Uploaded Model
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.875,
|
| 4 |
+
"eval_f1": 0.875,
|
| 5 |
+
"eval_loss": 0.3666810989379883,
|
| 6 |
+
"eval_precision": 0.875,
|
| 7 |
+
"eval_recall": 0.875,
|
| 8 |
+
"eval_runtime": 1.1147,
|
| 9 |
+
"eval_samples_per_second": 35.884,
|
| 10 |
+
"eval_steps_per_second": 4.485,
|
| 11 |
+
"total_flos": 5.05524198703104e+16,
|
| 12 |
+
"train_loss": 0.14953994750976562,
|
| 13 |
+
"train_runtime": 18.1531,
|
| 14 |
+
"train_samples_per_second": 35.256,
|
| 15 |
+
"train_steps_per_second": 2.203
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.875,
|
| 4 |
+
"eval_f1": 0.875,
|
| 5 |
+
"eval_loss": 0.3666810989379883,
|
| 6 |
+
"eval_precision": 0.875,
|
| 7 |
+
"eval_recall": 0.875,
|
| 8 |
+
"eval_runtime": 1.1147,
|
| 9 |
+
"eval_samples_per_second": 35.884,
|
| 10 |
+
"eval_steps_per_second": 4.485
|
| 11 |
+
}
|
runs/Jun27_11-13-11_a8668c8a8ada/events.out.tfevents.1687864439.a8668c8a8ada.936.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fbf1f5978e8e96b9776f2b32217b67c1f0b39a8e46b43fda25efd435bf3a5472
|
| 3 |
+
size 551
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"total_flos": 5.05524198703104e+16,
|
| 4 |
+
"train_loss": 0.14953994750976562,
|
| 5 |
+
"train_runtime": 18.1531,
|
| 6 |
+
"train_samples_per_second": 35.256,
|
| 7 |
+
"train_steps_per_second": 2.203
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 40,
|
| 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": 0.00015000000000000001,
|
| 13 |
+
"loss": 0.4049,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 2.0,
|
| 18 |
+
"learning_rate": 0.0001,
|
| 19 |
+
"loss": 0.1204,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 3.0,
|
| 24 |
+
"learning_rate": 5e-05,
|
| 25 |
+
"loss": 0.0492,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 4.0,
|
| 30 |
+
"learning_rate": 0.0,
|
| 31 |
+
"loss": 0.0236,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 4.0,
|
| 36 |
+
"step": 40,
|
| 37 |
+
"total_flos": 5.05524198703104e+16,
|
| 38 |
+
"train_loss": 0.14953994750976562,
|
| 39 |
+
"train_runtime": 18.1531,
|
| 40 |
+
"train_samples_per_second": 35.256,
|
| 41 |
+
"train_steps_per_second": 2.203
|
| 42 |
+
}
|
| 43 |
+
],
|
| 44 |
+
"max_steps": 40,
|
| 45 |
+
"num_train_epochs": 4,
|
| 46 |
+
"total_flos": 5.05524198703104e+16,
|
| 47 |
+
"trial_name": null,
|
| 48 |
+
"trial_params": null
|
| 49 |
+
}
|