Upload 2 files
Browse files- pytorch_model.bin +3 -0
- trainer_state.json +100 -0
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e38e3bb10b531f55268895b3eebef94a85001f9113c9b4aa9d13abd97422dcfe
|
| 3 |
+
size 891702929
|
trainer_state.json
ADDED
|
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.303820013999939,
|
| 3 |
+
"epoch": 1.0,
|
| 4 |
+
"global_step": 6230,
|
| 5 |
+
"is_hyper_param_search": false,
|
| 6 |
+
"is_local_process_zero": true,
|
| 7 |
+
"is_world_process_zero": true,
|
| 8 |
+
"log_history": [
|
| 9 |
+
{
|
| 10 |
+
"epoch": 0.08,
|
| 11 |
+
"learning_rate": 9.598715890850723e-06,
|
| 12 |
+
"loss": 1.528,
|
| 13 |
+
"step": 500
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"epoch": 0.16,
|
| 17 |
+
"learning_rate": 9.197431781701446e-06,
|
| 18 |
+
"loss": 1.3696,
|
| 19 |
+
"step": 1000
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 0.24,
|
| 23 |
+
"learning_rate": 8.796147672552167e-06,
|
| 24 |
+
"loss": 1.2899,
|
| 25 |
+
"step": 1500
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.32,
|
| 29 |
+
"learning_rate": 8.39486356340289e-06,
|
| 30 |
+
"loss": 1.2996,
|
| 31 |
+
"step": 2000
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.4,
|
| 35 |
+
"learning_rate": 7.993579454253612e-06,
|
| 36 |
+
"loss": 1.2637,
|
| 37 |
+
"step": 2500
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.48,
|
| 41 |
+
"learning_rate": 7.592295345104335e-06,
|
| 42 |
+
"loss": 1.2596,
|
| 43 |
+
"step": 3000
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.56,
|
| 47 |
+
"learning_rate": 7.191011235955056e-06,
|
| 48 |
+
"loss": 1.2513,
|
| 49 |
+
"step": 3500
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.64,
|
| 53 |
+
"learning_rate": 6.789727126805778e-06,
|
| 54 |
+
"loss": 1.2211,
|
| 55 |
+
"step": 4000
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.72,
|
| 59 |
+
"learning_rate": 6.3884430176565015e-06,
|
| 60 |
+
"loss": 1.2108,
|
| 61 |
+
"step": 4500
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.8,
|
| 65 |
+
"learning_rate": 5.987158908507223e-06,
|
| 66 |
+
"loss": 1.1853,
|
| 67 |
+
"step": 5000
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.88,
|
| 71 |
+
"learning_rate": 5.585874799357945e-06,
|
| 72 |
+
"loss": 1.2138,
|
| 73 |
+
"step": 5500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.96,
|
| 77 |
+
"learning_rate": 5.1845906902086685e-06,
|
| 78 |
+
"loss": 1.1879,
|
| 79 |
+
"step": 6000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.0,
|
| 83 |
+
"eval_gen_len": 18.758,
|
| 84 |
+
"eval_loss": 1.303820013999939,
|
| 85 |
+
"eval_rouge1": 38.7078,
|
| 86 |
+
"eval_rouge2": 14.3583,
|
| 87 |
+
"eval_rougeL": 32.2229,
|
| 88 |
+
"eval_rougeLsum": 34.3647,
|
| 89 |
+
"eval_runtime": 739.1192,
|
| 90 |
+
"eval_samples_per_second": 2.029,
|
| 91 |
+
"eval_steps_per_second": 1.015,
|
| 92 |
+
"step": 6230
|
| 93 |
+
}
|
| 94 |
+
],
|
| 95 |
+
"max_steps": 12460,
|
| 96 |
+
"num_train_epochs": 2,
|
| 97 |
+
"total_flos": 1.48195611648e+16,
|
| 98 |
+
"trial_name": null,
|
| 99 |
+
"trial_params": null
|
| 100 |
+
}
|