Upload checkpoint-3000\trainer_state.json with huggingface_hub
Browse files
checkpoint-3000//trainer_state.json
ADDED
|
@@ -0,0 +1,724 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 3000,
|
| 3 |
+
"best_metric": 0.8622779005977237,
|
| 4 |
+
"best_model_checkpoint": "./model_checkpoint\\checkpoint-3000",
|
| 5 |
+
"epoch": 3.927308447937132,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 3000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.06548788474132286,
|
| 14 |
+
"grad_norm": 0.932673454284668,
|
| 15 |
+
"learning_rate": 4.9000000000000005e-06,
|
| 16 |
+
"loss": 0.6777,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.13097576948264572,
|
| 21 |
+
"grad_norm": 1.4263807535171509,
|
| 22 |
+
"learning_rate": 9.900000000000002e-06,
|
| 23 |
+
"loss": 0.584,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.13097576948264572,
|
| 28 |
+
"eval_accuracy": 0.7383116351428806,
|
| 29 |
+
"eval_loss": 0.4755142033100128,
|
| 30 |
+
"eval_runtime": 16.8317,
|
| 31 |
+
"eval_samples_per_second": 725.593,
|
| 32 |
+
"eval_steps_per_second": 45.39,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.19646365422396855,
|
| 37 |
+
"grad_norm": 2.0286386013031006,
|
| 38 |
+
"learning_rate": 1.49e-05,
|
| 39 |
+
"loss": 0.4077,
|
| 40 |
+
"step": 150
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.26195153896529144,
|
| 44 |
+
"grad_norm": 1.7711361646652222,
|
| 45 |
+
"learning_rate": 1.9900000000000003e-05,
|
| 46 |
+
"loss": 0.2967,
|
| 47 |
+
"step": 200
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.26195153896529144,
|
| 51 |
+
"eval_accuracy": 0.8327192336035372,
|
| 52 |
+
"eval_loss": 0.2789490818977356,
|
| 53 |
+
"eval_runtime": 17.0163,
|
| 54 |
+
"eval_samples_per_second": 717.724,
|
| 55 |
+
"eval_steps_per_second": 44.898,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.3274394237066143,
|
| 60 |
+
"grad_norm": 1.2721681594848633,
|
| 61 |
+
"learning_rate": 2.4900000000000002e-05,
|
| 62 |
+
"loss": 0.2954,
|
| 63 |
+
"step": 250
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.3929273084479371,
|
| 67 |
+
"grad_norm": 0.7673302292823792,
|
| 68 |
+
"learning_rate": 2.9900000000000002e-05,
|
| 69 |
+
"loss": 0.2724,
|
| 70 |
+
"step": 300
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.3929273084479371,
|
| 74 |
+
"eval_accuracy": 0.8397609105051994,
|
| 75 |
+
"eval_loss": 0.2580530345439911,
|
| 76 |
+
"eval_runtime": 17.1319,
|
| 77 |
+
"eval_samples_per_second": 712.879,
|
| 78 |
+
"eval_steps_per_second": 44.595,
|
| 79 |
+
"step": 300
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.45841519318926,
|
| 83 |
+
"grad_norm": 0.6606993079185486,
|
| 84 |
+
"learning_rate": 3.49e-05,
|
| 85 |
+
"loss": 0.2524,
|
| 86 |
+
"step": 350
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.5239030779305829,
|
| 90 |
+
"grad_norm": 0.9855170249938965,
|
| 91 |
+
"learning_rate": 3.9800000000000005e-05,
|
| 92 |
+
"loss": 0.246,
|
| 93 |
+
"step": 400
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.5239030779305829,
|
| 97 |
+
"eval_accuracy": 0.8489314664701547,
|
| 98 |
+
"eval_loss": 0.24741230905056,
|
| 99 |
+
"eval_runtime": 17.2565,
|
| 100 |
+
"eval_samples_per_second": 707.732,
|
| 101 |
+
"eval_steps_per_second": 44.273,
|
| 102 |
+
"step": 400
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.5893909626719057,
|
| 106 |
+
"grad_norm": 0.863400936126709,
|
| 107 |
+
"learning_rate": 4.4800000000000005e-05,
|
| 108 |
+
"loss": 0.2509,
|
| 109 |
+
"step": 450
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.6548788474132285,
|
| 113 |
+
"grad_norm": 0.2728968858718872,
|
| 114 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 115 |
+
"loss": 0.2502,
|
| 116 |
+
"step": 500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.6548788474132285,
|
| 120 |
+
"eval_accuracy": 0.8481126668304266,
|
| 121 |
+
"eval_loss": 0.24498052895069122,
|
| 122 |
+
"eval_runtime": 17.4414,
|
| 123 |
+
"eval_samples_per_second": 700.229,
|
| 124 |
+
"eval_steps_per_second": 43.804,
|
| 125 |
+
"step": 500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.7203667321545514,
|
| 129 |
+
"grad_norm": 0.5378761291503906,
|
| 130 |
+
"learning_rate": 4.906103286384977e-05,
|
| 131 |
+
"loss": 0.2579,
|
| 132 |
+
"step": 550
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.7858546168958742,
|
| 136 |
+
"grad_norm": 0.2830002009868622,
|
| 137 |
+
"learning_rate": 4.808294209702661e-05,
|
| 138 |
+
"loss": 0.2367,
|
| 139 |
+
"step": 600
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.7858546168958742,
|
| 143 |
+
"eval_accuracy": 0.8522066650290674,
|
| 144 |
+
"eval_loss": 0.23902325332164764,
|
| 145 |
+
"eval_runtime": 17.3908,
|
| 146 |
+
"eval_samples_per_second": 702.27,
|
| 147 |
+
"eval_steps_per_second": 43.931,
|
| 148 |
+
"step": 600
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.8513425016371972,
|
| 152 |
+
"grad_norm": 0.5070816874504089,
|
| 153 |
+
"learning_rate": 4.710485133020345e-05,
|
| 154 |
+
"loss": 0.2405,
|
| 155 |
+
"step": 650
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.91683038637852,
|
| 159 |
+
"grad_norm": 6.335970401763916,
|
| 160 |
+
"learning_rate": 4.6126760563380286e-05,
|
| 161 |
+
"loss": 0.2453,
|
| 162 |
+
"step": 700
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.91683038637852,
|
| 166 |
+
"eval_accuracy": 0.850159665929747,
|
| 167 |
+
"eval_loss": 0.2547371983528137,
|
| 168 |
+
"eval_runtime": 17.4528,
|
| 169 |
+
"eval_samples_per_second": 699.773,
|
| 170 |
+
"eval_steps_per_second": 43.775,
|
| 171 |
+
"step": 700
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.9823182711198428,
|
| 175 |
+
"grad_norm": 0.29032933712005615,
|
| 176 |
+
"learning_rate": 4.514866979655713e-05,
|
| 177 |
+
"loss": 0.2473,
|
| 178 |
+
"step": 750
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 1.0471512770137525,
|
| 182 |
+
"grad_norm": 0.278637558221817,
|
| 183 |
+
"learning_rate": 4.417057902973396e-05,
|
| 184 |
+
"loss": 0.2596,
|
| 185 |
+
"step": 800
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.0471512770137525,
|
| 189 |
+
"eval_accuracy": 0.8526979448129043,
|
| 190 |
+
"eval_loss": 0.23811034858226776,
|
| 191 |
+
"eval_runtime": 17.4788,
|
| 192 |
+
"eval_samples_per_second": 698.732,
|
| 193 |
+
"eval_steps_per_second": 43.71,
|
| 194 |
+
"step": 800
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.1126391617550753,
|
| 198 |
+
"grad_norm": 2.2316906452178955,
|
| 199 |
+
"learning_rate": 4.3192488262910805e-05,
|
| 200 |
+
"loss": 0.2351,
|
| 201 |
+
"step": 850
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 1.1781270464963982,
|
| 205 |
+
"grad_norm": 6.879007339477539,
|
| 206 |
+
"learning_rate": 4.221439749608764e-05,
|
| 207 |
+
"loss": 0.2388,
|
| 208 |
+
"step": 900
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 1.1781270464963982,
|
| 212 |
+
"eval_accuracy": 0.8549905838041432,
|
| 213 |
+
"eval_loss": 0.2344694286584854,
|
| 214 |
+
"eval_runtime": 17.4998,
|
| 215 |
+
"eval_samples_per_second": 697.893,
|
| 216 |
+
"eval_steps_per_second": 43.658,
|
| 217 |
+
"step": 900
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 1.243614931237721,
|
| 221 |
+
"grad_norm": 0.19675926864147186,
|
| 222 |
+
"learning_rate": 4.123630672926448e-05,
|
| 223 |
+
"loss": 0.242,
|
| 224 |
+
"step": 950
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.3091028159790439,
|
| 228 |
+
"grad_norm": 0.2648178040981293,
|
| 229 |
+
"learning_rate": 4.0258215962441316e-05,
|
| 230 |
+
"loss": 0.2267,
|
| 231 |
+
"step": 1000
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 1.3091028159790439,
|
| 235 |
+
"eval_accuracy": 0.8549905838041432,
|
| 236 |
+
"eval_loss": 0.23283295333385468,
|
| 237 |
+
"eval_runtime": 17.5288,
|
| 238 |
+
"eval_samples_per_second": 696.739,
|
| 239 |
+
"eval_steps_per_second": 43.585,
|
| 240 |
+
"step": 1000
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.3745907007203666,
|
| 244 |
+
"grad_norm": 0.2951018214225769,
|
| 245 |
+
"learning_rate": 3.928012519561816e-05,
|
| 246 |
+
"loss": 0.2347,
|
| 247 |
+
"step": 1050
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.4400785854616895,
|
| 251 |
+
"grad_norm": 0.9093786478042603,
|
| 252 |
+
"learning_rate": 3.830203442879499e-05,
|
| 253 |
+
"loss": 0.2381,
|
| 254 |
+
"step": 1100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.4400785854616895,
|
| 258 |
+
"eval_accuracy": 0.8542536641283878,
|
| 259 |
+
"eval_loss": 0.2354062795639038,
|
| 260 |
+
"eval_runtime": 17.5883,
|
| 261 |
+
"eval_samples_per_second": 694.384,
|
| 262 |
+
"eval_steps_per_second": 43.438,
|
| 263 |
+
"step": 1100
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.5055664702030125,
|
| 267 |
+
"grad_norm": 0.43629541993141174,
|
| 268 |
+
"learning_rate": 3.7323943661971835e-05,
|
| 269 |
+
"loss": 0.2343,
|
| 270 |
+
"step": 1150
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.5710543549443354,
|
| 274 |
+
"grad_norm": 0.46896249055862427,
|
| 275 |
+
"learning_rate": 3.634585289514868e-05,
|
| 276 |
+
"loss": 0.2448,
|
| 277 |
+
"step": 1200
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 1.5710543549443354,
|
| 281 |
+
"eval_accuracy": 0.8555637435519529,
|
| 282 |
+
"eval_loss": 0.231441468000412,
|
| 283 |
+
"eval_runtime": 17.7028,
|
| 284 |
+
"eval_samples_per_second": 689.891,
|
| 285 |
+
"eval_steps_per_second": 43.157,
|
| 286 |
+
"step": 1200
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 1.6365422396856582,
|
| 290 |
+
"grad_norm": 3.0710296630859375,
|
| 291 |
+
"learning_rate": 3.536776212832551e-05,
|
| 292 |
+
"loss": 0.215,
|
| 293 |
+
"step": 1250
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.7020301244269809,
|
| 297 |
+
"grad_norm": 0.28947174549102783,
|
| 298 |
+
"learning_rate": 3.4389671361502353e-05,
|
| 299 |
+
"loss": 0.2341,
|
| 300 |
+
"step": 1300
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.7020301244269809,
|
| 304 |
+
"eval_accuracy": 0.8565463031196267,
|
| 305 |
+
"eval_loss": 0.22934912145137787,
|
| 306 |
+
"eval_runtime": 17.7508,
|
| 307 |
+
"eval_samples_per_second": 688.025,
|
| 308 |
+
"eval_steps_per_second": 43.04,
|
| 309 |
+
"step": 1300
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.7675180091683038,
|
| 313 |
+
"grad_norm": 0.916119396686554,
|
| 314 |
+
"learning_rate": 3.341158059467919e-05,
|
| 315 |
+
"loss": 0.2291,
|
| 316 |
+
"step": 1350
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.8330058939096268,
|
| 320 |
+
"grad_norm": 0.46822306513786316,
|
| 321 |
+
"learning_rate": 3.243348982785603e-05,
|
| 322 |
+
"loss": 0.2289,
|
| 323 |
+
"step": 1400
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.8330058939096268,
|
| 327 |
+
"eval_accuracy": 0.8552362236960616,
|
| 328 |
+
"eval_loss": 0.23168529570102692,
|
| 329 |
+
"eval_runtime": 17.7733,
|
| 330 |
+
"eval_samples_per_second": 687.153,
|
| 331 |
+
"eval_steps_per_second": 42.986,
|
| 332 |
+
"step": 1400
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.8984937786509497,
|
| 336 |
+
"grad_norm": 0.5012661218643188,
|
| 337 |
+
"learning_rate": 3.1455399061032865e-05,
|
| 338 |
+
"loss": 0.2325,
|
| 339 |
+
"step": 1450
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.9639816633922724,
|
| 343 |
+
"grad_norm": 1.2745088338851929,
|
| 344 |
+
"learning_rate": 3.0477308294209707e-05,
|
| 345 |
+
"loss": 0.24,
|
| 346 |
+
"step": 1500
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.9639816633922724,
|
| 350 |
+
"eval_accuracy": 0.8562187832637354,
|
| 351 |
+
"eval_loss": 0.22923418879508972,
|
| 352 |
+
"eval_runtime": 17.8388,
|
| 353 |
+
"eval_samples_per_second": 684.631,
|
| 354 |
+
"eval_steps_per_second": 42.828,
|
| 355 |
+
"step": 1500
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 2.028814669286182,
|
| 359 |
+
"grad_norm": 0.3664182722568512,
|
| 360 |
+
"learning_rate": 2.9499217527386542e-05,
|
| 361 |
+
"loss": 0.2262,
|
| 362 |
+
"step": 1550
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.094302554027505,
|
| 366 |
+
"grad_norm": 0.544244647026062,
|
| 367 |
+
"learning_rate": 2.8521126760563384e-05,
|
| 368 |
+
"loss": 0.229,
|
| 369 |
+
"step": 1600
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 2.094302554027505,
|
| 373 |
+
"eval_accuracy": 0.8558093834438713,
|
| 374 |
+
"eval_loss": 0.23112046718597412,
|
| 375 |
+
"eval_runtime": 17.8003,
|
| 376 |
+
"eval_samples_per_second": 686.112,
|
| 377 |
+
"eval_steps_per_second": 42.921,
|
| 378 |
+
"step": 1600
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 2.1597904387688276,
|
| 382 |
+
"grad_norm": 0.7998932600021362,
|
| 383 |
+
"learning_rate": 2.754303599374022e-05,
|
| 384 |
+
"loss": 0.2269,
|
| 385 |
+
"step": 1650
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 2.2252783235101505,
|
| 389 |
+
"grad_norm": 1.73899245262146,
|
| 390 |
+
"learning_rate": 2.656494522691706e-05,
|
| 391 |
+
"loss": 0.2266,
|
| 392 |
+
"step": 1700
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.2252783235101505,
|
| 396 |
+
"eval_accuracy": 0.8564644231556539,
|
| 397 |
+
"eval_loss": 0.23248396813869476,
|
| 398 |
+
"eval_runtime": 17.8458,
|
| 399 |
+
"eval_samples_per_second": 684.363,
|
| 400 |
+
"eval_steps_per_second": 42.811,
|
| 401 |
+
"step": 1700
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 2.2907662082514735,
|
| 405 |
+
"grad_norm": 0.7401416301727295,
|
| 406 |
+
"learning_rate": 2.5586854460093895e-05,
|
| 407 |
+
"loss": 0.2304,
|
| 408 |
+
"step": 1750
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 2.3562540929927964,
|
| 412 |
+
"grad_norm": 1.3307729959487915,
|
| 413 |
+
"learning_rate": 2.4608763693270737e-05,
|
| 414 |
+
"loss": 0.2229,
|
| 415 |
+
"step": 1800
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.3562540929927964,
|
| 419 |
+
"eval_accuracy": 0.8578563825431916,
|
| 420 |
+
"eval_loss": 0.22864140570163727,
|
| 421 |
+
"eval_runtime": 17.9078,
|
| 422 |
+
"eval_samples_per_second": 681.993,
|
| 423 |
+
"eval_steps_per_second": 42.663,
|
| 424 |
+
"step": 1800
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 2.4217419777341194,
|
| 428 |
+
"grad_norm": 0.16573481261730194,
|
| 429 |
+
"learning_rate": 2.3630672926447576e-05,
|
| 430 |
+
"loss": 0.2141,
|
| 431 |
+
"step": 1850
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 2.487229862475442,
|
| 435 |
+
"grad_norm": 0.6072602272033691,
|
| 436 |
+
"learning_rate": 2.2652582159624414e-05,
|
| 437 |
+
"loss": 0.2171,
|
| 438 |
+
"step": 1900
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 2.487229862475442,
|
| 442 |
+
"eval_accuracy": 0.856791943011545,
|
| 443 |
+
"eval_loss": 0.2283138483762741,
|
| 444 |
+
"eval_runtime": 17.9011,
|
| 445 |
+
"eval_samples_per_second": 682.247,
|
| 446 |
+
"eval_steps_per_second": 42.679,
|
| 447 |
+
"step": 1900
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 2.552717747216765,
|
| 451 |
+
"grad_norm": 0.3190251588821411,
|
| 452 |
+
"learning_rate": 2.1674491392801252e-05,
|
| 453 |
+
"loss": 0.2173,
|
| 454 |
+
"step": 1950
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 2.6182056319580878,
|
| 458 |
+
"grad_norm": 0.4229159355163574,
|
| 459 |
+
"learning_rate": 2.069640062597809e-05,
|
| 460 |
+
"loss": 0.2267,
|
| 461 |
+
"step": 2000
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 2.6182056319580878,
|
| 465 |
+
"eval_accuracy": 0.8599033816425121,
|
| 466 |
+
"eval_loss": 0.22846245765686035,
|
| 467 |
+
"eval_runtime": 17.8748,
|
| 468 |
+
"eval_samples_per_second": 683.254,
|
| 469 |
+
"eval_steps_per_second": 42.742,
|
| 470 |
+
"step": 2000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.6836935166994107,
|
| 474 |
+
"grad_norm": 0.6909223198890686,
|
| 475 |
+
"learning_rate": 1.971830985915493e-05,
|
| 476 |
+
"loss": 0.2287,
|
| 477 |
+
"step": 2050
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 2.749181401440733,
|
| 481 |
+
"grad_norm": 0.530602216720581,
|
| 482 |
+
"learning_rate": 1.874021909233177e-05,
|
| 483 |
+
"loss": 0.2332,
|
| 484 |
+
"step": 2100
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 2.749181401440733,
|
| 488 |
+
"eval_accuracy": 0.8573651027593547,
|
| 489 |
+
"eval_loss": 0.22631122171878815,
|
| 490 |
+
"eval_runtime": 17.8614,
|
| 491 |
+
"eval_samples_per_second": 683.766,
|
| 492 |
+
"eval_steps_per_second": 42.774,
|
| 493 |
+
"step": 2100
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 2.814669286182056,
|
| 497 |
+
"grad_norm": 0.4919154942035675,
|
| 498 |
+
"learning_rate": 1.776212832550861e-05,
|
| 499 |
+
"loss": 0.2116,
|
| 500 |
+
"step": 2150
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.880157170923379,
|
| 504 |
+
"grad_norm": 1.3828797340393066,
|
| 505 |
+
"learning_rate": 1.6784037558685448e-05,
|
| 506 |
+
"loss": 0.2218,
|
| 507 |
+
"step": 2200
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.880157170923379,
|
| 511 |
+
"eval_accuracy": 0.8577745025792188,
|
| 512 |
+
"eval_loss": 0.2293933629989624,
|
| 513 |
+
"eval_runtime": 17.8541,
|
| 514 |
+
"eval_samples_per_second": 684.044,
|
| 515 |
+
"eval_steps_per_second": 42.791,
|
| 516 |
+
"step": 2200
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 2.945645055664702,
|
| 520 |
+
"grad_norm": 0.24504908919334412,
|
| 521 |
+
"learning_rate": 1.5805946791862286e-05,
|
| 522 |
+
"loss": 0.2287,
|
| 523 |
+
"step": 2250
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 3.0104780615586115,
|
| 527 |
+
"grad_norm": 0.5326477885246277,
|
| 528 |
+
"learning_rate": 1.4827856025039124e-05,
|
| 529 |
+
"loss": 0.2272,
|
| 530 |
+
"step": 2300
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 3.0104780615586115,
|
| 534 |
+
"eval_accuracy": 0.8573651027593547,
|
| 535 |
+
"eval_loss": 0.2295144498348236,
|
| 536 |
+
"eval_runtime": 17.8719,
|
| 537 |
+
"eval_samples_per_second": 683.363,
|
| 538 |
+
"eval_steps_per_second": 42.749,
|
| 539 |
+
"step": 2300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 3.0759659462999345,
|
| 543 |
+
"grad_norm": 0.6676012277603149,
|
| 544 |
+
"learning_rate": 1.3849765258215963e-05,
|
| 545 |
+
"loss": 0.2126,
|
| 546 |
+
"step": 2350
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 3.1414538310412574,
|
| 550 |
+
"grad_norm": 0.3665286898612976,
|
| 551 |
+
"learning_rate": 1.2871674491392801e-05,
|
| 552 |
+
"loss": 0.2173,
|
| 553 |
+
"step": 2400
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 3.1414538310412574,
|
| 557 |
+
"eval_accuracy": 0.8602309014984033,
|
| 558 |
+
"eval_loss": 0.23410645127296448,
|
| 559 |
+
"eval_runtime": 17.8958,
|
| 560 |
+
"eval_samples_per_second": 682.449,
|
| 561 |
+
"eval_steps_per_second": 42.691,
|
| 562 |
+
"step": 2400
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 3.2069417157825804,
|
| 566 |
+
"grad_norm": 0.44082948565483093,
|
| 567 |
+
"learning_rate": 1.189358372456964e-05,
|
| 568 |
+
"loss": 0.2117,
|
| 569 |
+
"step": 2450
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.2724296005239033,
|
| 573 |
+
"grad_norm": 1.7528928518295288,
|
| 574 |
+
"learning_rate": 1.0915492957746478e-05,
|
| 575 |
+
"loss": 0.2176,
|
| 576 |
+
"step": 2500
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 3.2724296005239033,
|
| 580 |
+
"eval_accuracy": 0.8612134610660771,
|
| 581 |
+
"eval_loss": 0.22918100655078888,
|
| 582 |
+
"eval_runtime": 17.8879,
|
| 583 |
+
"eval_samples_per_second": 682.753,
|
| 584 |
+
"eval_steps_per_second": 42.71,
|
| 585 |
+
"step": 2500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.337917485265226,
|
| 589 |
+
"grad_norm": 1.001876711845398,
|
| 590 |
+
"learning_rate": 9.937402190923318e-06,
|
| 591 |
+
"loss": 0.2141,
|
| 592 |
+
"step": 2550
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 3.4034053700065487,
|
| 596 |
+
"grad_norm": 0.3455939292907715,
|
| 597 |
+
"learning_rate": 8.959311424100156e-06,
|
| 598 |
+
"loss": 0.2147,
|
| 599 |
+
"step": 2600
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 3.4034053700065487,
|
| 603 |
+
"eval_accuracy": 0.8619503807418325,
|
| 604 |
+
"eval_loss": 0.2289513647556305,
|
| 605 |
+
"eval_runtime": 17.8583,
|
| 606 |
+
"eval_samples_per_second": 683.883,
|
| 607 |
+
"eval_steps_per_second": 42.781,
|
| 608 |
+
"step": 2600
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.4688932547478717,
|
| 612 |
+
"grad_norm": 0.39306333661079407,
|
| 613 |
+
"learning_rate": 7.981220657276996e-06,
|
| 614 |
+
"loss": 0.2212,
|
| 615 |
+
"step": 2650
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.5343811394891946,
|
| 619 |
+
"grad_norm": 0.29782700538635254,
|
| 620 |
+
"learning_rate": 7.003129890453834e-06,
|
| 621 |
+
"loss": 0.2225,
|
| 622 |
+
"step": 2700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 3.5343811394891946,
|
| 626 |
+
"eval_accuracy": 0.8574469827233276,
|
| 627 |
+
"eval_loss": 0.2323383092880249,
|
| 628 |
+
"eval_runtime": 17.8359,
|
| 629 |
+
"eval_samples_per_second": 684.744,
|
| 630 |
+
"eval_steps_per_second": 42.835,
|
| 631 |
+
"step": 2700
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 3.599869024230517,
|
| 635 |
+
"grad_norm": 0.5452626943588257,
|
| 636 |
+
"learning_rate": 6.025039123630673e-06,
|
| 637 |
+
"loss": 0.2121,
|
| 638 |
+
"step": 2750
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.66535690897184,
|
| 642 |
+
"grad_norm": 3.8590140342712402,
|
| 643 |
+
"learning_rate": 5.046948356807512e-06,
|
| 644 |
+
"loss": 0.2208,
|
| 645 |
+
"step": 2800
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.66535690897184,
|
| 649 |
+
"eval_accuracy": 0.8591664619667567,
|
| 650 |
+
"eval_loss": 0.23002713918685913,
|
| 651 |
+
"eval_runtime": 17.874,
|
| 652 |
+
"eval_samples_per_second": 683.285,
|
| 653 |
+
"eval_steps_per_second": 42.744,
|
| 654 |
+
"step": 2800
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.730844793713163,
|
| 658 |
+
"grad_norm": 0.6683679223060608,
|
| 659 |
+
"learning_rate": 4.068857589984351e-06,
|
| 660 |
+
"loss": 0.2273,
|
| 661 |
+
"step": 2850
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 3.796332678454486,
|
| 665 |
+
"grad_norm": 0.24069799482822418,
|
| 666 |
+
"learning_rate": 3.0907668231611893e-06,
|
| 667 |
+
"loss": 0.2084,
|
| 668 |
+
"step": 2900
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.796332678454486,
|
| 672 |
+
"eval_accuracy": 0.8621960206337509,
|
| 673 |
+
"eval_loss": 0.22760269045829773,
|
| 674 |
+
"eval_runtime": 17.8864,
|
| 675 |
+
"eval_samples_per_second": 682.809,
|
| 676 |
+
"eval_steps_per_second": 42.714,
|
| 677 |
+
"step": 2900
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 3.861820563195809,
|
| 681 |
+
"grad_norm": 1.0248568058013916,
|
| 682 |
+
"learning_rate": 2.112676056338028e-06,
|
| 683 |
+
"loss": 0.2082,
|
| 684 |
+
"step": 2950
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.927308447937132,
|
| 688 |
+
"grad_norm": 1.0928360223770142,
|
| 689 |
+
"learning_rate": 1.1541471048513302e-06,
|
| 690 |
+
"loss": 0.2159,
|
| 691 |
+
"step": 3000
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 3.927308447937132,
|
| 695 |
+
"eval_accuracy": 0.8622779005977237,
|
| 696 |
+
"eval_loss": 0.22847115993499756,
|
| 697 |
+
"eval_runtime": 17.9258,
|
| 698 |
+
"eval_samples_per_second": 681.307,
|
| 699 |
+
"eval_steps_per_second": 42.62,
|
| 700 |
+
"step": 3000
|
| 701 |
+
}
|
| 702 |
+
],
|
| 703 |
+
"logging_steps": 50,
|
| 704 |
+
"max_steps": 3056,
|
| 705 |
+
"num_input_tokens_seen": 0,
|
| 706 |
+
"num_train_epochs": 4,
|
| 707 |
+
"save_steps": 500,
|
| 708 |
+
"stateful_callbacks": {
|
| 709 |
+
"TrainerControl": {
|
| 710 |
+
"args": {
|
| 711 |
+
"should_epoch_stop": false,
|
| 712 |
+
"should_evaluate": false,
|
| 713 |
+
"should_log": false,
|
| 714 |
+
"should_save": true,
|
| 715 |
+
"should_training_stop": false
|
| 716 |
+
},
|
| 717 |
+
"attributes": {}
|
| 718 |
+
}
|
| 719 |
+
},
|
| 720 |
+
"total_flos": 6354063711332352.0,
|
| 721 |
+
"train_batch_size": 16,
|
| 722 |
+
"trial_name": null,
|
| 723 |
+
"trial_params": null
|
| 724 |
+
}
|