Upload folder using huggingface_hub
Browse files
slots/16/checkpoint-1731/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/16/checkpoint-1731/trainer_state.json
ADDED
|
@@ -0,0 +1,643 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.23142291334174486,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1731,
|
| 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": 150.77503967285156,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002673863816773482,
|
| 21 |
+
"grad_norm": 65.5747299194336,
|
| 22 |
+
"learning_rate": 4.999995392022967e-05,
|
| 23 |
+
"loss": 65.57097424958882,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.005347727633546964,
|
| 28 |
+
"grad_norm": 29.183452606201172,
|
| 29 |
+
"learning_rate": 4.999976672145381e-05,
|
| 30 |
+
"loss": 12.781006622314454,
|
| 31 |
+
"step": 40
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.008021591450320446,
|
| 35 |
+
"grad_norm": 22.14234161376953,
|
| 36 |
+
"learning_rate": 4.999943552476422e-05,
|
| 37 |
+
"loss": 7.058085632324219,
|
| 38 |
+
"step": 60
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.010695455267093928,
|
| 42 |
+
"grad_norm": 25.116792678833008,
|
| 43 |
+
"learning_rate": 4.999896033206858e-05,
|
| 44 |
+
"loss": 5.17547492980957,
|
| 45 |
+
"step": 80
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.01336931908386741,
|
| 49 |
+
"grad_norm": 17.95206642150879,
|
| 50 |
+
"learning_rate": 4.999834114610398e-05,
|
| 51 |
+
"loss": 3.6641620635986327,
|
| 52 |
+
"step": 100
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.01604318290064089,
|
| 56 |
+
"grad_norm": 13.571253776550293,
|
| 57 |
+
"learning_rate": 4.999757797043691e-05,
|
| 58 |
+
"loss": 2.9684438705444336,
|
| 59 |
+
"step": 120
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.018717046717414372,
|
| 63 |
+
"grad_norm": 11.740418434143066,
|
| 64 |
+
"learning_rate": 4.999667080946324e-05,
|
| 65 |
+
"loss": 2.658764457702637,
|
| 66 |
+
"step": 140
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.021390910534187856,
|
| 70 |
+
"grad_norm": 8.911559104919434,
|
| 71 |
+
"learning_rate": 4.999561966840821e-05,
|
| 72 |
+
"loss": 2.0916315078735352,
|
| 73 |
+
"step": 160
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.024064774350961337,
|
| 77 |
+
"grad_norm": 7.613410472869873,
|
| 78 |
+
"learning_rate": 4.9994424553326335e-05,
|
| 79 |
+
"loss": 1.730122184753418,
|
| 80 |
+
"step": 180
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.02673863816773482,
|
| 84 |
+
"grad_norm": 10.990550994873047,
|
| 85 |
+
"learning_rate": 4.999308547110146e-05,
|
| 86 |
+
"loss": 1.689904022216797,
|
| 87 |
+
"step": 200
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.029412501984508302,
|
| 91 |
+
"grad_norm": 13.414973258972168,
|
| 92 |
+
"learning_rate": 4.999160242944665e-05,
|
| 93 |
+
"loss": 1.331161880493164,
|
| 94 |
+
"step": 220
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.03208636580128178,
|
| 98 |
+
"grad_norm": 9.434646606445312,
|
| 99 |
+
"learning_rate": 4.998997543690418e-05,
|
| 100 |
+
"loss": 1.285347843170166,
|
| 101 |
+
"step": 240
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.03476022961805526,
|
| 105 |
+
"grad_norm": 15.048566818237305,
|
| 106 |
+
"learning_rate": 4.998820450284549e-05,
|
| 107 |
+
"loss": 1.1024582862854004,
|
| 108 |
+
"step": 260
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.037434093434828744,
|
| 112 |
+
"grad_norm": 7.001116752624512,
|
| 113 |
+
"learning_rate": 4.99862896374711e-05,
|
| 114 |
+
"loss": 1.106878662109375,
|
| 115 |
+
"step": 280
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.04010795725160223,
|
| 119 |
+
"grad_norm": 10.976016998291016,
|
| 120 |
+
"learning_rate": 4.998423085181056e-05,
|
| 121 |
+
"loss": 1.0516901969909669,
|
| 122 |
+
"step": 300
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.04278182106837571,
|
| 126 |
+
"grad_norm": 10.106513977050781,
|
| 127 |
+
"learning_rate": 4.998202815772245e-05,
|
| 128 |
+
"loss": 0.8515177726745605,
|
| 129 |
+
"step": 320
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.04545568488514919,
|
| 133 |
+
"grad_norm": 3.9024946689605713,
|
| 134 |
+
"learning_rate": 4.9979681567894195e-05,
|
| 135 |
+
"loss": 0.851468849182129,
|
| 136 |
+
"step": 340
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.048129548701922674,
|
| 140 |
+
"grad_norm": 7.649858474731445,
|
| 141 |
+
"learning_rate": 4.997719109584209e-05,
|
| 142 |
+
"loss": 0.8077478408813477,
|
| 143 |
+
"step": 360
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.050803412518696155,
|
| 147 |
+
"grad_norm": 11.689629554748535,
|
| 148 |
+
"learning_rate": 4.997455675591119e-05,
|
| 149 |
+
"loss": 0.6951910495758057,
|
| 150 |
+
"step": 380
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.05347727633546964,
|
| 154 |
+
"grad_norm": 12.950127601623535,
|
| 155 |
+
"learning_rate": 4.9971778563275204e-05,
|
| 156 |
+
"loss": 0.6042420387268066,
|
| 157 |
+
"step": 400
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.05615114015224312,
|
| 161 |
+
"grad_norm": 6.282811641693115,
|
| 162 |
+
"learning_rate": 4.9968856533936436e-05,
|
| 163 |
+
"loss": 0.6595364093780518,
|
| 164 |
+
"step": 420
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.058825003969016604,
|
| 168 |
+
"grad_norm": 3.553151845932007,
|
| 169 |
+
"learning_rate": 4.99657906847257e-05,
|
| 170 |
+
"loss": 0.71713285446167,
|
| 171 |
+
"step": 440
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.061498867785790085,
|
| 175 |
+
"grad_norm": 5.715928554534912,
|
| 176 |
+
"learning_rate": 4.996258103330218e-05,
|
| 177 |
+
"loss": 0.5080134391784668,
|
| 178 |
+
"step": 460
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.06417273160256357,
|
| 182 |
+
"grad_norm": 8.924579620361328,
|
| 183 |
+
"learning_rate": 4.995922759815339e-05,
|
| 184 |
+
"loss": 0.4877929210662842,
|
| 185 |
+
"step": 480
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.06684659541933705,
|
| 189 |
+
"grad_norm": 4.2473649978637695,
|
| 190 |
+
"learning_rate": 4.995573039859501e-05,
|
| 191 |
+
"loss": 0.5352929592132568,
|
| 192 |
+
"step": 500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.06952045923611053,
|
| 196 |
+
"grad_norm": 5.880599021911621,
|
| 197 |
+
"learning_rate": 4.995208945477081e-05,
|
| 198 |
+
"loss": 0.520254373550415,
|
| 199 |
+
"step": 520
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.07219432305288401,
|
| 203 |
+
"grad_norm": 8.42399787902832,
|
| 204 |
+
"learning_rate": 4.994830478765251e-05,
|
| 205 |
+
"loss": 0.5289288520812988,
|
| 206 |
+
"step": 540
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.07486818686965749,
|
| 210 |
+
"grad_norm": 5.284440040588379,
|
| 211 |
+
"learning_rate": 4.9944376419039684e-05,
|
| 212 |
+
"loss": 0.4858880043029785,
|
| 213 |
+
"step": 560
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.07754205068643098,
|
| 217 |
+
"grad_norm": 4.662903308868408,
|
| 218 |
+
"learning_rate": 4.994030437155961e-05,
|
| 219 |
+
"loss": 0.4253716468811035,
|
| 220 |
+
"step": 580
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.08021591450320446,
|
| 224 |
+
"grad_norm": 2.9051289558410645,
|
| 225 |
+
"learning_rate": 4.993608866866718e-05,
|
| 226 |
+
"loss": 0.32270302772521975,
|
| 227 |
+
"step": 600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.08288977831997794,
|
| 231 |
+
"grad_norm": 2.804189920425415,
|
| 232 |
+
"learning_rate": 4.993172933464471e-05,
|
| 233 |
+
"loss": 0.329010009765625,
|
| 234 |
+
"step": 620
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.08556364213675142,
|
| 238 |
+
"grad_norm": 1.455569863319397,
|
| 239 |
+
"learning_rate": 4.9927226394601815e-05,
|
| 240 |
+
"loss": 0.35200178623199463,
|
| 241 |
+
"step": 640
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0882375059535249,
|
| 245 |
+
"grad_norm": 11.18322467803955,
|
| 246 |
+
"learning_rate": 4.992257987447532e-05,
|
| 247 |
+
"loss": 0.30103607177734376,
|
| 248 |
+
"step": 660
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.09091136977029839,
|
| 252 |
+
"grad_norm": 6.4625773429870605,
|
| 253 |
+
"learning_rate": 4.991778980102904e-05,
|
| 254 |
+
"loss": 0.3078582525253296,
|
| 255 |
+
"step": 680
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.09358523358707187,
|
| 259 |
+
"grad_norm": 11.98585033416748,
|
| 260 |
+
"learning_rate": 4.9912856201853644e-05,
|
| 261 |
+
"loss": 0.30774364471435545,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.09625909740384535,
|
| 266 |
+
"grad_norm": 10.583540916442871,
|
| 267 |
+
"learning_rate": 4.990777910536653e-05,
|
| 268 |
+
"loss": 0.30326850414276124,
|
| 269 |
+
"step": 720
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.09893296122061883,
|
| 273 |
+
"grad_norm": 6.04772424697876,
|
| 274 |
+
"learning_rate": 4.990255854081161e-05,
|
| 275 |
+
"loss": 0.37041378021240234,
|
| 276 |
+
"step": 740
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.10160682503739231,
|
| 280 |
+
"grad_norm": 7.155930519104004,
|
| 281 |
+
"learning_rate": 4.989719453825918e-05,
|
| 282 |
+
"loss": 0.33830211162567136,
|
| 283 |
+
"step": 760
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.10428068885416579,
|
| 287 |
+
"grad_norm": 6.669356346130371,
|
| 288 |
+
"learning_rate": 4.9891687128605744e-05,
|
| 289 |
+
"loss": 0.3225187540054321,
|
| 290 |
+
"step": 780
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.10695455267093928,
|
| 294 |
+
"grad_norm": 2.8060507774353027,
|
| 295 |
+
"learning_rate": 4.988603634357383e-05,
|
| 296 |
+
"loss": 0.27278661727905273,
|
| 297 |
+
"step": 800
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.10962841648771277,
|
| 301 |
+
"grad_norm": 4.8071699142456055,
|
| 302 |
+
"learning_rate": 4.988024221571177e-05,
|
| 303 |
+
"loss": 0.2937103986740112,
|
| 304 |
+
"step": 820
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.11230228030448625,
|
| 308 |
+
"grad_norm": 5.857818603515625,
|
| 309 |
+
"learning_rate": 4.9874304778393574e-05,
|
| 310 |
+
"loss": 0.17092934846878052,
|
| 311 |
+
"step": 840
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.11497614412125973,
|
| 315 |
+
"grad_norm": 3.4762802124023438,
|
| 316 |
+
"learning_rate": 4.9868224065818706e-05,
|
| 317 |
+
"loss": 0.1812203049659729,
|
| 318 |
+
"step": 860
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.11765000793803321,
|
| 322 |
+
"grad_norm": 6.1022186279296875,
|
| 323 |
+
"learning_rate": 4.98620001130119e-05,
|
| 324 |
+
"loss": 0.2638218879699707,
|
| 325 |
+
"step": 880
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.12032387175480669,
|
| 329 |
+
"grad_norm": 7.3963799476623535,
|
| 330 |
+
"learning_rate": 4.9855632955822916e-05,
|
| 331 |
+
"loss": 0.25242090225219727,
|
| 332 |
+
"step": 900
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.12299773557158017,
|
| 336 |
+
"grad_norm": 1.6199634075164795,
|
| 337 |
+
"learning_rate": 4.984912263092641e-05,
|
| 338 |
+
"loss": 0.2603979349136353,
|
| 339 |
+
"step": 920
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.12567159938835365,
|
| 343 |
+
"grad_norm": 8.293638229370117,
|
| 344 |
+
"learning_rate": 4.984246917582166e-05,
|
| 345 |
+
"loss": 0.2023364543914795,
|
| 346 |
+
"step": 940
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.12834546320512713,
|
| 350 |
+
"grad_norm": 8.61099910736084,
|
| 351 |
+
"learning_rate": 4.9835672628832366e-05,
|
| 352 |
+
"loss": 0.16875293254852294,
|
| 353 |
+
"step": 960
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.1310193270219006,
|
| 357 |
+
"grad_norm": 0.5758953094482422,
|
| 358 |
+
"learning_rate": 4.9828733029106434e-05,
|
| 359 |
+
"loss": 0.24272210597991944,
|
| 360 |
+
"step": 980
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.1336931908386741,
|
| 364 |
+
"grad_norm": 6.830381870269775,
|
| 365 |
+
"learning_rate": 4.982165041661575e-05,
|
| 366 |
+
"loss": 0.1900513768196106,
|
| 367 |
+
"step": 1000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.13636705465544757,
|
| 371 |
+
"grad_norm": 3.979989528656006,
|
| 372 |
+
"learning_rate": 4.981442483215595e-05,
|
| 373 |
+
"loss": 0.23804061412811278,
|
| 374 |
+
"step": 1020
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.13904091847222105,
|
| 378 |
+
"grad_norm": 3.480762004852295,
|
| 379 |
+
"learning_rate": 4.98070563173462e-05,
|
| 380 |
+
"loss": 0.19694632291793823,
|
| 381 |
+
"step": 1040
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.14171478228899453,
|
| 385 |
+
"grad_norm": 2.7838668823242188,
|
| 386 |
+
"learning_rate": 4.979954491462892e-05,
|
| 387 |
+
"loss": 0.20703573226928712,
|
| 388 |
+
"step": 1060
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.14438864610576801,
|
| 392 |
+
"grad_norm": 4.724070072174072,
|
| 393 |
+
"learning_rate": 4.979189066726955e-05,
|
| 394 |
+
"loss": 0.135595703125,
|
| 395 |
+
"step": 1080
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.1470625099225415,
|
| 399 |
+
"grad_norm": 6.755579471588135,
|
| 400 |
+
"learning_rate": 4.978409361935636e-05,
|
| 401 |
+
"loss": 0.20947463512420655,
|
| 402 |
+
"step": 1100
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.14973637373931498,
|
| 406 |
+
"grad_norm": 2.989082098007202,
|
| 407 |
+
"learning_rate": 4.9776153815800075e-05,
|
| 408 |
+
"loss": 0.15855485200881958,
|
| 409 |
+
"step": 1120
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.15241023755608848,
|
| 413 |
+
"grad_norm": 2.310991048812866,
|
| 414 |
+
"learning_rate": 4.976807130233375e-05,
|
| 415 |
+
"loss": 0.1401826858520508,
|
| 416 |
+
"step": 1140
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.15508410137286197,
|
| 420 |
+
"grad_norm": 3.802248477935791,
|
| 421 |
+
"learning_rate": 4.975984612551243e-05,
|
| 422 |
+
"loss": 0.17620396614074707,
|
| 423 |
+
"step": 1160
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.15775796518963545,
|
| 427 |
+
"grad_norm": 2.0629191398620605,
|
| 428 |
+
"learning_rate": 4.975147833271288e-05,
|
| 429 |
+
"loss": 0.21862537860870362,
|
| 430 |
+
"step": 1180
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.16043182900640893,
|
| 434 |
+
"grad_norm": 2.2114040851593018,
|
| 435 |
+
"learning_rate": 4.9742967972133335e-05,
|
| 436 |
+
"loss": 0.15507786273956298,
|
| 437 |
+
"step": 1200
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.1631056928231824,
|
| 441 |
+
"grad_norm": 10.362643241882324,
|
| 442 |
+
"learning_rate": 4.973431509279323e-05,
|
| 443 |
+
"loss": 0.15983071327209472,
|
| 444 |
+
"step": 1220
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.1657795566399559,
|
| 448 |
+
"grad_norm": 0.3497254252433777,
|
| 449 |
+
"learning_rate": 4.972551974453287e-05,
|
| 450 |
+
"loss": 0.20826480388641358,
|
| 451 |
+
"step": 1240
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.16845342045672937,
|
| 455 |
+
"grad_norm": 0.3897332549095154,
|
| 456 |
+
"learning_rate": 4.971658197801322e-05,
|
| 457 |
+
"loss": 0.12974958419799804,
|
| 458 |
+
"step": 1260
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.17112728427350285,
|
| 462 |
+
"grad_norm": 6.043999195098877,
|
| 463 |
+
"learning_rate": 4.9707501844715554e-05,
|
| 464 |
+
"loss": 0.1621025800704956,
|
| 465 |
+
"step": 1280
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.17380114809027633,
|
| 469 |
+
"grad_norm": 1.7294467687606812,
|
| 470 |
+
"learning_rate": 4.969827939694115e-05,
|
| 471 |
+
"loss": 0.17147176265716552,
|
| 472 |
+
"step": 1300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.1764750119070498,
|
| 476 |
+
"grad_norm": 0.8541621565818787,
|
| 477 |
+
"learning_rate": 4.968891468781105e-05,
|
| 478 |
+
"loss": 0.1330248236656189,
|
| 479 |
+
"step": 1320
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.1791488757238233,
|
| 483 |
+
"grad_norm": 2.197866439819336,
|
| 484 |
+
"learning_rate": 4.967940777126569e-05,
|
| 485 |
+
"loss": 0.12178658246994019,
|
| 486 |
+
"step": 1340
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.18182273954059677,
|
| 490 |
+
"grad_norm": 0.8711172342300415,
|
| 491 |
+
"learning_rate": 4.9669758702064636e-05,
|
| 492 |
+
"loss": 0.19501737356185914,
|
| 493 |
+
"step": 1360
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.18449660335737025,
|
| 497 |
+
"grad_norm": 5.356688022613525,
|
| 498 |
+
"learning_rate": 4.965996753578623e-05,
|
| 499 |
+
"loss": 0.1602308750152588,
|
| 500 |
+
"step": 1380
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.18717046717414373,
|
| 504 |
+
"grad_norm": 13.699594497680664,
|
| 505 |
+
"learning_rate": 4.9650034328827305e-05,
|
| 506 |
+
"loss": 0.12051005363464355,
|
| 507 |
+
"step": 1400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.18984433099091721,
|
| 511 |
+
"grad_norm": 1.738416314125061,
|
| 512 |
+
"learning_rate": 4.963995913840284e-05,
|
| 513 |
+
"loss": 0.1177564263343811,
|
| 514 |
+
"step": 1420
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.1925181948076907,
|
| 518 |
+
"grad_norm": 5.713410377502441,
|
| 519 |
+
"learning_rate": 4.9629742022545623e-05,
|
| 520 |
+
"loss": 0.19124844074249267,
|
| 521 |
+
"step": 1440
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.19519205862446418,
|
| 525 |
+
"grad_norm": 4.171947479248047,
|
| 526 |
+
"learning_rate": 4.961938304010595e-05,
|
| 527 |
+
"loss": 0.12269804477691651,
|
| 528 |
+
"step": 1460
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.19786592244123766,
|
| 532 |
+
"grad_norm": 2.8535916805267334,
|
| 533 |
+
"learning_rate": 4.9608882250751245e-05,
|
| 534 |
+
"loss": 0.0892150342464447,
|
| 535 |
+
"step": 1480
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.20053978625801114,
|
| 539 |
+
"grad_norm": 0.6551859378814697,
|
| 540 |
+
"learning_rate": 4.959823971496574e-05,
|
| 541 |
+
"loss": 0.10962753295898438,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.20321365007478462,
|
| 546 |
+
"grad_norm": 6.525365352630615,
|
| 547 |
+
"learning_rate": 4.9587455494050136e-05,
|
| 548 |
+
"loss": 0.10057367086410522,
|
| 549 |
+
"step": 1520
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.2058875138915581,
|
| 553 |
+
"grad_norm": 4.142037391662598,
|
| 554 |
+
"learning_rate": 4.9576529650121214e-05,
|
| 555 |
+
"loss": 0.1188935399055481,
|
| 556 |
+
"step": 1540
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.20856137770833158,
|
| 560 |
+
"grad_norm": 1.3209469318389893,
|
| 561 |
+
"learning_rate": 4.956546224611152e-05,
|
| 562 |
+
"loss": 0.10144028663635254,
|
| 563 |
+
"step": 1560
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.21123524152510506,
|
| 567 |
+
"grad_norm": 2.7724432945251465,
|
| 568 |
+
"learning_rate": 4.9554253345768965e-05,
|
| 569 |
+
"loss": 0.09022048711776734,
|
| 570 |
+
"step": 1580
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.21390910534187857,
|
| 574 |
+
"grad_norm": 1.7022371292114258,
|
| 575 |
+
"learning_rate": 4.9542903013656486e-05,
|
| 576 |
+
"loss": 0.13176963329315186,
|
| 577 |
+
"step": 1600
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.21658296915865205,
|
| 581 |
+
"grad_norm": 4.1148681640625,
|
| 582 |
+
"learning_rate": 4.9531411315151654e-05,
|
| 583 |
+
"loss": 0.08941383361816406,
|
| 584 |
+
"step": 1620
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.21925683297542553,
|
| 588 |
+
"grad_norm": 1.5328305959701538,
|
| 589 |
+
"learning_rate": 4.951977831644632e-05,
|
| 590 |
+
"loss": 0.12352118492126465,
|
| 591 |
+
"step": 1640
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.221930696792199,
|
| 595 |
+
"grad_norm": 2.8182177543640137,
|
| 596 |
+
"learning_rate": 4.95080040845462e-05,
|
| 597 |
+
"loss": 0.12917338609695433,
|
| 598 |
+
"step": 1660
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.2246045606089725,
|
| 602 |
+
"grad_norm": 2.4821066856384277,
|
| 603 |
+
"learning_rate": 4.949608868727053e-05,
|
| 604 |
+
"loss": 0.12498313188552856,
|
| 605 |
+
"step": 1680
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.22727842442574597,
|
| 609 |
+
"grad_norm": 2.292102336883545,
|
| 610 |
+
"learning_rate": 4.948403219325163e-05,
|
| 611 |
+
"loss": 0.10659204721450806,
|
| 612 |
+
"step": 1700
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.22995228824251945,
|
| 616 |
+
"grad_norm": 0.4930734634399414,
|
| 617 |
+
"learning_rate": 4.947183467193456e-05,
|
| 618 |
+
"loss": 0.0782310426235199,
|
| 619 |
+
"step": 1720
|
| 620 |
+
}
|
| 621 |
+
],
|
| 622 |
+
"logging_steps": 20,
|
| 623 |
+
"max_steps": 26180,
|
| 624 |
+
"num_input_tokens_seen": 0,
|
| 625 |
+
"num_train_epochs": 4,
|
| 626 |
+
"save_steps": 1000000000,
|
| 627 |
+
"stateful_callbacks": {
|
| 628 |
+
"TrainerControl": {
|
| 629 |
+
"args": {
|
| 630 |
+
"should_epoch_stop": false,
|
| 631 |
+
"should_evaluate": false,
|
| 632 |
+
"should_log": false,
|
| 633 |
+
"should_save": true,
|
| 634 |
+
"should_training_stop": false
|
| 635 |
+
},
|
| 636 |
+
"attributes": {}
|
| 637 |
+
}
|
| 638 |
+
},
|
| 639 |
+
"total_flos": 6638921592496128.0,
|
| 640 |
+
"train_batch_size": 1,
|
| 641 |
+
"trial_name": null,
|
| 642 |
+
"trial_params": null
|
| 643 |
+
}
|