Upload folder using huggingface_hub
Browse files
shared/checkpoints/latest/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AxaiForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"dtype": "float32",
|
| 7 |
+
"hidden_dropout": 0.0,
|
| 8 |
+
"hidden_size": 768,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 3072,
|
| 11 |
+
"max_position_embeddings": 128,
|
| 12 |
+
"model_type": "axai",
|
| 13 |
+
"neftune_alpha": 0.0,
|
| 14 |
+
"num_attention_heads": 12,
|
| 15 |
+
"num_hidden_layers": 16,
|
| 16 |
+
"num_key_value_heads": 6,
|
| 17 |
+
"qk_norm": true,
|
| 18 |
+
"rezero_init": 1.0,
|
| 19 |
+
"rms_norm_eps": 1e-06,
|
| 20 |
+
"rope_theta": 10000.0,
|
| 21 |
+
"transformers_version": "5.0.0",
|
| 22 |
+
"use_cache": false,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
shared/checkpoints/latest/trainer_state.json
ADDED
|
@@ -0,0 +1,391 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.4145493132111064,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0014148440723928551,
|
| 14 |
+
"grad_norm": 46.59520721435547,
|
| 15 |
+
"learning_rate": 5e-05,
|
| 16 |
+
"loss": 408.9247131347656,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0282968814478571,
|
| 21 |
+
"grad_norm": 43.969398498535156,
|
| 22 |
+
"learning_rate": 4.996525250695594e-05,
|
| 23 |
+
"loss": 385.9318205180921,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0565937628957142,
|
| 28 |
+
"grad_norm": 82.97836303710938,
|
| 29 |
+
"learning_rate": 4.985370748593809e-05,
|
| 30 |
+
"loss": 385.9879638671875,
|
| 31 |
+
"step": 40
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.0848906443435713,
|
| 35 |
+
"grad_norm": 145.0235595703125,
|
| 36 |
+
"learning_rate": 4.966561222361462e-05,
|
| 37 |
+
"loss": 368.023681640625,
|
| 38 |
+
"step": 60
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1131875257914284,
|
| 42 |
+
"grad_norm": 156.58584594726562,
|
| 43 |
+
"learning_rate": 4.940154605961632e-05,
|
| 44 |
+
"loss": 363.8541259765625,
|
| 45 |
+
"step": 80
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14148440723928551,
|
| 49 |
+
"grad_norm": 72.1839828491211,
|
| 50 |
+
"learning_rate": 4.906232232646455e-05,
|
| 51 |
+
"loss": 367.31640625,
|
| 52 |
+
"step": 100
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1697812886871426,
|
| 56 |
+
"grad_norm": 74.4588394165039,
|
| 57 |
+
"learning_rate": 4.864898584448015e-05,
|
| 58 |
+
"loss": 358.9984375,
|
| 59 |
+
"step": 120
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.1980781701349997,
|
| 63 |
+
"grad_norm": 76.57646179199219,
|
| 64 |
+
"learning_rate": 4.8162809703702284e-05,
|
| 65 |
+
"loss": 359.79423828125,
|
| 66 |
+
"step": 140
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.2263750515828568,
|
| 70 |
+
"grad_norm": 61.2109260559082,
|
| 71 |
+
"learning_rate": 4.760529134272895e-05,
|
| 72 |
+
"loss": 347.1311767578125,
|
| 73 |
+
"step": 160
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.2546719330307139,
|
| 77 |
+
"grad_norm": 70.05961608886719,
|
| 78 |
+
"learning_rate": 4.697814793655667e-05,
|
| 79 |
+
"loss": 336.2667236328125,
|
| 80 |
+
"step": 180
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.28296881447857103,
|
| 84 |
+
"grad_norm": 68.51240539550781,
|
| 85 |
+
"learning_rate": 4.628331110762456e-05,
|
| 86 |
+
"loss": 340.3522216796875,
|
| 87 |
+
"step": 200
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.3112656959264281,
|
| 91 |
+
"grad_norm": 75.84900665283203,
|
| 92 |
+
"learning_rate": 4.9999614897993864e-05,
|
| 93 |
+
"loss": 339.859619140625,
|
| 94 |
+
"step": 220
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.3395625773742852,
|
| 98 |
+
"grad_norm": 73.60946655273438,
|
| 99 |
+
"learning_rate": 4.995341701131627e-05,
|
| 100 |
+
"loss": 327.986181640625,
|
| 101 |
+
"step": 240
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.3678594588221423,
|
| 105 |
+
"grad_norm": 75.10031127929688,
|
| 106 |
+
"learning_rate": 4.983036177469612e-05,
|
| 107 |
+
"loss": 327.8923828125,
|
| 108 |
+
"step": 260
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.3961563402699994,
|
| 112 |
+
"grad_norm": 80.19892883300781,
|
| 113 |
+
"learning_rate": 4.963082820233332e-05,
|
| 114 |
+
"loss": 318.4505615234375,
|
| 115 |
+
"step": 280
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.4244532217178565,
|
| 119 |
+
"grad_norm": 59.05309295654297,
|
| 120 |
+
"learning_rate": 4.935543086423201e-05,
|
| 121 |
+
"loss": 320.59248046875,
|
| 122 |
+
"step": 300
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.4527501031657136,
|
| 126 |
+
"grad_norm": 46.86222457885742,
|
| 127 |
+
"learning_rate": 4.9005017993304684e-05,
|
| 128 |
+
"loss": 315.8417724609375,
|
| 129 |
+
"step": 320
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.4810469846135707,
|
| 133 |
+
"grad_norm": 52.75422668457031,
|
| 134 |
+
"learning_rate": 4.8580668872786686e-05,
|
| 135 |
+
"loss": 315.3454833984375,
|
| 136 |
+
"step": 340
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.5093438660614278,
|
| 140 |
+
"grad_norm": 53.963768005371094,
|
| 141 |
+
"learning_rate": 4.80836905120082e-05,
|
| 142 |
+
"loss": 312.086181640625,
|
| 143 |
+
"step": 360
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.5376407475092849,
|
| 147 |
+
"grad_norm": 47.19234085083008,
|
| 148 |
+
"learning_rate": 4.751561362076222e-05,
|
| 149 |
+
"loss": 310.32763671875,
|
| 150 |
+
"step": 380
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.5659376289571421,
|
| 154 |
+
"grad_norm": 54.37424850463867,
|
| 155 |
+
"learning_rate": 4.68781878946678e-05,
|
| 156 |
+
"loss": 298.73134765625,
|
| 157 |
+
"step": 400
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.5942345104049991,
|
| 161 |
+
"grad_norm": 69.96826171875,
|
| 162 |
+
"learning_rate": 4.6173376626049615e-05,
|
| 163 |
+
"loss": 304.4895263671875,
|
| 164 |
+
"step": 420
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.6225313918528562,
|
| 168 |
+
"grad_norm": 46.4427490234375,
|
| 169 |
+
"learning_rate": 4.5403350656932655e-05,
|
| 170 |
+
"loss": 296.0695068359375,
|
| 171 |
+
"step": 440
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.6508282733007134,
|
| 175 |
+
"grad_norm": 60.224464416503906,
|
| 176 |
+
"learning_rate": 4.457048169277701e-05,
|
| 177 |
+
"loss": 303.4669921875,
|
| 178 |
+
"step": 460
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.6791251547485704,
|
| 182 |
+
"grad_norm": 48.74308776855469,
|
| 183 |
+
"learning_rate": 4.367733499754655e-05,
|
| 184 |
+
"loss": 298.897607421875,
|
| 185 |
+
"step": 480
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.7074220361964275,
|
| 189 |
+
"grad_norm": 42.0952033996582,
|
| 190 |
+
"learning_rate": 4.996525250695594e-05,
|
| 191 |
+
"loss": 296.8440673828125,
|
| 192 |
+
"step": 500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.7357189176442847,
|
| 196 |
+
"grad_norm": 59.525245666503906,
|
| 197 |
+
"learning_rate": 4.985370748593809e-05,
|
| 198 |
+
"loss": 289.252197265625,
|
| 199 |
+
"step": 520
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.7640157990921417,
|
| 203 |
+
"grad_norm": 41.16178512573242,
|
| 204 |
+
"learning_rate": 4.966561222361462e-05,
|
| 205 |
+
"loss": 285.179833984375,
|
| 206 |
+
"step": 540
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.7923126805399988,
|
| 210 |
+
"grad_norm": 46.047428131103516,
|
| 211 |
+
"learning_rate": 4.940154605961632e-05,
|
| 212 |
+
"loss": 283.002978515625,
|
| 213 |
+
"step": 560
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.820609561987856,
|
| 217 |
+
"grad_norm": 50.66560363769531,
|
| 218 |
+
"learning_rate": 4.906232232646455e-05,
|
| 219 |
+
"loss": 277.0517333984375,
|
| 220 |
+
"step": 580
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.848906443435713,
|
| 224 |
+
"grad_norm": 47.08381271362305,
|
| 225 |
+
"learning_rate": 4.864898584448015e-05,
|
| 226 |
+
"loss": 279.8613037109375,
|
| 227 |
+
"step": 600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.8772033248835701,
|
| 231 |
+
"grad_norm": 50.696502685546875,
|
| 232 |
+
"learning_rate": 4.8162809703702284e-05,
|
| 233 |
+
"loss": 277.827294921875,
|
| 234 |
+
"step": 620
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.9055002063314272,
|
| 238 |
+
"grad_norm": 62.32334518432617,
|
| 239 |
+
"learning_rate": 4.760529134272895e-05,
|
| 240 |
+
"loss": 280.201611328125,
|
| 241 |
+
"step": 640
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.9337970877792843,
|
| 245 |
+
"grad_norm": 48.96547317504883,
|
| 246 |
+
"learning_rate": 4.697814793655667e-05,
|
| 247 |
+
"loss": 275.6574462890625,
|
| 248 |
+
"step": 660
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.9620939692271414,
|
| 252 |
+
"grad_norm": 50.21261215209961,
|
| 253 |
+
"learning_rate": 4.628331110762456e-05,
|
| 254 |
+
"loss": 275.423828125,
|
| 255 |
+
"step": 680
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.9903908506749985,
|
| 259 |
+
"grad_norm": 53.1383056640625,
|
| 260 |
+
"learning_rate": 4.5522920976353423e-05,
|
| 261 |
+
"loss": 272.92900390625,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 1.0183929729411072,
|
| 266 |
+
"grad_norm": 53.41257095336914,
|
| 267 |
+
"learning_rate": 4.4699319569503964e-05,
|
| 268 |
+
"loss": 259.012353515625,
|
| 269 |
+
"step": 720
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.0466898543889642,
|
| 273 |
+
"grad_norm": 43.35313034057617,
|
| 274 |
+
"learning_rate": 4.3815043606656954e-05,
|
| 275 |
+
"loss": 267.3876953125,
|
| 276 |
+
"step": 740
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.0749867358368212,
|
| 280 |
+
"grad_norm": 40.72585678100586,
|
| 281 |
+
"learning_rate": 4.2872816687032934e-05,
|
| 282 |
+
"loss": 263.577294921875,
|
| 283 |
+
"step": 760
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 1.1032836172846785,
|
| 287 |
+
"grad_norm": 39.94559860229492,
|
| 288 |
+
"learning_rate": 4.1875540900716586e-05,
|
| 289 |
+
"loss": 262.37470703125,
|
| 290 |
+
"step": 780
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.1315804987325355,
|
| 294 |
+
"grad_norm": 45.3363151550293,
|
| 295 |
+
"learning_rate": 4.082628789012317e-05,
|
| 296 |
+
"loss": 257.188037109375,
|
| 297 |
+
"step": 800
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.1598773801803925,
|
| 301 |
+
"grad_norm": 54.727962493896484,
|
| 302 |
+
"learning_rate": 3.972828938923834e-05,
|
| 303 |
+
"loss": 256.655029296875,
|
| 304 |
+
"step": 820
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 1.1881742616282498,
|
| 308 |
+
"grad_norm": 46.29310607910156,
|
| 309 |
+
"learning_rate": 3.858492726977051e-05,
|
| 310 |
+
"loss": 264.1218994140625,
|
| 311 |
+
"step": 840
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.2164711430761068,
|
| 315 |
+
"grad_norm": 57.223209381103516,
|
| 316 |
+
"learning_rate": 3.739972312487412e-05,
|
| 317 |
+
"loss": 250.47421875,
|
| 318 |
+
"step": 860
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.2447680245239638,
|
| 322 |
+
"grad_norm": 55.62833023071289,
|
| 323 |
+
"learning_rate": 3.6176327422526226e-05,
|
| 324 |
+
"loss": 255.1604248046875,
|
| 325 |
+
"step": 880
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.273064905971821,
|
| 329 |
+
"grad_norm": 56.45187759399414,
|
| 330 |
+
"learning_rate": 3.491850826196432e-05,
|
| 331 |
+
"loss": 254.080322265625,
|
| 332 |
+
"step": 900
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.3013617874196781,
|
| 336 |
+
"grad_norm": 60.60747146606445,
|
| 337 |
+
"learning_rate": 3.363013976781608e-05,
|
| 338 |
+
"loss": 250.2458251953125,
|
| 339 |
+
"step": 920
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.3296586688675351,
|
| 343 |
+
"grad_norm": 47.585941314697266,
|
| 344 |
+
"learning_rate": 3.231519015766727e-05,
|
| 345 |
+
"loss": 258.33154296875,
|
| 346 |
+
"step": 940
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.3579555503153924,
|
| 350 |
+
"grad_norm": 72.65380096435547,
|
| 351 |
+
"learning_rate": 3.097770951982037e-05,
|
| 352 |
+
"loss": 253.3177734375,
|
| 353 |
+
"step": 960
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.3862524317632494,
|
| 357 |
+
"grad_norm": 61.72402572631836,
|
| 358 |
+
"learning_rate": 2.9621817338888546e-05,
|
| 359 |
+
"loss": 246.7986572265625,
|
| 360 |
+
"step": 980
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.4145493132111064,
|
| 364 |
+
"grad_norm": 46.793060302734375,
|
| 365 |
+
"learning_rate": 2.825168980764682e-05,
|
| 366 |
+
"loss": 247.830126953125,
|
| 367 |
+
"step": 1000
|
| 368 |
+
}
|
| 369 |
+
],
|
| 370 |
+
"logging_steps": 20,
|
| 371 |
+
"max_steps": 1132,
|
| 372 |
+
"num_input_tokens_seen": 0,
|
| 373 |
+
"num_train_epochs": 2,
|
| 374 |
+
"save_steps": 40,
|
| 375 |
+
"stateful_callbacks": {
|
| 376 |
+
"TrainerControl": {
|
| 377 |
+
"args": {
|
| 378 |
+
"should_epoch_stop": false,
|
| 379 |
+
"should_evaluate": false,
|
| 380 |
+
"should_log": false,
|
| 381 |
+
"should_save": true,
|
| 382 |
+
"should_training_stop": false
|
| 383 |
+
},
|
| 384 |
+
"attributes": {}
|
| 385 |
+
}
|
| 386 |
+
},
|
| 387 |
+
"total_flos": 1.294819840548864e+16,
|
| 388 |
+
"train_batch_size": 1,
|
| 389 |
+
"trial_name": null,
|
| 390 |
+
"trial_params": null
|
| 391 |
+
}
|