delete unnecessary files
Browse files- all_results.json +0 -34
- eval_results.json +0 -13
- test_generations.txt +0 -0
- test_results.json +0 -12
- train_results.json +0 -14
- trainer_state.json +0 -239
- training_args.bin +0 -3
all_results.json
DELETED
|
@@ -1,34 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"epoch": 10.0,
|
| 3 |
-
"eval_bleu": 20.563,
|
| 4 |
-
"eval_gen_len": 28.4174,
|
| 5 |
-
"eval_loss": 1.523200511932373,
|
| 6 |
-
"eval_mem_cpu_alloc_delta": 11807807,
|
| 7 |
-
"eval_mem_cpu_peaked_delta": 8618611,
|
| 8 |
-
"eval_mem_gpu_alloc_delta": 0,
|
| 9 |
-
"eval_mem_gpu_peaked_delta": 2386463232,
|
| 10 |
-
"eval_runtime": 222.6162,
|
| 11 |
-
"eval_samples": 2753,
|
| 12 |
-
"eval_samples_per_second": 12.367,
|
| 13 |
-
"init_mem_cpu_alloc_delta": 2595413,
|
| 14 |
-
"init_mem_cpu_peaked_delta": 18258,
|
| 15 |
-
"init_mem_gpu_alloc_delta": 242026496,
|
| 16 |
-
"init_mem_gpu_peaked_delta": 0,
|
| 17 |
-
"test_bleu": 37.3827,
|
| 18 |
-
"test_gen_len": 30.7292,
|
| 19 |
-
"test_loss": 1.712681531906128,
|
| 20 |
-
"test_mem_cpu_alloc_delta": 9661242,
|
| 21 |
-
"test_mem_cpu_peaked_delta": 7664493,
|
| 22 |
-
"test_mem_gpu_alloc_delta": 0,
|
| 23 |
-
"test_mem_gpu_peaked_delta": 2097349120,
|
| 24 |
-
"test_runtime": 191.4724,
|
| 25 |
-
"test_samples": 2345,
|
| 26 |
-
"test_samples_per_second": 12.247,
|
| 27 |
-
"train_mem_cpu_alloc_delta": 1526492,
|
| 28 |
-
"train_mem_cpu_peaked_delta": 67831945,
|
| 29 |
-
"train_mem_gpu_alloc_delta": 968303104,
|
| 30 |
-
"train_mem_gpu_peaked_delta": 2622328832,
|
| 31 |
-
"train_runtime": 352.0036,
|
| 32 |
-
"train_samples": 28883,
|
| 33 |
-
"train_samples_per_second": 51.306
|
| 34 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
eval_results.json
DELETED
|
@@ -1,13 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"epoch": 10.0,
|
| 3 |
-
"eval_bleu": 20.563,
|
| 4 |
-
"eval_gen_len": 28.4174,
|
| 5 |
-
"eval_loss": 1.523200511932373,
|
| 6 |
-
"eval_mem_cpu_alloc_delta": 11807807,
|
| 7 |
-
"eval_mem_cpu_peaked_delta": 8618611,
|
| 8 |
-
"eval_mem_gpu_alloc_delta": 0,
|
| 9 |
-
"eval_mem_gpu_peaked_delta": 2386463232,
|
| 10 |
-
"eval_runtime": 222.6162,
|
| 11 |
-
"eval_samples": 2753,
|
| 12 |
-
"eval_samples_per_second": 12.367
|
| 13 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
test_generations.txt
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
test_results.json
DELETED
|
@@ -1,12 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"test_bleu": 37.3827,
|
| 3 |
-
"test_gen_len": 30.7292,
|
| 4 |
-
"test_loss": 1.712681531906128,
|
| 5 |
-
"test_mem_cpu_alloc_delta": 9661242,
|
| 6 |
-
"test_mem_cpu_peaked_delta": 7664493,
|
| 7 |
-
"test_mem_gpu_alloc_delta": 0,
|
| 8 |
-
"test_mem_gpu_peaked_delta": 2097349120,
|
| 9 |
-
"test_runtime": 191.4724,
|
| 10 |
-
"test_samples": 2345,
|
| 11 |
-
"test_samples_per_second": 12.247
|
| 12 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
train_results.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"epoch": 10.0,
|
| 3 |
-
"init_mem_cpu_alloc_delta": 2595413,
|
| 4 |
-
"init_mem_cpu_peaked_delta": 18258,
|
| 5 |
-
"init_mem_gpu_alloc_delta": 242026496,
|
| 6 |
-
"init_mem_gpu_peaked_delta": 0,
|
| 7 |
-
"train_mem_cpu_alloc_delta": 1526492,
|
| 8 |
-
"train_mem_cpu_peaked_delta": 67831945,
|
| 9 |
-
"train_mem_gpu_alloc_delta": 968303104,
|
| 10 |
-
"train_mem_gpu_peaked_delta": 2622328832,
|
| 11 |
-
"train_runtime": 352.0036,
|
| 12 |
-
"train_samples": 28883,
|
| 13 |
-
"train_samples_per_second": 51.306
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
trainer_state.json
DELETED
|
@@ -1,239 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 10.0,
|
| 5 |
-
"global_step": 18060,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.28,
|
| 12 |
-
"learning_rate": 4.8615725359911405e-05,
|
| 13 |
-
"loss": 2.4674,
|
| 14 |
-
"step": 500
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.55,
|
| 18 |
-
"learning_rate": 4.7231450719822815e-05,
|
| 19 |
-
"loss": 2.2643,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.83,
|
| 24 |
-
"learning_rate": 4.5847176079734225e-05,
|
| 25 |
-
"loss": 2.1852,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 1.11,
|
| 30 |
-
"learning_rate": 4.446290143964563e-05,
|
| 31 |
-
"loss": 2.1008,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 1.38,
|
| 36 |
-
"learning_rate": 4.307862679955704e-05,
|
| 37 |
-
"loss": 2.0497,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 1.66,
|
| 42 |
-
"learning_rate": 4.169435215946844e-05,
|
| 43 |
-
"loss": 2.0143,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 1.94,
|
| 48 |
-
"learning_rate": 4.0310077519379843e-05,
|
| 49 |
-
"loss": 1.9799,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 2.21,
|
| 54 |
-
"learning_rate": 3.892580287929125e-05,
|
| 55 |
-
"loss": 1.9401,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 2.49,
|
| 60 |
-
"learning_rate": 3.754152823920266e-05,
|
| 61 |
-
"loss": 1.9205,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 2.77,
|
| 66 |
-
"learning_rate": 3.6157253599114066e-05,
|
| 67 |
-
"loss": 1.9005,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 3.05,
|
| 72 |
-
"learning_rate": 3.477297895902547e-05,
|
| 73 |
-
"loss": 1.8781,
|
| 74 |
-
"step": 5500
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 3.32,
|
| 78 |
-
"learning_rate": 3.338870431893688e-05,
|
| 79 |
-
"loss": 1.8538,
|
| 80 |
-
"step": 6000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 3.6,
|
| 84 |
-
"learning_rate": 3.200442967884829e-05,
|
| 85 |
-
"loss": 1.8398,
|
| 86 |
-
"step": 6500
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 3.88,
|
| 90 |
-
"learning_rate": 3.062015503875969e-05,
|
| 91 |
-
"loss": 1.822,
|
| 92 |
-
"step": 7000
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 4.15,
|
| 96 |
-
"learning_rate": 2.92358803986711e-05,
|
| 97 |
-
"loss": 1.8072,
|
| 98 |
-
"step": 7500
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 4.43,
|
| 102 |
-
"learning_rate": 2.7851605758582504e-05,
|
| 103 |
-
"loss": 1.8134,
|
| 104 |
-
"step": 8000
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 4.71,
|
| 108 |
-
"learning_rate": 2.646733111849391e-05,
|
| 109 |
-
"loss": 1.7785,
|
| 110 |
-
"step": 8500
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 4.98,
|
| 114 |
-
"learning_rate": 2.5083056478405313e-05,
|
| 115 |
-
"loss": 1.7787,
|
| 116 |
-
"step": 9000
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 5.26,
|
| 120 |
-
"learning_rate": 2.3698781838316723e-05,
|
| 121 |
-
"loss": 1.7587,
|
| 122 |
-
"step": 9500
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 5.54,
|
| 126 |
-
"learning_rate": 2.231450719822813e-05,
|
| 127 |
-
"loss": 1.7583,
|
| 128 |
-
"step": 10000
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 5.81,
|
| 132 |
-
"learning_rate": 2.0930232558139536e-05,
|
| 133 |
-
"loss": 1.7558,
|
| 134 |
-
"step": 10500
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 6.09,
|
| 138 |
-
"learning_rate": 1.9545957918050942e-05,
|
| 139 |
-
"loss": 1.7438,
|
| 140 |
-
"step": 11000
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 6.37,
|
| 144 |
-
"learning_rate": 1.816168327796235e-05,
|
| 145 |
-
"loss": 1.7299,
|
| 146 |
-
"step": 11500
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 6.64,
|
| 150 |
-
"learning_rate": 1.6777408637873755e-05,
|
| 151 |
-
"loss": 1.753,
|
| 152 |
-
"step": 12000
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 6.92,
|
| 156 |
-
"learning_rate": 1.539313399778516e-05,
|
| 157 |
-
"loss": 1.6899,
|
| 158 |
-
"step": 12500
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 7.2,
|
| 162 |
-
"learning_rate": 1.4008859357696569e-05,
|
| 163 |
-
"loss": 1.7248,
|
| 164 |
-
"step": 13000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 7.48,
|
| 168 |
-
"learning_rate": 1.2624584717607974e-05,
|
| 169 |
-
"loss": 1.6908,
|
| 170 |
-
"step": 13500
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 7.75,
|
| 174 |
-
"learning_rate": 1.1240310077519382e-05,
|
| 175 |
-
"loss": 1.6977,
|
| 176 |
-
"step": 14000
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 8.03,
|
| 180 |
-
"learning_rate": 9.856035437430786e-06,
|
| 181 |
-
"loss": 1.6908,
|
| 182 |
-
"step": 14500
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 8.31,
|
| 186 |
-
"learning_rate": 8.471760797342193e-06,
|
| 187 |
-
"loss": 1.7079,
|
| 188 |
-
"step": 15000
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 8.58,
|
| 192 |
-
"learning_rate": 7.087486157253599e-06,
|
| 193 |
-
"loss": 1.6921,
|
| 194 |
-
"step": 15500
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 8.86,
|
| 198 |
-
"learning_rate": 5.703211517165006e-06,
|
| 199 |
-
"loss": 1.6902,
|
| 200 |
-
"step": 16000
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 9.14,
|
| 204 |
-
"learning_rate": 4.318936877076412e-06,
|
| 205 |
-
"loss": 1.6771,
|
| 206 |
-
"step": 16500
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 9.41,
|
| 210 |
-
"learning_rate": 2.9346622369878186e-06,
|
| 211 |
-
"loss": 1.6853,
|
| 212 |
-
"step": 17000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 9.69,
|
| 216 |
-
"learning_rate": 1.550387596899225e-06,
|
| 217 |
-
"loss": 1.6893,
|
| 218 |
-
"step": 17500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 9.97,
|
| 222 |
-
"learning_rate": 1.6611295681063123e-07,
|
| 223 |
-
"loss": 1.6818,
|
| 224 |
-
"step": 18000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 10.0,
|
| 228 |
-
"step": 18060,
|
| 229 |
-
"total_flos": 1.1559596046910464e+16,
|
| 230 |
-
"train_runtime": 352.0036,
|
| 231 |
-
"train_samples_per_second": 51.306
|
| 232 |
-
}
|
| 233 |
-
],
|
| 234 |
-
"max_steps": 18060,
|
| 235 |
-
"num_train_epochs": 10,
|
| 236 |
-
"total_flos": 1.1559596046910464e+16,
|
| 237 |
-
"trial_name": null,
|
| 238 |
-
"trial_params": null
|
| 239 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d1890a229aea27e81568a5b8407044e323a2e668222918f7d20c8b288bbc676e
|
| 3 |
-
size 2351
|
|
|
|
|
|
|
|
|
|
|
|