End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +61 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"eval_loss": 2.175971508026123,
|
| 4 |
+
"eval_runtime": 4.3357,
|
| 5 |
+
"eval_samples": 1991,
|
| 6 |
+
"eval_samples_per_second": 459.213,
|
| 7 |
+
"eval_steps_per_second": 14.531,
|
| 8 |
+
"train_loss": 1.5467751831825067,
|
| 9 |
+
"train_runtime": 653.0728,
|
| 10 |
+
"train_samples": 15931,
|
| 11 |
+
"train_samples_per_second": 170.757,
|
| 12 |
+
"train_steps_per_second": 5.338
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"eval_loss": 2.175971508026123,
|
| 4 |
+
"eval_runtime": 4.3357,
|
| 5 |
+
"eval_samples": 1991,
|
| 6 |
+
"eval_samples_per_second": 459.213,
|
| 7 |
+
"eval_steps_per_second": 14.531
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"train_loss": 1.5467751831825067,
|
| 4 |
+
"train_runtime": 653.0728,
|
| 5 |
+
"train_samples": 15931,
|
| 6 |
+
"train_samples_per_second": 170.757,
|
| 7 |
+
"train_steps_per_second": 5.338
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 3486,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 4.2828456683878374e-05,
|
| 13 |
+
"loss": 2.2005,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 2.01,
|
| 18 |
+
"learning_rate": 3.5656913367756745e-05,
|
| 19 |
+
"loss": 1.8288,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 3.01,
|
| 24 |
+
"learning_rate": 2.8485370051635113e-05,
|
| 25 |
+
"loss": 1.6132,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 4.02,
|
| 30 |
+
"learning_rate": 2.1313826735513484e-05,
|
| 31 |
+
"loss": 1.4468,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 5.02,
|
| 36 |
+
"learning_rate": 1.4142283419391852e-05,
|
| 37 |
+
"loss": 1.3287,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 6.02,
|
| 42 |
+
"learning_rate": 6.970740103270223e-06,
|
| 43 |
+
"loss": 1.2313,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 7.0,
|
| 48 |
+
"step": 3486,
|
| 49 |
+
"total_flos": 4463221452318720.0,
|
| 50 |
+
"train_loss": 1.5467751831825067,
|
| 51 |
+
"train_runtime": 653.0728,
|
| 52 |
+
"train_samples_per_second": 170.757,
|
| 53 |
+
"train_steps_per_second": 5.338
|
| 54 |
+
}
|
| 55 |
+
],
|
| 56 |
+
"max_steps": 3486,
|
| 57 |
+
"num_train_epochs": 7,
|
| 58 |
+
"total_flos": 4463221452318720.0,
|
| 59 |
+
"trial_name": null,
|
| 60 |
+
"trial_params": null
|
| 61 |
+
}
|