 2.4, - "grad_norm": 0.251953125, - "learning_rate": 1.3263157894736844e-06, - "loss": 1.9499, - "step": 1874 - }, - { - "epoch": 2.4, - "grad_norm": 0.2734375, - "learning_rate": 1.3157894736842106e-06, - "loss": 1.9411, - "step": 1875 - }, - { - "epoch": 2.4, - "grad_norm": 0.2255859375, - "learning_rate": 1.3052631578947369e-06, - "loss": 1.8485, - "step": 1876 - }, - { - "epoch": 2.4, - "grad_norm": 0.2890625, - "learning_rate": 1.2947368421052634e-06, - "loss": 1.8595, - "step": 1877 - }, - { - "epoch": 2.4, - "grad_norm": 0.259765625, - "learning_rate": 1.2842105263157895e-06, - "loss": 1.8305, - "step": 1878 - }, - { - "epoch": 2.4, - "grad_norm": 0.2421875, - "learning_rate": 1.2736842105263159e-06, - "loss": 1.8645, - "step": 1879 - }, - { - "epoch": 2.41, - "grad_norm": 0.2431640625, - "learning_rate": 1.2631578947368422e-06, - "loss": 1.9243, - "step": 1880 - }, - { - "epoch": 2.41, - "grad_norm": 0.2119140625, - "learning_rate": 1.2526315789473687e-06, - "loss": 1.9302, - "step": 1881 - }, - { - "epoch": 2.41, - "grad_norm": 0.2578125, - "learning_rate": 1.2421052631578948e-06, - "loss": 1.8223, - "step": 1882 - }, - { - "epoch": 2.41, - "grad_norm": 0.2333984375, - "learning_rate": 1.2315789473684212e-06, - "loss": 1.8794, - "step": 1883 - }, - { - "epoch": 2.41, - "grad_norm": 0.2197265625, - "learning_rate": 1.2210526315789475e-06, - "loss": 1.8524, - "step": 1884 - }, - { - "epoch": 2.41, - "grad_norm": 0.2353515625, - "learning_rate": 1.2105263157894738e-06, - "loss": 1.916, - "step": 1885 - }, - { - "epoch": 2.41, - "grad_norm": 0.23828125, - "learning_rate": 1.2000000000000002e-06, - "loss": 1.9615, - "step": 1886 - }, - { - "epoch": 2.41, - "grad_norm": 0.2265625, - "learning_rate": 1.1894736842105265e-06, - "loss": 1.9673, - "step": 1887 - }, - { - "epoch": 2.42, - "grad_norm": 0.236328125, - "learning_rate": 1.1789473684210526e-06, - "loss": 1.9517, - "step": 1888 - }, - { - "epoch": 2.42, - "grad_norm": 0.25, - "learning_rate": 1.1684210526315791e-06, - "loss": 1.8982, - "step": 1889 - }, - { - "epoch": 2.42, - "grad_norm": 0.251953125, - "learning_rate": 1.1578947368421053e-06, - "loss": 1.8753, - "step": 1890 - }, - { - "epoch": 2.42, - "grad_norm": 0.2578125, - "learning_rate": 1.1473684210526316e-06, - "loss": 1.8638, - "step": 1891 - }, - { - "epoch": 2.42, - "grad_norm": 0.271484375, - "learning_rate": 1.136842105263158e-06, - "loss": 1.8975, - "step": 1892 - }, - { - "epoch": 2.42, - "grad_norm": 0.21875, - "learning_rate": 1.1263157894736842e-06, - "loss": 1.9828, - "step": 1893 - }, - { - "epoch": 2.42, - "grad_norm": 0.2333984375, - "learning_rate": 1.1157894736842106e-06, - "loss": 1.8852, - "step": 1894 - }, - { - "epoch": 2.42, - "grad_norm": 0.240234375, - "learning_rate": 1.1052631578947369e-06, - "loss": 1.8857, - "step": 1895 - }, - { - "epoch": 2.43, - "grad_norm": 0.3046875, - "learning_rate": 1.0947368421052632e-06, - "loss": 1.8387, - "step": 1896 - }, - { - "epoch": 2.43, - "grad_norm": 0.30859375, - "learning_rate": 1.0842105263157895e-06, - "loss": 1.9522, - "step": 1897 - }, - { - "epoch": 2.43, - "grad_norm": 0.224609375, - "learning_rate": 1.0736842105263159e-06, - "loss": 1.9089, - "step": 1898 - }, - { - "epoch": 2.43, - "grad_norm": 0.232421875, - "learning_rate": 1.0631578947368422e-06, - "loss": 1.9909, - "step": 1899 - }, - { - "epoch": 2.43, - "grad_norm": 0.22265625, - "learning_rate": 1.0526315789473685e-06, - "loss": 1.9401, - "step": 1900 - }, - { - "epoch": 2.43, - "grad_norm": 0.2373046875, - "learning_rate": 1.0421052631578949e-06, - "loss": 1.8911, - "step": 1901 - }, - { - "epoch": 2.43, - "grad_norm": 0.2578125, - "learning_rate": 1.0315789473684212e-06, - "loss": 1.8742, - "step": 1902 - }, - { - "epoch": 2.44, - "grad_norm": 0.24609375, - "learning_rate": 1.0210526315789475e-06, - "loss": 1.8874, - "step": 1903 - }, - { - "epoch": 2.44, - "grad_norm": 0.2373046875, - "learning_rate": 1.0105263157894738e-06, - "loss": 1.9229, - "step": 1904 - }, - { - "epoch": 2.44, - "grad_norm": 0.251953125, - "learning_rate": 1.0000000000000002e-06, - "loss": 1.9311, - "step": 1905 - }, - { - "epoch": 2.44, - "grad_norm": 0.212890625, - "learning_rate": 9.894736842105265e-07, - "loss": 1.9127, - "step": 1906 - }, - { - "epoch": 2.44, - "grad_norm": 0.267578125, - "learning_rate": 9.789473684210526e-07, - "loss": 1.8882, - "step": 1907 - }, - { - "epoch": 2.44, - "grad_norm": 0.22265625, - "learning_rate": 9.68421052631579e-07, - "loss": 1.8046, - "step": 1908 - }, - { - "epoch": 2.44, - "grad_norm": 0.23046875, - "learning_rate": 9.578947368421053e-07, - "loss": 1.9357, - "step": 1909 - }, - { - "epoch": 2.44, - "grad_norm": 0.24609375, - "learning_rate": 9.473684210526317e-07, - "loss": 1.9565, - "step": 1910 - }, - { - "epoch": 2.45, - "grad_norm": 0.287109375, - "learning_rate": 9.368421052631579e-07, - "loss": 1.8328, - "step": 1911 - }, - { - "epoch": 2.45, - "grad_norm": 0.2431640625, - "learning_rate": 9.263157894736844e-07, - "loss": 2.0138, - "step": 1912 - }, - { - "epoch": 2.45, - "grad_norm": 0.248046875, - "learning_rate": 9.157894736842106e-07, - "loss": 1.8319, - "step": 1913 - }, - { - "epoch": 2.45, - "grad_norm": 0.2294921875, - "learning_rate": 9.05263157894737e-07, - "loss": 1.8363, - "step": 1914 - }, - { - "epoch": 2.45, - "grad_norm": 0.216796875, - "learning_rate": 8.947368421052632e-07, - "loss": 1.929, - "step": 1915 - }, - { - "epoch": 2.45, - "grad_norm": 0.234375, - "learning_rate": 8.842105263157895e-07, - "loss": 1.9703, - "step": 1916 - }, - { - "epoch": 2.45, - "grad_norm": 0.2470703125, - "learning_rate": 8.736842105263159e-07, - "loss": 1.8959, - "step": 1917 - }, - { - "epoch": 2.45, - "grad_norm": 0.27734375, - "learning_rate": 8.631578947368421e-07, - "loss": 1.8569, - "step": 1918 - }, - { - "epoch": 2.46, - "grad_norm": 0.2353515625, - "learning_rate": 8.526315789473685e-07, - "loss": 1.9756, - "step": 1919 - }, - { - "epoch": 2.46, - "grad_norm": 0.22265625, - "learning_rate": 8.421052631578948e-07, - "loss": 1.8456, - "step": 1920 - }, - { - "epoch": 2.46, - "grad_norm": 0.2353515625, - "learning_rate": 8.315789473684212e-07, - "loss": 1.9389, - "step": 1921 - }, - { - "epoch": 2.46, - "grad_norm": 0.28515625, - "learning_rate": 8.210526315789474e-07, - "loss": 1.8881, - "step": 1922 - }, - { - "epoch": 2.46, - "grad_norm": 0.2216796875, - "learning_rate": 8.105263157894736e-07, - "loss": 1.8839, - "step": 1923 - }, - { - "epoch": 2.46, - "grad_norm": 0.310546875, - "learning_rate": 8.000000000000001e-07, - "loss": 1.9732, - "step": 1924 - }, - { - "epoch": 2.46, - "grad_norm": 0.2216796875, - "learning_rate": 7.894736842105263e-07, - "loss": 1.8881, - "step": 1925 - }, - { - "epoch": 2.46, - "grad_norm": 0.244140625, - "learning_rate": 7.789473684210527e-07, - "loss": 1.8409, - "step": 1926 - }, - { - "epoch": 2.47, - "grad_norm": 0.251953125, - "learning_rate": 7.684210526315789e-07, - "loss": 1.9125, - "step": 1927 - }, - { - "epoch": 2.47, - "grad_norm": 0.2216796875, - "learning_rate": 7.578947368421054e-07, - "loss": 1.9222, - "step": 1928 - }, - { - "epoch": 2.47, - "grad_norm": 0.279296875, - "learning_rate": 7.473684210526316e-07, - "loss": 1.8613, - "step": 1929 - }, - { - "epoch": 2.47, - "grad_norm": 0.2421875, - "learning_rate": 7.368421052631579e-07, - "loss": 1.8954, - "step": 1930 - }, - { - "epoch": 2.47, - "grad_norm": 0.2470703125, - "learning_rate": 7.263157894736843e-07, - "loss": 1.9187, - "step": 1931 - }, - { - "epoch": 2.47, - "grad_norm": 0.267578125, - "learning_rate": 7.157894736842106e-07, - "loss": 1.8443, - "step": 1932 - }, - { - "epoch": 2.47, - "grad_norm": 0.22265625, - "learning_rate": 7.052631578947369e-07, - "loss": 1.9389, - "step": 1933 - }, - { - "epoch": 2.47, - "grad_norm": 0.2080078125, - "learning_rate": 6.947368421052631e-07, - "loss": 1.8359, - "step": 1934 - }, - { - "epoch": 2.48, - "grad_norm": 0.2236328125, - "learning_rate": 6.842105263157896e-07, - "loss": 1.9023, - "step": 1935 - }, - { - "epoch": 2.48, - "grad_norm": 0.248046875, - "learning_rate": 6.736842105263158e-07, - "loss": 1.8365, - "step": 1936 - }, - { - "epoch": 2.48, - "grad_norm": 0.23046875, - "learning_rate": 6.631578947368422e-07, - "loss": 1.9567, - "step": 1937 - }, - { - "epoch": 2.48, - "grad_norm": 0.275390625, - "learning_rate": 6.526315789473684e-07, - "loss": 1.8556, - "step": 1938 - }, - { - "epoch": 2.48, - "grad_norm": 0.23046875, - "learning_rate": 6.421052631578948e-07, - "loss": 1.8906, - "step": 1939 - }, - { - "epoch": 2.48, - "grad_norm": 0.21484375, - "learning_rate": 6.315789473684211e-07, - "loss": 1.9299, - "step": 1940 - }, - { - "epoch": 2.48, - "grad_norm": 0.302734375, - "learning_rate": 6.210526315789474e-07, - "loss": 1.9133, - "step": 1941 - }, - { - "epoch": 2.48, - "grad_norm": 0.232421875, - "learning_rate": 6.105263157894738e-07, - "loss": 1.8596, - "step": 1942 - }, - { - "epoch": 2.49, - "grad_norm": 0.2119140625, - "learning_rate": 6.000000000000001e-07, - "loss": 1.8562, - "step": 1943 - }, - { - "epoch": 2.49, - "grad_norm": 0.23828125, - "learning_rate": 5.894736842105263e-07, - "loss": 1.8824, - "step": 1944 - }, - { - "epoch": 2.49, - "grad_norm": 0.283203125, - "learning_rate": 5.789473684210526e-07, - "loss": 1.8897, - "step": 1945 - }, - { - "epoch": 2.49, - "grad_norm": 0.251953125, - "learning_rate": 5.68421052631579e-07, - "loss": 1.9513, - "step": 1946 - }, - { - "epoch": 2.49, - "grad_norm": 0.2314453125, - "learning_rate": 5.578947368421053e-07, - "loss": 1.881, - "step": 1947 - }, - { - "epoch": 2.49, - "grad_norm": 0.25390625, - "learning_rate": 5.473684210526316e-07, - "loss": 1.9103, - "step": 1948 - }, - { - "epoch": 2.49, - "grad_norm": 0.21875, - "learning_rate": 5.368421052631579e-07, - "loss": 1.8638, - "step": 1949 - }, - { - "epoch": 2.5, - "grad_norm": 0.23046875, - "learning_rate": 5.263157894736843e-07, - "loss": 1.8953, - "step": 1950 - }, - { - "epoch": 2.5, - "grad_norm": 0.263671875, - "learning_rate": 5.157894736842106e-07, - "loss": 1.8558, - "step": 1951 - }, - { - "epoch": 2.5, - "grad_norm": 0.248046875, - "learning_rate": 5.052631578947369e-07, - "loss": 1.8183, - "step": 1952 - }, - { - "epoch": 2.5, - "grad_norm": 0.384765625, - "learning_rate": 4.947368421052632e-07, - "loss": 1.983, - "step": 1953 - }, - { - "epoch": 2.5, - "grad_norm": 0.267578125, - "learning_rate": 4.842105263157895e-07, - "loss": 1.9494, - "step": 1954 - }, - { - "epoch": 2.5, - "grad_norm": 0.27734375, - "learning_rate": 4.7368421052631585e-07, - "loss": 1.9199, - "step": 1955 - }, - { - "epoch": 2.5, - "grad_norm": 0.294921875, - "learning_rate": 4.631578947368422e-07, - "loss": 1.8667, - "step": 1956 - }, - { - "epoch": 2.5, - "grad_norm": 0.30859375, - "learning_rate": 4.526315789473685e-07, - "loss": 1.8608, - "step": 1957 - }, - { - "epoch": 2.51, - "grad_norm": 0.2314453125, - "learning_rate": 4.421052631578947e-07, - "loss": 1.8885, - "step": 1958 - }, - { - "epoch": 2.51, - "grad_norm": 0.251953125, - "learning_rate": 4.3157894736842105e-07, - "loss": 1.9525, - "step": 1959 - }, - { - "epoch": 2.51, - "grad_norm": 0.2041015625, - "learning_rate": 4.210526315789474e-07, - "loss": 1.9011, - "step": 1960 - }, - { - "epoch": 2.51, - "grad_norm": 0.2255859375, - "learning_rate": 4.105263157894737e-07, - "loss": 1.9285, - "step": 1961 - }, - { - "epoch": 2.51, - "grad_norm": 0.224609375, - "learning_rate": 4.0000000000000003e-07, - "loss": 1.7551, - "step": 1962 - }, - { - "epoch": 2.51, - "grad_norm": 0.30078125, - "learning_rate": 3.8947368421052636e-07, - "loss": 1.9228, - "step": 1963 - }, - { - "epoch": 2.51, - "grad_norm": 0.2353515625, - "learning_rate": 3.789473684210527e-07, - "loss": 1.8818, - "step": 1964 - }, - { - "epoch": 2.51, - "grad_norm": 0.25, - "learning_rate": 3.6842105263157896e-07, - "loss": 1.9237, - "step": 1965 - }, - { - "epoch": 2.52, - "grad_norm": 0.255859375, - "learning_rate": 3.578947368421053e-07, - "loss": 1.9512, - "step": 1966 - }, - { - "epoch": 2.52, - "grad_norm": 0.2431640625, - "learning_rate": 3.4736842105263157e-07, - "loss": 1.8773, - "step": 1967 - }, - { - "epoch": 2.52, - "grad_norm": 0.2294921875, - "learning_rate": 3.368421052631579e-07, - "loss": 1.9225, - "step": 1968 - }, - { - "epoch": 2.52, - "grad_norm": 0.318359375, - "learning_rate": 3.263157894736842e-07, - "loss": 1.8592, - "step": 1969 - }, - { - "epoch": 2.52, - "grad_norm": 0.236328125, - "learning_rate": 3.1578947368421055e-07, - "loss": 1.9107, - "step": 1970 - }, - { - "epoch": 2.52, - "grad_norm": 0.2490234375, - "learning_rate": 3.052631578947369e-07, - "loss": 1.9027, - "step": 1971 - }, - { - "epoch": 2.52, - "grad_norm": 0.236328125, - "learning_rate": 2.9473684210526315e-07, - "loss": 1.8329, - "step": 1972 - }, - { - "epoch": 2.52, - "grad_norm": 0.2158203125, - "learning_rate": 2.842105263157895e-07, - "loss": 1.865, - "step": 1973 - }, - { - "epoch": 2.53, - "grad_norm": 0.2265625, - "learning_rate": 2.736842105263158e-07, - "loss": 1.8684, - "step": 1974 - }, - { - "epoch": 2.53, - "grad_norm": 0.24609375, - "learning_rate": 2.6315789473684213e-07, - "loss": 1.8998, - "step": 1975 - }, - { - "epoch": 2.53, - "grad_norm": 0.314453125, - "learning_rate": 2.5263157894736846e-07, - "loss": 1.884, - "step": 1976 - }, - { - "epoch": 2.53, - "grad_norm": 0.2275390625, - "learning_rate": 2.4210526315789473e-07, - "loss": 1.8763, - "step": 1977 - }, - { - "epoch": 2.53, - "grad_norm": 0.30078125, - "learning_rate": 2.315789473684211e-07, - "loss": 1.8355, - "step": 1978 - }, - { - "epoch": 2.53, - "grad_norm": 0.251953125, - "learning_rate": 2.2105263157894736e-07, - "loss": 1.878, - "step": 1979 - }, - { - "epoch": 2.53, - "grad_norm": 0.228515625, - "learning_rate": 2.105263157894737e-07, - "loss": 1.8831, - "step": 1980 - }, - { - "epoch": 2.53, - "grad_norm": 0.259765625, - "learning_rate": 2.0000000000000002e-07, - "loss": 1.9792, - "step": 1981 - }, - { - "epoch": 2.54, - "grad_norm": 0.255859375, - "learning_rate": 1.8947368421052634e-07, - "loss": 1.9032, - "step": 1982 - }, - { - "epoch": 2.54, - "grad_norm": 0.291015625, - "learning_rate": 1.7894736842105265e-07, - "loss": 2.0402, - "step": 1983 - }, - { - "epoch": 2.54, - "grad_norm": 0.2431640625, - "learning_rate": 1.6842105263157895e-07, - "loss": 1.894, - "step": 1984 - }, - { - "epoch": 2.54, - "grad_norm": 0.251953125, - "learning_rate": 1.5789473684210527e-07, - "loss": 1.8959, - "step": 1985 - }, - { - "epoch": 2.54, - "grad_norm": 0.2138671875, - "learning_rate": 1.4736842105263158e-07, - "loss": 1.8814, - "step": 1986 - }, - { - "epoch": 2.54, - "grad_norm": 0.25, - "learning_rate": 1.368421052631579e-07, - "loss": 1.8637, - "step": 1987 - }, - { - "epoch": 2.54, - "grad_norm": 0.25390625, - "learning_rate": 1.2631578947368423e-07, - "loss": 1.8855, - "step": 1988 - }, - { - "epoch": 2.55, - "grad_norm": 0.2578125, - "learning_rate": 1.1578947368421054e-07, - "loss": 1.938, - "step": 1989 - }, - { - "epoch": 2.55, - "grad_norm": 0.224609375, - "learning_rate": 1.0526315789473685e-07, - "loss": 1.8612, - "step": 1990 - }, - { - "epoch": 2.55, - "grad_norm": 0.2041015625, - "learning_rate": 9.473684210526317e-08, - "loss": 1.8995, - "step": 1991 - }, - { - "epoch": 2.55, - "grad_norm": 0.251953125, - "learning_rate": 8.421052631578947e-08, - "loss": 1.8756, - "step": 1992 - }, - { - "epoch": 2.55, - "grad_norm": 0.291015625, - "learning_rate": 7.368421052631579e-08, - "loss": 1.8318, - "step": 1993 - }, - { - "epoch": 2.55, - "grad_norm": 0.30859375, - "learning_rate": 6.315789473684211e-08, - "loss": 1.9934, - "step": 1994 - }, - { - "epoch": 2.55, - "grad_norm": 0.234375, - "learning_rate": 5.263157894736842e-08, - "loss": 1.8822, - "step": 1995 - }, - { - "epoch": 2.55, - "grad_norm": 0.2431640625, - "learning_rate": 4.2105263157894737e-08, - "loss": 1.9347, - "step": 1996 - }, - { - "epoch": 2.56, - "grad_norm": 0.2216796875, - "learning_rate": 3.157894736842106e-08, - "loss": 1.8804, - "step": 1997 - }, - { - "epoch": 2.56, - "grad_norm": 0.251953125, - "learning_rate": 2.1052631578947368e-08, - "loss": 1.8495, - "step": 1998 - }, - { - "epoch": 2.56, - "grad_norm": 0.259765625, - "learning_rate": 1.0526315789473684e-08, - "loss": 1.8955, - "step": 1999 - }, - { - "epoch": 2.56, - "grad_norm": 0.25390625, - "learning_rate": 0.0, - "loss": 1.9032, - "step": 2000 - }, - { - "epoch": 2.56, - "eval_loss": 1.8948355913162231, - "eval_runtime": 125.5957, - "eval_samples_per_second": 39.81, - "eval_steps_per_second": 1.25, - "step": 2000 - }, - { - "epoch": 2.56, - "step": 2000, - "total_flos": 1.4606300151763108e+18, - "train_loss": 2.081684485554695, - "train_runtime": 11414.8095, - "train_samples_per_second": 11.214, - "train_steps_per_second": 0.175 - }, - { - "epoch": 2.56, - "eval_loss": 1.8946523666381836, - "eval_runtime": 125.3616, - "eval_samples_per_second": 39.885, - "eval_steps_per_second": 1.252, - "step": 2000 + "eval_loss": 1.7247449159622192, + "eval_runtime": 125.8509, + "eval_samples_per_second": 39.73, + "eval_steps_per_second": 1.248, + "step": 1024 } ], "logging_steps": 1, - "max_steps": 2000, + "max_steps": 1024, "num_input_tokens_seen": 0, - "num_train_epochs": 3, - "save_steps": 200, - "total_flos": 1.4606300151763108e+18, + "num_train_epochs": 2, + "save_steps": 50, + "total_flos": 7.46638513614422e+17, "train_batch_size": 32, "trial_name": null, "trial_params": null