Upload folder using huggingface_hub
Browse files
slots/3/checkpoint-1535/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"TwinyForCausalLM"
|
| 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": "twiny",
|
| 13 |
+
"neftune_alpha": 0.0,
|
| 14 |
+
"num_attention_heads": 12,
|
| 15 |
+
"num_hidden_layers": 3,
|
| 16 |
+
"num_key_value_heads": 3,
|
| 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 |
+
}
|
slots/3/checkpoint-1535/trainer_state.json
ADDED
|
@@ -0,0 +1,573 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.20521904793736473,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1535,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0001336931908386741,
|
| 14 |
+
"grad_norm": Infinity,
|
| 15 |
+
"learning_rate": 5e-05,
|
| 16 |
+
"loss": 129.80032348632812,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002673863816773482,
|
| 21 |
+
"grad_norm": 63.64365768432617,
|
| 22 |
+
"learning_rate": 4.999995392022967e-05,
|
| 23 |
+
"loss": 63.88374408922697,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.005347727633546964,
|
| 28 |
+
"grad_norm": 24.627853393554688,
|
| 29 |
+
"learning_rate": 4.999976672145381e-05,
|
| 30 |
+
"loss": 12.65963363647461,
|
| 31 |
+
"step": 40
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.008021591450320446,
|
| 35 |
+
"grad_norm": 14.29983901977539,
|
| 36 |
+
"learning_rate": 4.999943552476422e-05,
|
| 37 |
+
"loss": 5.90204963684082,
|
| 38 |
+
"step": 60
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.010695455267093928,
|
| 42 |
+
"grad_norm": 15.690323829650879,
|
| 43 |
+
"learning_rate": 4.999896033206858e-05,
|
| 44 |
+
"loss": 3.9918922424316405,
|
| 45 |
+
"step": 80
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.01336931908386741,
|
| 49 |
+
"grad_norm": 31.583160400390625,
|
| 50 |
+
"learning_rate": 4.999834114610398e-05,
|
| 51 |
+
"loss": 2.9675426483154297,
|
| 52 |
+
"step": 100
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.01604318290064089,
|
| 56 |
+
"grad_norm": 13.034649848937988,
|
| 57 |
+
"learning_rate": 4.999757797043691e-05,
|
| 58 |
+
"loss": 2.725296401977539,
|
| 59 |
+
"step": 120
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.018717046717414372,
|
| 63 |
+
"grad_norm": 8.362203598022461,
|
| 64 |
+
"learning_rate": 4.999667080946324e-05,
|
| 65 |
+
"loss": 2.2478992462158205,
|
| 66 |
+
"step": 140
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.021390910534187856,
|
| 70 |
+
"grad_norm": 8.726786613464355,
|
| 71 |
+
"learning_rate": 4.999561966840821e-05,
|
| 72 |
+
"loss": 1.8447845458984375,
|
| 73 |
+
"step": 160
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.024064774350961337,
|
| 77 |
+
"grad_norm": 10.092752456665039,
|
| 78 |
+
"learning_rate": 4.9994424553326335e-05,
|
| 79 |
+
"loss": 1.5611843109130858,
|
| 80 |
+
"step": 180
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.02673863816773482,
|
| 84 |
+
"grad_norm": 9.090085983276367,
|
| 85 |
+
"learning_rate": 4.999308547110146e-05,
|
| 86 |
+
"loss": 1.520334815979004,
|
| 87 |
+
"step": 200
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.029412501984508302,
|
| 91 |
+
"grad_norm": 9.668124198913574,
|
| 92 |
+
"learning_rate": 4.999160242944665e-05,
|
| 93 |
+
"loss": 1.2818055152893066,
|
| 94 |
+
"step": 220
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.03208636580128178,
|
| 98 |
+
"grad_norm": 9.182533264160156,
|
| 99 |
+
"learning_rate": 4.998997543690418e-05,
|
| 100 |
+
"loss": 1.0428407669067383,
|
| 101 |
+
"step": 240
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.03476022961805526,
|
| 105 |
+
"grad_norm": 5.745838165283203,
|
| 106 |
+
"learning_rate": 4.998820450284549e-05,
|
| 107 |
+
"loss": 1.2343652725219727,
|
| 108 |
+
"step": 260
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.037434093434828744,
|
| 112 |
+
"grad_norm": 8.651643753051758,
|
| 113 |
+
"learning_rate": 4.99862896374711e-05,
|
| 114 |
+
"loss": 0.8859601020812988,
|
| 115 |
+
"step": 280
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.04010795725160223,
|
| 119 |
+
"grad_norm": 10.765266418457031,
|
| 120 |
+
"learning_rate": 4.998423085181056e-05,
|
| 121 |
+
"loss": 0.989600658416748,
|
| 122 |
+
"step": 300
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.04278182106837571,
|
| 126 |
+
"grad_norm": 6.092499256134033,
|
| 127 |
+
"learning_rate": 4.998202815772245e-05,
|
| 128 |
+
"loss": 0.7189463615417481,
|
| 129 |
+
"step": 320
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.04545568488514919,
|
| 133 |
+
"grad_norm": 6.352876663208008,
|
| 134 |
+
"learning_rate": 4.9979681567894195e-05,
|
| 135 |
+
"loss": 0.7489545345306396,
|
| 136 |
+
"step": 340
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.048129548701922674,
|
| 140 |
+
"grad_norm": 4.620656490325928,
|
| 141 |
+
"learning_rate": 4.997719109584209e-05,
|
| 142 |
+
"loss": 0.7381401538848877,
|
| 143 |
+
"step": 360
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.050803412518696155,
|
| 147 |
+
"grad_norm": 7.796917915344238,
|
| 148 |
+
"learning_rate": 4.997455675591119e-05,
|
| 149 |
+
"loss": 0.5687405109405518,
|
| 150 |
+
"step": 380
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.05347727633546964,
|
| 154 |
+
"grad_norm": 2.837172508239746,
|
| 155 |
+
"learning_rate": 4.9971778563275204e-05,
|
| 156 |
+
"loss": 0.5686865329742432,
|
| 157 |
+
"step": 400
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.05615114015224312,
|
| 161 |
+
"grad_norm": 3.3103690147399902,
|
| 162 |
+
"learning_rate": 4.9968856533936436e-05,
|
| 163 |
+
"loss": 0.625730562210083,
|
| 164 |
+
"step": 420
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.058825003969016604,
|
| 168 |
+
"grad_norm": 3.5682132244110107,
|
| 169 |
+
"learning_rate": 4.99657906847257e-05,
|
| 170 |
+
"loss": 0.6125466346740722,
|
| 171 |
+
"step": 440
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.061498867785790085,
|
| 175 |
+
"grad_norm": 5.63640832901001,
|
| 176 |
+
"learning_rate": 4.996258103330218e-05,
|
| 177 |
+
"loss": 0.6182214260101319,
|
| 178 |
+
"step": 460
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.06417273160256357,
|
| 182 |
+
"grad_norm": 4.698945999145508,
|
| 183 |
+
"learning_rate": 4.995922759815339e-05,
|
| 184 |
+
"loss": 0.43828091621398924,
|
| 185 |
+
"step": 480
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.06684659541933705,
|
| 189 |
+
"grad_norm": 2.1976189613342285,
|
| 190 |
+
"learning_rate": 4.995573039859501e-05,
|
| 191 |
+
"loss": 0.4459230899810791,
|
| 192 |
+
"step": 500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.06952045923611053,
|
| 196 |
+
"grad_norm": 3.8809523582458496,
|
| 197 |
+
"learning_rate": 4.995208945477081e-05,
|
| 198 |
+
"loss": 0.3821882963180542,
|
| 199 |
+
"step": 520
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.07219432305288401,
|
| 203 |
+
"grad_norm": 3.75144100189209,
|
| 204 |
+
"learning_rate": 4.994830478765251e-05,
|
| 205 |
+
"loss": 0.5800807476043701,
|
| 206 |
+
"step": 540
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.07486818686965749,
|
| 210 |
+
"grad_norm": 3.0038585662841797,
|
| 211 |
+
"learning_rate": 4.9944376419039684e-05,
|
| 212 |
+
"loss": 0.3928264617919922,
|
| 213 |
+
"step": 560
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.07754205068643098,
|
| 217 |
+
"grad_norm": 3.614591598510742,
|
| 218 |
+
"learning_rate": 4.994030437155961e-05,
|
| 219 |
+
"loss": 0.48637890815734863,
|
| 220 |
+
"step": 580
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.08021591450320446,
|
| 224 |
+
"grad_norm": 4.143443584442139,
|
| 225 |
+
"learning_rate": 4.993608866866718e-05,
|
| 226 |
+
"loss": 0.3650153160095215,
|
| 227 |
+
"step": 600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.08288977831997794,
|
| 231 |
+
"grad_norm": 6.692712783813477,
|
| 232 |
+
"learning_rate": 4.993172933464471e-05,
|
| 233 |
+
"loss": 0.3677916288375854,
|
| 234 |
+
"step": 620
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.08556364213675142,
|
| 238 |
+
"grad_norm": 8.383441925048828,
|
| 239 |
+
"learning_rate": 4.9927226394601815e-05,
|
| 240 |
+
"loss": 0.3399480104446411,
|
| 241 |
+
"step": 640
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0882375059535249,
|
| 245 |
+
"grad_norm": 5.566338062286377,
|
| 246 |
+
"learning_rate": 4.992257987447532e-05,
|
| 247 |
+
"loss": 0.28104052543640134,
|
| 248 |
+
"step": 660
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.09091136977029839,
|
| 252 |
+
"grad_norm": 3.1196420192718506,
|
| 253 |
+
"learning_rate": 4.991778980102904e-05,
|
| 254 |
+
"loss": 0.351950478553772,
|
| 255 |
+
"step": 680
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.09358523358707187,
|
| 259 |
+
"grad_norm": 3.47979736328125,
|
| 260 |
+
"learning_rate": 4.9912856201853644e-05,
|
| 261 |
+
"loss": 0.27501535415649414,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.09625909740384535,
|
| 266 |
+
"grad_norm": 5.446717262268066,
|
| 267 |
+
"learning_rate": 4.990777910536653e-05,
|
| 268 |
+
"loss": 0.2651593923568726,
|
| 269 |
+
"step": 720
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.09893296122061883,
|
| 273 |
+
"grad_norm": 7.6145339012146,
|
| 274 |
+
"learning_rate": 4.990255854081161e-05,
|
| 275 |
+
"loss": 0.35140380859375,
|
| 276 |
+
"step": 740
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.10160682503739231,
|
| 280 |
+
"grad_norm": 8.445616722106934,
|
| 281 |
+
"learning_rate": 4.989719453825918e-05,
|
| 282 |
+
"loss": 0.2961219072341919,
|
| 283 |
+
"step": 760
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.10428068885416579,
|
| 287 |
+
"grad_norm": 6.339537620544434,
|
| 288 |
+
"learning_rate": 4.9891687128605744e-05,
|
| 289 |
+
"loss": 0.24962289333343507,
|
| 290 |
+
"step": 780
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.10695455267093928,
|
| 294 |
+
"grad_norm": 3.3369436264038086,
|
| 295 |
+
"learning_rate": 4.988603634357383e-05,
|
| 296 |
+
"loss": 0.2124847412109375,
|
| 297 |
+
"step": 800
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.10962841648771277,
|
| 301 |
+
"grad_norm": 2.2909045219421387,
|
| 302 |
+
"learning_rate": 4.988024221571177e-05,
|
| 303 |
+
"loss": 0.24679112434387207,
|
| 304 |
+
"step": 820
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.11230228030448625,
|
| 308 |
+
"grad_norm": 3.1149911880493164,
|
| 309 |
+
"learning_rate": 4.9874304778393574e-05,
|
| 310 |
+
"loss": 0.22161397933959961,
|
| 311 |
+
"step": 840
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.11497614412125973,
|
| 315 |
+
"grad_norm": 14.802160263061523,
|
| 316 |
+
"learning_rate": 4.9868224065818706e-05,
|
| 317 |
+
"loss": 0.2623537302017212,
|
| 318 |
+
"step": 860
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.11765000793803321,
|
| 322 |
+
"grad_norm": 5.586325168609619,
|
| 323 |
+
"learning_rate": 4.98620001130119e-05,
|
| 324 |
+
"loss": 0.3560942649841309,
|
| 325 |
+
"step": 880
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.12032387175480669,
|
| 329 |
+
"grad_norm": 3.390017032623291,
|
| 330 |
+
"learning_rate": 4.9855632955822916e-05,
|
| 331 |
+
"loss": 0.16934787034988402,
|
| 332 |
+
"step": 900
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.12299773557158017,
|
| 336 |
+
"grad_norm": 6.070940971374512,
|
| 337 |
+
"learning_rate": 4.984912263092641e-05,
|
| 338 |
+
"loss": 0.2131197214126587,
|
| 339 |
+
"step": 920
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.12567159938835365,
|
| 343 |
+
"grad_norm": 1.4912281036376953,
|
| 344 |
+
"learning_rate": 4.984246917582166e-05,
|
| 345 |
+
"loss": 0.25128653049468996,
|
| 346 |
+
"step": 940
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.12834546320512713,
|
| 350 |
+
"grad_norm": 7.000472545623779,
|
| 351 |
+
"learning_rate": 4.9835672628832366e-05,
|
| 352 |
+
"loss": 0.2653592586517334,
|
| 353 |
+
"step": 960
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.1310193270219006,
|
| 357 |
+
"grad_norm": 5.427223205566406,
|
| 358 |
+
"learning_rate": 4.9828733029106434e-05,
|
| 359 |
+
"loss": 0.1653295636177063,
|
| 360 |
+
"step": 980
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.1336931908386741,
|
| 364 |
+
"grad_norm": 1.9502102136611938,
|
| 365 |
+
"learning_rate": 4.982165041661575e-05,
|
| 366 |
+
"loss": 0.2250870943069458,
|
| 367 |
+
"step": 1000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.13636705465544757,
|
| 371 |
+
"grad_norm": 0.6216259598731995,
|
| 372 |
+
"learning_rate": 4.981442483215595e-05,
|
| 373 |
+
"loss": 0.18943849802017212,
|
| 374 |
+
"step": 1020
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.13904091847222105,
|
| 378 |
+
"grad_norm": 2.3363687992095947,
|
| 379 |
+
"learning_rate": 4.98070563173462e-05,
|
| 380 |
+
"loss": 0.1673592209815979,
|
| 381 |
+
"step": 1040
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.14171478228899453,
|
| 385 |
+
"grad_norm": 1.040717601776123,
|
| 386 |
+
"learning_rate": 4.979954491462892e-05,
|
| 387 |
+
"loss": 0.2113173007965088,
|
| 388 |
+
"step": 1060
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.14438864610576801,
|
| 392 |
+
"grad_norm": 2.735522747039795,
|
| 393 |
+
"learning_rate": 4.979189066726955e-05,
|
| 394 |
+
"loss": 0.17504971027374266,
|
| 395 |
+
"step": 1080
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.1470625099225415,
|
| 399 |
+
"grad_norm": 4.701151371002197,
|
| 400 |
+
"learning_rate": 4.978409361935636e-05,
|
| 401 |
+
"loss": 0.15881222486495972,
|
| 402 |
+
"step": 1100
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.14973637373931498,
|
| 406 |
+
"grad_norm": 2.735919237136841,
|
| 407 |
+
"learning_rate": 4.9776153815800075e-05,
|
| 408 |
+
"loss": 0.14044179916381835,
|
| 409 |
+
"step": 1120
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.15241023755608848,
|
| 413 |
+
"grad_norm": 3.5479538440704346,
|
| 414 |
+
"learning_rate": 4.976807130233375e-05,
|
| 415 |
+
"loss": 0.18565714359283447,
|
| 416 |
+
"step": 1140
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.15508410137286197,
|
| 420 |
+
"grad_norm": 3.2167458534240723,
|
| 421 |
+
"learning_rate": 4.975984612551243e-05,
|
| 422 |
+
"loss": 0.13236271142959594,
|
| 423 |
+
"step": 1160
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.15775796518963545,
|
| 427 |
+
"grad_norm": 1.0206760168075562,
|
| 428 |
+
"learning_rate": 4.975147833271288e-05,
|
| 429 |
+
"loss": 0.19124728441238403,
|
| 430 |
+
"step": 1180
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.16043182900640893,
|
| 434 |
+
"grad_norm": 4.194457530975342,
|
| 435 |
+
"learning_rate": 4.9742967972133335e-05,
|
| 436 |
+
"loss": 0.144741427898407,
|
| 437 |
+
"step": 1200
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.1631056928231824,
|
| 441 |
+
"grad_norm": 3.0225746631622314,
|
| 442 |
+
"learning_rate": 4.973431509279323e-05,
|
| 443 |
+
"loss": 0.1374324679374695,
|
| 444 |
+
"step": 1220
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.1657795566399559,
|
| 448 |
+
"grad_norm": 4.243523120880127,
|
| 449 |
+
"learning_rate": 4.972551974453287e-05,
|
| 450 |
+
"loss": 0.13663809299468993,
|
| 451 |
+
"step": 1240
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.16845342045672937,
|
| 455 |
+
"grad_norm": 2.4990086555480957,
|
| 456 |
+
"learning_rate": 4.971658197801322e-05,
|
| 457 |
+
"loss": 0.16817957162857056,
|
| 458 |
+
"step": 1260
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.17112728427350285,
|
| 462 |
+
"grad_norm": 4.983982563018799,
|
| 463 |
+
"learning_rate": 4.9707501844715554e-05,
|
| 464 |
+
"loss": 0.13795313835144044,
|
| 465 |
+
"step": 1280
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.17380114809027633,
|
| 469 |
+
"grad_norm": 3.6780316829681396,
|
| 470 |
+
"learning_rate": 4.969827939694115e-05,
|
| 471 |
+
"loss": 0.1637880802154541,
|
| 472 |
+
"step": 1300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.1764750119070498,
|
| 476 |
+
"grad_norm": 0.7950732707977295,
|
| 477 |
+
"learning_rate": 4.968891468781105e-05,
|
| 478 |
+
"loss": 0.10979138612747193,
|
| 479 |
+
"step": 1320
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.1791488757238233,
|
| 483 |
+
"grad_norm": 1.2414121627807617,
|
| 484 |
+
"learning_rate": 4.967940777126569e-05,
|
| 485 |
+
"loss": 0.13692171573638917,
|
| 486 |
+
"step": 1340
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.18182273954059677,
|
| 490 |
+
"grad_norm": 2.1383633613586426,
|
| 491 |
+
"learning_rate": 4.9669758702064636e-05,
|
| 492 |
+
"loss": 0.07821698188781738,
|
| 493 |
+
"step": 1360
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.18449660335737025,
|
| 497 |
+
"grad_norm": 5.061275959014893,
|
| 498 |
+
"learning_rate": 4.965996753578623e-05,
|
| 499 |
+
"loss": 0.19053516387939454,
|
| 500 |
+
"step": 1380
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.18717046717414373,
|
| 504 |
+
"grad_norm": 6.151792049407959,
|
| 505 |
+
"learning_rate": 4.9650034328827305e-05,
|
| 506 |
+
"loss": 0.11360721588134766,
|
| 507 |
+
"step": 1400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.18984433099091721,
|
| 511 |
+
"grad_norm": 1.0604305267333984,
|
| 512 |
+
"learning_rate": 4.963995913840284e-05,
|
| 513 |
+
"loss": 0.13138024806976317,
|
| 514 |
+
"step": 1420
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.1925181948076907,
|
| 518 |
+
"grad_norm": 1.7159489393234253,
|
| 519 |
+
"learning_rate": 4.9629742022545623e-05,
|
| 520 |
+
"loss": 0.08657677173614502,
|
| 521 |
+
"step": 1440
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.19519205862446418,
|
| 525 |
+
"grad_norm": 2.4207754135131836,
|
| 526 |
+
"learning_rate": 4.961938304010595e-05,
|
| 527 |
+
"loss": 0.10309149026870727,
|
| 528 |
+
"step": 1460
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.19786592244123766,
|
| 532 |
+
"grad_norm": 1.532060146331787,
|
| 533 |
+
"learning_rate": 4.9608882250751245e-05,
|
| 534 |
+
"loss": 0.13628544807434081,
|
| 535 |
+
"step": 1480
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.20053978625801114,
|
| 539 |
+
"grad_norm": 6.409943580627441,
|
| 540 |
+
"learning_rate": 4.959823971496574e-05,
|
| 541 |
+
"loss": 0.10584845542907714,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.20321365007478462,
|
| 546 |
+
"grad_norm": 2.452012538909912,
|
| 547 |
+
"learning_rate": 4.9587455494050136e-05,
|
| 548 |
+
"loss": 0.06506187915802002,
|
| 549 |
+
"step": 1520
|
| 550 |
+
}
|
| 551 |
+
],
|
| 552 |
+
"logging_steps": 20,
|
| 553 |
+
"max_steps": 26180,
|
| 554 |
+
"num_input_tokens_seen": 0,
|
| 555 |
+
"num_train_epochs": 4,
|
| 556 |
+
"save_steps": 1000000000,
|
| 557 |
+
"stateful_callbacks": {
|
| 558 |
+
"TrainerControl": {
|
| 559 |
+
"args": {
|
| 560 |
+
"should_epoch_stop": false,
|
| 561 |
+
"should_evaluate": false,
|
| 562 |
+
"should_log": false,
|
| 563 |
+
"should_save": true,
|
| 564 |
+
"should_training_stop": false
|
| 565 |
+
},
|
| 566 |
+
"attributes": {}
|
| 567 |
+
}
|
| 568 |
+
},
|
| 569 |
+
"total_flos": 5887200834478080.0,
|
| 570 |
+
"train_batch_size": 1,
|
| 571 |
+
"trial_name": null,
|
| 572 |
+
"trial_params": null
|
| 573 |
+
}
|