Commit
•
44b996c
1
Parent(s):
3432431
Upload 5 files
Browse files- all_results.json +24 -0
- eval_results.json +11 -0
- predict_results.json +10 -0
- train_results.json +8 -0
- trainer_state.json +59 -0
all_results.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 32.0,
|
| 3 |
+
"eval_bleu": 9.6508,
|
| 4 |
+
"eval_gen_len": 47.087,
|
| 5 |
+
"eval_loss": 2.915083885192871,
|
| 6 |
+
"eval_meteor": 0.2957,
|
| 7 |
+
"eval_runtime": 450.4115,
|
| 8 |
+
"eval_samples": 1816,
|
| 9 |
+
"eval_samples_per_second": 4.032,
|
| 10 |
+
"eval_steps_per_second": 0.337,
|
| 11 |
+
"predict_bleu": 9.8708,
|
| 12 |
+
"predict_gen_len": 47.7877,
|
| 13 |
+
"predict_loss": 2.9007568359375,
|
| 14 |
+
"predict_meteor": 0.2982,
|
| 15 |
+
"predict_runtime": 564.6061,
|
| 16 |
+
"predict_samples": 2270,
|
| 17 |
+
"predict_samples_per_second": 4.021,
|
| 18 |
+
"predict_steps_per_second": 0.337,
|
| 19 |
+
"train_loss": 2.1759391255898053,
|
| 20 |
+
"train_runtime": 8987.7149,
|
| 21 |
+
"train_samples": 7263,
|
| 22 |
+
"train_samples_per_second": 25.859,
|
| 23 |
+
"train_steps_per_second": 1.079
|
| 24 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 32.0,
|
| 3 |
+
"eval_bleu": 9.6508,
|
| 4 |
+
"eval_gen_len": 47.087,
|
| 5 |
+
"eval_loss": 2.915083885192871,
|
| 6 |
+
"eval_meteor": 0.2957,
|
| 7 |
+
"eval_runtime": 450.4115,
|
| 8 |
+
"eval_samples": 1816,
|
| 9 |
+
"eval_samples_per_second": 4.032,
|
| 10 |
+
"eval_steps_per_second": 0.337
|
| 11 |
+
}
|
predict_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"predict_bleu": 9.8708,
|
| 3 |
+
"predict_gen_len": 47.7877,
|
| 4 |
+
"predict_loss": 2.9007568359375,
|
| 5 |
+
"predict_meteor": 0.2982,
|
| 6 |
+
"predict_runtime": 564.6061,
|
| 7 |
+
"predict_samples": 2270,
|
| 8 |
+
"predict_samples_per_second": 4.021,
|
| 9 |
+
"predict_steps_per_second": 0.337
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 32.0,
|
| 3 |
+
"train_loss": 2.1759391255898053,
|
| 4 |
+
"train_runtime": 8987.7149,
|
| 5 |
+
"train_samples": 7263,
|
| 6 |
+
"train_samples_per_second": 25.859,
|
| 7 |
+
"train_steps_per_second": 1.079
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 32.0,
|
| 5 |
+
"global_step": 9696,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 11.24,
|
| 12 |
+
"learning_rate": 1.2976485148514853e-05,
|
| 13 |
+
"loss": 2.9964,
|
| 14 |
+
"step": 3406
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 11.24,
|
| 18 |
+
"eval_bleu": 8.3668,
|
| 19 |
+
"eval_gen_len": 49.1713,
|
| 20 |
+
"eval_loss": 2.9263081550598145,
|
| 21 |
+
"eval_meteor": 0.2788,
|
| 22 |
+
"eval_runtime": 534.66,
|
| 23 |
+
"eval_samples_per_second": 3.397,
|
| 24 |
+
"eval_steps_per_second": 0.284,
|
| 25 |
+
"step": 3406
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 22.48,
|
| 29 |
+
"learning_rate": 5.9529702970297035e-06,
|
| 30 |
+
"loss": 1.8732,
|
| 31 |
+
"step": 6812
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 22.48,
|
| 35 |
+
"eval_bleu": 9.2938,
|
| 36 |
+
"eval_gen_len": 48.9251,
|
| 37 |
+
"eval_loss": 2.9048359394073486,
|
| 38 |
+
"eval_meteor": 0.2971,
|
| 39 |
+
"eval_runtime": 526.9982,
|
| 40 |
+
"eval_samples_per_second": 3.446,
|
| 41 |
+
"eval_steps_per_second": 0.288,
|
| 42 |
+
"step": 6812
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 32.0,
|
| 46 |
+
"step": 9696,
|
| 47 |
+
"total_flos": 2.37387465916416e+16,
|
| 48 |
+
"train_loss": 2.1759391255898053,
|
| 49 |
+
"train_runtime": 8987.7149,
|
| 50 |
+
"train_samples_per_second": 25.859,
|
| 51 |
+
"train_steps_per_second": 1.079
|
| 52 |
+
}
|
| 53 |
+
],
|
| 54 |
+
"max_steps": 9696,
|
| 55 |
+
"num_train_epochs": 32,
|
| 56 |
+
"total_flos": 2.37387465916416e+16,
|
| 57 |
+
"trial_name": null,
|
| 58 |
+
"trial_params": null
|
| 59 |
+
}
|