End of training
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Aug14_16-52-14_Phoenix/events.out.tfevents.1723658464.Phoenix.9472.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1030 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"Validation_accuracy": 0.8372543455125931,
|
| 3 |
+
"Validation_f1": 0.8919393816628117,
|
| 4 |
+
"Validation_loss": 0.9663852453231812,
|
| 5 |
+
"Validation_precision": 0.9221433233005075,
|
| 6 |
+
"Validation_recall": 0.8636512917822216,
|
| 7 |
+
"Validation_runtime": 717.1984,
|
| 8 |
+
"Validation_samples_per_second": 196.529,
|
| 9 |
+
"Validation_steps_per_second": 24.566,
|
| 10 |
+
"epoch": 3.0,
|
| 11 |
+
"total_flos": 4.2393620250444464e+19,
|
| 12 |
+
"train_loss": 0.011377156182601716,
|
| 13 |
+
"train_runtime": 7007.9208,
|
| 14 |
+
"train_samples_per_second": 78.065,
|
| 15 |
+
"train_steps_per_second": 9.758
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"Validation_accuracy": 0.8372543455125931,
|
| 3 |
+
"Validation_f1": 0.8919393816628117,
|
| 4 |
+
"Validation_loss": 0.9663852453231812,
|
| 5 |
+
"Validation_precision": 0.9221433233005075,
|
| 6 |
+
"Validation_recall": 0.8636512917822216,
|
| 7 |
+
"Validation_runtime": 717.1984,
|
| 8 |
+
"Validation_samples_per_second": 196.529,
|
| 9 |
+
"Validation_steps_per_second": 24.566,
|
| 10 |
+
"epoch": 3.0
|
| 11 |
+
}
|
runs/Aug14_16-52-14_Phoenix/events.out.tfevents.1723658464.Phoenix.9472.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5889fc09dde99e7cc02a4a25d83191ea598cf49961edc04a2499fc5fdd5a1df
|
| 3 |
+
size 665
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 4.2393620250444464e+19,
|
| 4 |
+
"train_loss": 0.011377156182601716,
|
| 5 |
+
"train_runtime": 7007.9208,
|
| 6 |
+
"train_samples_per_second": 78.065,
|
| 7 |
+
"train_steps_per_second": 9.758
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1030 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9663852453231812,
|
| 3 |
+
"best_model_checkpoint": "../res/Transformers/VIT-ASVspoof5-Mel_Spectrogram-Synthetic-Voice-Detection\\checkpoint-22795",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 68385,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.021934634788330774,
|
| 13 |
+
"grad_norm": 0.34741124510765076,
|
| 14 |
+
"learning_rate": 4.963442275352782e-05,
|
| 15 |
+
"loss": 0.1315,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04386926957666155,
|
| 20 |
+
"grad_norm": 0.0708213597536087,
|
| 21 |
+
"learning_rate": 4.926884550705564e-05,
|
| 22 |
+
"loss": 0.074,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.06580390436499232,
|
| 27 |
+
"grad_norm": 0.017688876017928123,
|
| 28 |
+
"learning_rate": 4.890326826058346e-05,
|
| 29 |
+
"loss": 0.0627,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0877385391533231,
|
| 34 |
+
"grad_norm": 0.009881118312478065,
|
| 35 |
+
"learning_rate": 4.853769101411129e-05,
|
| 36 |
+
"loss": 0.054,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10967317394165386,
|
| 41 |
+
"grad_norm": 0.017528673633933067,
|
| 42 |
+
"learning_rate": 4.817211376763911e-05,
|
| 43 |
+
"loss": 0.0529,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.13160780872998465,
|
| 48 |
+
"grad_norm": 0.05024711415171623,
|
| 49 |
+
"learning_rate": 4.780653652116692e-05,
|
| 50 |
+
"loss": 0.0436,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.15354244351831542,
|
| 55 |
+
"grad_norm": 0.0056371367536485195,
|
| 56 |
+
"learning_rate": 4.7440959274694746e-05,
|
| 57 |
+
"loss": 0.0478,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.1754770783066462,
|
| 62 |
+
"grad_norm": 0.0035752628464251757,
|
| 63 |
+
"learning_rate": 4.7075382028222566e-05,
|
| 64 |
+
"loss": 0.0304,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.19741171309497696,
|
| 69 |
+
"grad_norm": 0.006441359408199787,
|
| 70 |
+
"learning_rate": 4.6709804781750386e-05,
|
| 71 |
+
"loss": 0.0379,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.21934634788330773,
|
| 76 |
+
"grad_norm": 0.0028403999749571085,
|
| 77 |
+
"learning_rate": 4.6344227535278205e-05,
|
| 78 |
+
"loss": 0.0359,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.24128098267163853,
|
| 83 |
+
"grad_norm": 0.003119397209957242,
|
| 84 |
+
"learning_rate": 4.5978650288806025e-05,
|
| 85 |
+
"loss": 0.0242,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.2632156174599693,
|
| 90 |
+
"grad_norm": 0.006989920046180487,
|
| 91 |
+
"learning_rate": 4.561307304233385e-05,
|
| 92 |
+
"loss": 0.0352,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.28515025224830004,
|
| 97 |
+
"grad_norm": 0.054010841995477676,
|
| 98 |
+
"learning_rate": 4.5247495795861664e-05,
|
| 99 |
+
"loss": 0.029,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.30708488703663084,
|
| 104 |
+
"grad_norm": 0.002598833991214633,
|
| 105 |
+
"learning_rate": 4.488191854938949e-05,
|
| 106 |
+
"loss": 0.0292,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.32901952182496164,
|
| 111 |
+
"grad_norm": 0.0669151097536087,
|
| 112 |
+
"learning_rate": 4.451634130291731e-05,
|
| 113 |
+
"loss": 0.0316,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.3509541566132924,
|
| 118 |
+
"grad_norm": 0.03422193601727486,
|
| 119 |
+
"learning_rate": 4.415076405644513e-05,
|
| 120 |
+
"loss": 0.0219,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3728887914016232,
|
| 125 |
+
"grad_norm": 0.0012628569966182113,
|
| 126 |
+
"learning_rate": 4.378518680997295e-05,
|
| 127 |
+
"loss": 0.0266,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3948234261899539,
|
| 132 |
+
"grad_norm": 0.04995453730225563,
|
| 133 |
+
"learning_rate": 4.341960956350077e-05,
|
| 134 |
+
"loss": 0.0255,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.4167580609782847,
|
| 139 |
+
"grad_norm": 0.0008714247378520668,
|
| 140 |
+
"learning_rate": 4.3054032317028595e-05,
|
| 141 |
+
"loss": 0.0259,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.43869269576661546,
|
| 146 |
+
"grad_norm": 0.002895859070122242,
|
| 147 |
+
"learning_rate": 4.268845507055641e-05,
|
| 148 |
+
"loss": 0.0279,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.46062733055494626,
|
| 153 |
+
"grad_norm": 0.002743455581367016,
|
| 154 |
+
"learning_rate": 4.232287782408423e-05,
|
| 155 |
+
"loss": 0.0176,
|
| 156 |
+
"step": 10500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.48256196534327706,
|
| 160 |
+
"grad_norm": 0.0018967551877722144,
|
| 161 |
+
"learning_rate": 4.1957300577612054e-05,
|
| 162 |
+
"loss": 0.0167,
|
| 163 |
+
"step": 11000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.5044966001316078,
|
| 167 |
+
"grad_norm": 0.08448649197816849,
|
| 168 |
+
"learning_rate": 4.159172333113987e-05,
|
| 169 |
+
"loss": 0.0178,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.5264312349199386,
|
| 174 |
+
"grad_norm": 0.001365090487524867,
|
| 175 |
+
"learning_rate": 4.122614608466769e-05,
|
| 176 |
+
"loss": 0.0169,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5483658697082694,
|
| 181 |
+
"grad_norm": 0.0005711843841709197,
|
| 182 |
+
"learning_rate": 4.086056883819551e-05,
|
| 183 |
+
"loss": 0.018,
|
| 184 |
+
"step": 12500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5703005044966001,
|
| 188 |
+
"grad_norm": 0.0006833109073340893,
|
| 189 |
+
"learning_rate": 4.049499159172334e-05,
|
| 190 |
+
"loss": 0.0173,
|
| 191 |
+
"step": 13000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5922351392849309,
|
| 195 |
+
"grad_norm": 0.0005675885477103293,
|
| 196 |
+
"learning_rate": 4.012941434525115e-05,
|
| 197 |
+
"loss": 0.0222,
|
| 198 |
+
"step": 13500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.6141697740732617,
|
| 202 |
+
"grad_norm": 0.02867540344595909,
|
| 203 |
+
"learning_rate": 3.976383709877897e-05,
|
| 204 |
+
"loss": 0.0221,
|
| 205 |
+
"step": 14000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.6361044088615925,
|
| 209 |
+
"grad_norm": 0.0008733811555430293,
|
| 210 |
+
"learning_rate": 3.93982598523068e-05,
|
| 211 |
+
"loss": 0.0128,
|
| 212 |
+
"step": 14500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6580390436499233,
|
| 216 |
+
"grad_norm": 0.0015775271458551288,
|
| 217 |
+
"learning_rate": 3.903268260583462e-05,
|
| 218 |
+
"loss": 0.0209,
|
| 219 |
+
"step": 15000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.679973678438254,
|
| 223 |
+
"grad_norm": 0.0007889546104706824,
|
| 224 |
+
"learning_rate": 3.866710535936244e-05,
|
| 225 |
+
"loss": 0.0143,
|
| 226 |
+
"step": 15500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.7019083132265848,
|
| 230 |
+
"grad_norm": 0.005065426230430603,
|
| 231 |
+
"learning_rate": 3.8301528112890256e-05,
|
| 232 |
+
"loss": 0.0142,
|
| 233 |
+
"step": 16000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.7238429480149156,
|
| 237 |
+
"grad_norm": 0.0019233387429267168,
|
| 238 |
+
"learning_rate": 3.7935950866418076e-05,
|
| 239 |
+
"loss": 0.0167,
|
| 240 |
+
"step": 16500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7457775828032464,
|
| 244 |
+
"grad_norm": 0.004881173837929964,
|
| 245 |
+
"learning_rate": 3.7570373619945896e-05,
|
| 246 |
+
"loss": 0.0182,
|
| 247 |
+
"step": 17000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7677122175915772,
|
| 251 |
+
"grad_norm": 0.004554756451398134,
|
| 252 |
+
"learning_rate": 3.7204796373473715e-05,
|
| 253 |
+
"loss": 0.0164,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7896468523799078,
|
| 258 |
+
"grad_norm": 0.0010554809123277664,
|
| 259 |
+
"learning_rate": 3.683921912700154e-05,
|
| 260 |
+
"loss": 0.015,
|
| 261 |
+
"step": 18000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.8115814871682386,
|
| 265 |
+
"grad_norm": 0.0011578754056245089,
|
| 266 |
+
"learning_rate": 3.6473641880529354e-05,
|
| 267 |
+
"loss": 0.0137,
|
| 268 |
+
"step": 18500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.8335161219565694,
|
| 272 |
+
"grad_norm": 0.0028496412560343742,
|
| 273 |
+
"learning_rate": 3.6108064634057174e-05,
|
| 274 |
+
"loss": 0.0128,
|
| 275 |
+
"step": 19000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8554507567449002,
|
| 279 |
+
"grad_norm": 0.0005675283609889448,
|
| 280 |
+
"learning_rate": 3.5742487387585e-05,
|
| 281 |
+
"loss": 0.0079,
|
| 282 |
+
"step": 19500
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8773853915332309,
|
| 286 |
+
"grad_norm": 0.000873275741469115,
|
| 287 |
+
"learning_rate": 3.537691014111282e-05,
|
| 288 |
+
"loss": 0.0133,
|
| 289 |
+
"step": 20000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.8993200263215617,
|
| 293 |
+
"grad_norm": 0.002682998776435852,
|
| 294 |
+
"learning_rate": 3.501133289464064e-05,
|
| 295 |
+
"loss": 0.0176,
|
| 296 |
+
"step": 20500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.9212546611098925,
|
| 300 |
+
"grad_norm": 0.00047966442070901394,
|
| 301 |
+
"learning_rate": 3.464575564816846e-05,
|
| 302 |
+
"loss": 0.0034,
|
| 303 |
+
"step": 21000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.9431892958982233,
|
| 307 |
+
"grad_norm": 0.00044119294034317136,
|
| 308 |
+
"learning_rate": 3.428017840169628e-05,
|
| 309 |
+
"loss": 0.0081,
|
| 310 |
+
"step": 21500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9651239306865541,
|
| 314 |
+
"grad_norm": 0.0005569157656282187,
|
| 315 |
+
"learning_rate": 3.39146011552241e-05,
|
| 316 |
+
"loss": 0.0164,
|
| 317 |
+
"step": 22000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9870585654748848,
|
| 321 |
+
"grad_norm": 0.0005497235688380897,
|
| 322 |
+
"learning_rate": 3.354902390875192e-05,
|
| 323 |
+
"loss": 0.0047,
|
| 324 |
+
"step": 22500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.0,
|
| 328 |
+
"eval_accuracy": 0.8372543455125931,
|
| 329 |
+
"eval_f1": 0.8919393816628117,
|
| 330 |
+
"eval_loss": 0.9663852453231812,
|
| 331 |
+
"eval_precision": 0.9221433233005075,
|
| 332 |
+
"eval_recall": 0.8636512917822216,
|
| 333 |
+
"eval_runtime": 782.1941,
|
| 334 |
+
"eval_samples_per_second": 180.198,
|
| 335 |
+
"eval_steps_per_second": 22.525,
|
| 336 |
+
"step": 22795
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.0089932002632156,
|
| 340 |
+
"grad_norm": 0.0004964213585481048,
|
| 341 |
+
"learning_rate": 3.3183446662279744e-05,
|
| 342 |
+
"loss": 0.017,
|
| 343 |
+
"step": 23000
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 1.0309278350515463,
|
| 347 |
+
"grad_norm": 0.0005278219468891621,
|
| 348 |
+
"learning_rate": 3.2817869415807564e-05,
|
| 349 |
+
"loss": 0.0042,
|
| 350 |
+
"step": 23500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.0528624698398772,
|
| 354 |
+
"grad_norm": 0.0016331393271684647,
|
| 355 |
+
"learning_rate": 3.245229216933538e-05,
|
| 356 |
+
"loss": 0.007,
|
| 357 |
+
"step": 24000
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.0747971046282079,
|
| 361 |
+
"grad_norm": 0.00023824520758353174,
|
| 362 |
+
"learning_rate": 3.20867149228632e-05,
|
| 363 |
+
"loss": 0.0027,
|
| 364 |
+
"step": 24500
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 1.0967317394165388,
|
| 368 |
+
"grad_norm": 0.00027841408154927194,
|
| 369 |
+
"learning_rate": 3.172113767639102e-05,
|
| 370 |
+
"loss": 0.0166,
|
| 371 |
+
"step": 25000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.1186663742048695,
|
| 375 |
+
"grad_norm": 0.00036192481638863683,
|
| 376 |
+
"learning_rate": 3.135556042991884e-05,
|
| 377 |
+
"loss": 0.0075,
|
| 378 |
+
"step": 25500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.1406010089932002,
|
| 382 |
+
"grad_norm": 0.0002430744789307937,
|
| 383 |
+
"learning_rate": 3.098998318344666e-05,
|
| 384 |
+
"loss": 0.0119,
|
| 385 |
+
"step": 26000
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 1.162535643781531,
|
| 389 |
+
"grad_norm": 0.0004526956763584167,
|
| 390 |
+
"learning_rate": 3.062440593697448e-05,
|
| 391 |
+
"loss": 0.0137,
|
| 392 |
+
"step": 26500
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.1844702785698618,
|
| 396 |
+
"grad_norm": 0.10140929371118546,
|
| 397 |
+
"learning_rate": 3.0258828690502304e-05,
|
| 398 |
+
"loss": 0.0093,
|
| 399 |
+
"step": 27000
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.2064049133581927,
|
| 403 |
+
"grad_norm": 0.00041032847366295755,
|
| 404 |
+
"learning_rate": 2.9893251444030124e-05,
|
| 405 |
+
"loss": 0.0053,
|
| 406 |
+
"step": 27500
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.2283395481465234,
|
| 410 |
+
"grad_norm": 0.0003452656092122197,
|
| 411 |
+
"learning_rate": 2.9527674197557947e-05,
|
| 412 |
+
"loss": 0.0076,
|
| 413 |
+
"step": 28000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.250274182934854,
|
| 417 |
+
"grad_norm": 0.0002765257377177477,
|
| 418 |
+
"learning_rate": 2.9162096951085767e-05,
|
| 419 |
+
"loss": 0.0049,
|
| 420 |
+
"step": 28500
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.272208817723185,
|
| 424 |
+
"grad_norm": 0.00047475536121055484,
|
| 425 |
+
"learning_rate": 2.8796519704613583e-05,
|
| 426 |
+
"loss": 0.0114,
|
| 427 |
+
"step": 29000
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.2941434525115156,
|
| 431 |
+
"grad_norm": 0.0001869210391305387,
|
| 432 |
+
"learning_rate": 2.8430942458141406e-05,
|
| 433 |
+
"loss": 0.0053,
|
| 434 |
+
"step": 29500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.3160780872998465,
|
| 438 |
+
"grad_norm": 0.0001772296818671748,
|
| 439 |
+
"learning_rate": 2.8065365211669225e-05,
|
| 440 |
+
"loss": 0.0026,
|
| 441 |
+
"step": 30000
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.3380127220881772,
|
| 445 |
+
"grad_norm": 0.0003674278559628874,
|
| 446 |
+
"learning_rate": 2.769978796519705e-05,
|
| 447 |
+
"loss": 0.003,
|
| 448 |
+
"step": 30500
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 1.359947356876508,
|
| 452 |
+
"grad_norm": 0.0005148735363036394,
|
| 453 |
+
"learning_rate": 2.7334210718724868e-05,
|
| 454 |
+
"loss": 0.0089,
|
| 455 |
+
"step": 31000
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.3818819916648388,
|
| 459 |
+
"grad_norm": 0.00596097856760025,
|
| 460 |
+
"learning_rate": 2.6968633472252684e-05,
|
| 461 |
+
"loss": 0.0074,
|
| 462 |
+
"step": 31500
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.4038166264531695,
|
| 466 |
+
"grad_norm": 0.009417912922799587,
|
| 467 |
+
"learning_rate": 2.660305622578051e-05,
|
| 468 |
+
"loss": 0.0065,
|
| 469 |
+
"step": 32000
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 1.4257512612415004,
|
| 473 |
+
"grad_norm": 0.0004612387565430254,
|
| 474 |
+
"learning_rate": 2.6237478979308327e-05,
|
| 475 |
+
"loss": 0.0054,
|
| 476 |
+
"step": 32500
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.447685896029831,
|
| 480 |
+
"grad_norm": 0.000312151329126209,
|
| 481 |
+
"learning_rate": 2.587190173283615e-05,
|
| 482 |
+
"loss": 0.0013,
|
| 483 |
+
"step": 33000
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.4696205308181618,
|
| 487 |
+
"grad_norm": 0.00043960358016192913,
|
| 488 |
+
"learning_rate": 2.550632448636397e-05,
|
| 489 |
+
"loss": 0.0081,
|
| 490 |
+
"step": 33500
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 1.4915551656064927,
|
| 494 |
+
"grad_norm": 0.00014296828885562718,
|
| 495 |
+
"learning_rate": 2.514074723989179e-05,
|
| 496 |
+
"loss": 0.0012,
|
| 497 |
+
"step": 34000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.5134898003948234,
|
| 501 |
+
"grad_norm": 0.00014170966460369527,
|
| 502 |
+
"learning_rate": 2.4775169993419612e-05,
|
| 503 |
+
"loss": 0.0067,
|
| 504 |
+
"step": 34500
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.5354244351831543,
|
| 508 |
+
"grad_norm": 0.00012841432180721313,
|
| 509 |
+
"learning_rate": 2.440959274694743e-05,
|
| 510 |
+
"loss": 0.0021,
|
| 511 |
+
"step": 35000
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.557359069971485,
|
| 515 |
+
"grad_norm": 0.00016817337018437684,
|
| 516 |
+
"learning_rate": 2.404401550047525e-05,
|
| 517 |
+
"loss": 0.0103,
|
| 518 |
+
"step": 35500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.5792937047598157,
|
| 522 |
+
"grad_norm": 0.00028819849831052125,
|
| 523 |
+
"learning_rate": 2.367843825400307e-05,
|
| 524 |
+
"loss": 0.0055,
|
| 525 |
+
"step": 36000
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 1.6012283395481464,
|
| 529 |
+
"grad_norm": 0.00032831408316269517,
|
| 530 |
+
"learning_rate": 2.3312861007530894e-05,
|
| 531 |
+
"loss": 0.0049,
|
| 532 |
+
"step": 36500
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 1.6231629743364773,
|
| 536 |
+
"grad_norm": 0.0003929148952011019,
|
| 537 |
+
"learning_rate": 2.2947283761058713e-05,
|
| 538 |
+
"loss": 0.011,
|
| 539 |
+
"step": 37000
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 1.6450976091248082,
|
| 543 |
+
"grad_norm": 0.015144962817430496,
|
| 544 |
+
"learning_rate": 2.2581706514586533e-05,
|
| 545 |
+
"loss": 0.0009,
|
| 546 |
+
"step": 37500
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.6670322439131389,
|
| 550 |
+
"grad_norm": 0.0004016929306089878,
|
| 551 |
+
"learning_rate": 2.2216129268114352e-05,
|
| 552 |
+
"loss": 0.0039,
|
| 553 |
+
"step": 38000
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 1.6889668787014696,
|
| 557 |
+
"grad_norm": 0.0003622666699811816,
|
| 558 |
+
"learning_rate": 2.1850552021642172e-05,
|
| 559 |
+
"loss": 0.011,
|
| 560 |
+
"step": 38500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.7109015134898002,
|
| 564 |
+
"grad_norm": 0.00015891093062236905,
|
| 565 |
+
"learning_rate": 2.1484974775169995e-05,
|
| 566 |
+
"loss": 0.0039,
|
| 567 |
+
"step": 39000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.7328361482781311,
|
| 571 |
+
"grad_norm": 2.2137434482574463,
|
| 572 |
+
"learning_rate": 2.1119397528697815e-05,
|
| 573 |
+
"loss": 0.0064,
|
| 574 |
+
"step": 39500
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.754770783066462,
|
| 578 |
+
"grad_norm": 0.01450470369309187,
|
| 579 |
+
"learning_rate": 2.0753820282225638e-05,
|
| 580 |
+
"loss": 0.0019,
|
| 581 |
+
"step": 40000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.7767054178547927,
|
| 585 |
+
"grad_norm": 0.0001268786145374179,
|
| 586 |
+
"learning_rate": 2.0388243035753457e-05,
|
| 587 |
+
"loss": 0.006,
|
| 588 |
+
"step": 40500
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 1.7986400526431234,
|
| 592 |
+
"grad_norm": 0.00022799715225119144,
|
| 593 |
+
"learning_rate": 2.0022665789281277e-05,
|
| 594 |
+
"loss": 0.0,
|
| 595 |
+
"step": 41000
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 1.8205746874314541,
|
| 599 |
+
"grad_norm": 0.0007139018853195012,
|
| 600 |
+
"learning_rate": 1.9657088542809096e-05,
|
| 601 |
+
"loss": 0.0105,
|
| 602 |
+
"step": 41500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.842509322219785,
|
| 606 |
+
"grad_norm": 0.000262285815551877,
|
| 607 |
+
"learning_rate": 1.9291511296336916e-05,
|
| 608 |
+
"loss": 0.0002,
|
| 609 |
+
"step": 42000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.864443957008116,
|
| 613 |
+
"grad_norm": 9.916254930431023e-05,
|
| 614 |
+
"learning_rate": 1.892593404986474e-05,
|
| 615 |
+
"loss": 0.0034,
|
| 616 |
+
"step": 42500
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.8863785917964466,
|
| 620 |
+
"grad_norm": 0.00010086688416777179,
|
| 621 |
+
"learning_rate": 1.856035680339256e-05,
|
| 622 |
+
"loss": 0.0103,
|
| 623 |
+
"step": 43000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.9083132265847773,
|
| 627 |
+
"grad_norm": 0.0003273077018093318,
|
| 628 |
+
"learning_rate": 1.8194779556920378e-05,
|
| 629 |
+
"loss": 0.0043,
|
| 630 |
+
"step": 43500
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.930247861373108,
|
| 634 |
+
"grad_norm": 0.0004520936054177582,
|
| 635 |
+
"learning_rate": 1.7829202310448198e-05,
|
| 636 |
+
"loss": 0.0012,
|
| 637 |
+
"step": 44000
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.952182496161439,
|
| 641 |
+
"grad_norm": 0.00017915925127454102,
|
| 642 |
+
"learning_rate": 1.7463625063976017e-05,
|
| 643 |
+
"loss": 0.0027,
|
| 644 |
+
"step": 44500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.9741171309497698,
|
| 648 |
+
"grad_norm": 8.093972428468987e-05,
|
| 649 |
+
"learning_rate": 1.709804781750384e-05,
|
| 650 |
+
"loss": 0.0,
|
| 651 |
+
"step": 45000
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.9960517657381005,
|
| 655 |
+
"grad_norm": 0.00023188079649116844,
|
| 656 |
+
"learning_rate": 1.673247057103166e-05,
|
| 657 |
+
"loss": 0.0064,
|
| 658 |
+
"step": 45500
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 2.0,
|
| 662 |
+
"eval_accuracy": 0.7830365377793543,
|
| 663 |
+
"eval_f1": 0.8420981974585766,
|
| 664 |
+
"eval_loss": 1.6013481616973877,
|
| 665 |
+
"eval_precision": 0.9701385997263696,
|
| 666 |
+
"eval_recall": 0.7439151218800175,
|
| 667 |
+
"eval_runtime": 713.4707,
|
| 668 |
+
"eval_samples_per_second": 197.555,
|
| 669 |
+
"eval_steps_per_second": 24.695,
|
| 670 |
+
"step": 45590
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 2.017986400526431,
|
| 674 |
+
"grad_norm": 0.00013298647536430508,
|
| 675 |
+
"learning_rate": 1.636689332455948e-05,
|
| 676 |
+
"loss": 0.0,
|
| 677 |
+
"step": 46000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.039921035314762,
|
| 681 |
+
"grad_norm": 0.002622125670313835,
|
| 682 |
+
"learning_rate": 1.60013160780873e-05,
|
| 683 |
+
"loss": 0.0049,
|
| 684 |
+
"step": 46500
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 2.0618556701030926,
|
| 688 |
+
"grad_norm": 0.0016739132115617394,
|
| 689 |
+
"learning_rate": 1.5635738831615122e-05,
|
| 690 |
+
"loss": 0.0033,
|
| 691 |
+
"step": 47000
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 2.0837903048914237,
|
| 695 |
+
"grad_norm": 7.326369086513296e-05,
|
| 696 |
+
"learning_rate": 1.5270161585142942e-05,
|
| 697 |
+
"loss": 0.0,
|
| 698 |
+
"step": 47500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.1057249396797544,
|
| 702 |
+
"grad_norm": 7.18442679499276e-05,
|
| 703 |
+
"learning_rate": 1.4904584338670763e-05,
|
| 704 |
+
"loss": 0.0016,
|
| 705 |
+
"step": 48000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.127659574468085,
|
| 709 |
+
"grad_norm": 0.00014385154645424336,
|
| 710 |
+
"learning_rate": 1.4539007092198581e-05,
|
| 711 |
+
"loss": 0.0,
|
| 712 |
+
"step": 48500
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 2.1495942092564158,
|
| 716 |
+
"grad_norm": 0.00024707347620278597,
|
| 717 |
+
"learning_rate": 1.4173429845726402e-05,
|
| 718 |
+
"loss": 0.0,
|
| 719 |
+
"step": 49000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 2.1715288440447464,
|
| 723 |
+
"grad_norm": 0.00010678445687517524,
|
| 724 |
+
"learning_rate": 1.3807852599254223e-05,
|
| 725 |
+
"loss": 0.0,
|
| 726 |
+
"step": 49500
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 2.1934634788330776,
|
| 730 |
+
"grad_norm": 5.570024222834036e-05,
|
| 731 |
+
"learning_rate": 1.3442275352782043e-05,
|
| 732 |
+
"loss": 0.0,
|
| 733 |
+
"step": 50000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 2.2153981136214083,
|
| 737 |
+
"grad_norm": 0.03718312457203865,
|
| 738 |
+
"learning_rate": 1.3076698106309864e-05,
|
| 739 |
+
"loss": 0.0106,
|
| 740 |
+
"step": 50500
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 2.237332748409739,
|
| 744 |
+
"grad_norm": 5.917974704061635e-05,
|
| 745 |
+
"learning_rate": 1.2711120859837682e-05,
|
| 746 |
+
"loss": 0.0001,
|
| 747 |
+
"step": 51000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 2.2592673831980696,
|
| 751 |
+
"grad_norm": 8.80313673405908e-05,
|
| 752 |
+
"learning_rate": 1.2345543613365505e-05,
|
| 753 |
+
"loss": 0.0,
|
| 754 |
+
"step": 51500
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 2.2812020179864003,
|
| 758 |
+
"grad_norm": 8.454511407762766e-05,
|
| 759 |
+
"learning_rate": 1.1979966366893325e-05,
|
| 760 |
+
"loss": 0.0,
|
| 761 |
+
"step": 52000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 2.3031366527747315,
|
| 765 |
+
"grad_norm": 4.547155913314782e-05,
|
| 766 |
+
"learning_rate": 1.1614389120421146e-05,
|
| 767 |
+
"loss": 0.0013,
|
| 768 |
+
"step": 52500
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 2.325071287563062,
|
| 772 |
+
"grad_norm": 9.757141378941014e-05,
|
| 773 |
+
"learning_rate": 1.1248811873948966e-05,
|
| 774 |
+
"loss": 0.0,
|
| 775 |
+
"step": 53000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 2.347005922351393,
|
| 779 |
+
"grad_norm": 6.546088116010651e-05,
|
| 780 |
+
"learning_rate": 1.0883234627476787e-05,
|
| 781 |
+
"loss": 0.0026,
|
| 782 |
+
"step": 53500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.3689405571397235,
|
| 786 |
+
"grad_norm": 3.695198756759055e-05,
|
| 787 |
+
"learning_rate": 1.0517657381004607e-05,
|
| 788 |
+
"loss": 0.0012,
|
| 789 |
+
"step": 54000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.390875191928054,
|
| 793 |
+
"grad_norm": 3.3266915124841034e-05,
|
| 794 |
+
"learning_rate": 1.0152080134532426e-05,
|
| 795 |
+
"loss": 0.0,
|
| 796 |
+
"step": 54500
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.4128098267163853,
|
| 800 |
+
"grad_norm": 3.605923484428786e-05,
|
| 801 |
+
"learning_rate": 9.786502888060247e-06,
|
| 802 |
+
"loss": 0.0048,
|
| 803 |
+
"step": 55000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 2.434744461504716,
|
| 807 |
+
"grad_norm": 8.535337110515684e-05,
|
| 808 |
+
"learning_rate": 9.420925641588069e-06,
|
| 809 |
+
"loss": 0.0028,
|
| 810 |
+
"step": 55500
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 2.4566790962930467,
|
| 814 |
+
"grad_norm": 7.798920705681667e-05,
|
| 815 |
+
"learning_rate": 9.055348395115888e-06,
|
| 816 |
+
"loss": 0.0011,
|
| 817 |
+
"step": 56000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 2.4786137310813774,
|
| 821 |
+
"grad_norm": 3.1817657145438716e-05,
|
| 822 |
+
"learning_rate": 8.68977114864371e-06,
|
| 823 |
+
"loss": 0.0016,
|
| 824 |
+
"step": 56500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.500548365869708,
|
| 828 |
+
"grad_norm": 3.148000905639492e-05,
|
| 829 |
+
"learning_rate": 8.32419390217153e-06,
|
| 830 |
+
"loss": 0.0,
|
| 831 |
+
"step": 57000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.522483000658039,
|
| 835 |
+
"grad_norm": 0.21145348250865936,
|
| 836 |
+
"learning_rate": 7.95861665569935e-06,
|
| 837 |
+
"loss": 0.0035,
|
| 838 |
+
"step": 57500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 2.54441763544637,
|
| 842 |
+
"grad_norm": 4.887134491582401e-05,
|
| 843 |
+
"learning_rate": 7.593039409227171e-06,
|
| 844 |
+
"loss": 0.0006,
|
| 845 |
+
"step": 58000
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 2.5663522702347006,
|
| 849 |
+
"grad_norm": 2.885399771912489e-05,
|
| 850 |
+
"learning_rate": 7.2274621627549905e-06,
|
| 851 |
+
"loss": 0.0025,
|
| 852 |
+
"step": 58500
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 2.5882869050230313,
|
| 856 |
+
"grad_norm": 2.8404225304257125e-05,
|
| 857 |
+
"learning_rate": 6.861884916282811e-06,
|
| 858 |
+
"loss": 0.001,
|
| 859 |
+
"step": 59000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 2.610221539811362,
|
| 863 |
+
"grad_norm": 6.570501864189282e-05,
|
| 864 |
+
"learning_rate": 6.496307669810631e-06,
|
| 865 |
+
"loss": 0.0,
|
| 866 |
+
"step": 59500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.632156174599693,
|
| 870 |
+
"grad_norm": 4.453480869415216e-05,
|
| 871 |
+
"learning_rate": 6.130730423338452e-06,
|
| 872 |
+
"loss": 0.0,
|
| 873 |
+
"step": 60000
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.6540908093880238,
|
| 877 |
+
"grad_norm": 4.257899854565039e-05,
|
| 878 |
+
"learning_rate": 5.7651531768662715e-06,
|
| 879 |
+
"loss": 0.0,
|
| 880 |
+
"step": 60500
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 2.6760254441763545,
|
| 884 |
+
"grad_norm": 4.08503801736515e-05,
|
| 885 |
+
"learning_rate": 5.399575930394093e-06,
|
| 886 |
+
"loss": 0.0012,
|
| 887 |
+
"step": 61000
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 2.697960078964685,
|
| 891 |
+
"grad_norm": 2.521094575058669e-05,
|
| 892 |
+
"learning_rate": 5.033998683921913e-06,
|
| 893 |
+
"loss": 0.0001,
|
| 894 |
+
"step": 61500
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 2.719894713753016,
|
| 898 |
+
"grad_norm": 5.6092027080012485e-05,
|
| 899 |
+
"learning_rate": 4.668421437449734e-06,
|
| 900 |
+
"loss": 0.0011,
|
| 901 |
+
"step": 62000
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 2.741829348541347,
|
| 905 |
+
"grad_norm": 3.7315894587663934e-05,
|
| 906 |
+
"learning_rate": 4.302844190977553e-06,
|
| 907 |
+
"loss": 0.0,
|
| 908 |
+
"step": 62500
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 2.7637639833296777,
|
| 912 |
+
"grad_norm": 3.713283149409108e-05,
|
| 913 |
+
"learning_rate": 3.9372669445053745e-06,
|
| 914 |
+
"loss": 0.0,
|
| 915 |
+
"step": 63000
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 2.7856986181180083,
|
| 919 |
+
"grad_norm": 3.7352405342971906e-05,
|
| 920 |
+
"learning_rate": 3.5716896980331945e-06,
|
| 921 |
+
"loss": 0.0028,
|
| 922 |
+
"step": 63500
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 2.807633252906339,
|
| 926 |
+
"grad_norm": 2.116698851750698e-05,
|
| 927 |
+
"learning_rate": 3.206112451561015e-06,
|
| 928 |
+
"loss": 0.0,
|
| 929 |
+
"step": 64000
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 2.8295678876946697,
|
| 933 |
+
"grad_norm": 2.0813607989111915e-05,
|
| 934 |
+
"learning_rate": 2.8405352050888354e-06,
|
| 935 |
+
"loss": 0.0,
|
| 936 |
+
"step": 64500
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 2.851502522483001,
|
| 940 |
+
"grad_norm": 2.0640662114601582e-05,
|
| 941 |
+
"learning_rate": 2.474957958616656e-06,
|
| 942 |
+
"loss": 0.0018,
|
| 943 |
+
"step": 65000
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 2.8734371572713315,
|
| 947 |
+
"grad_norm": 4.867030293098651e-05,
|
| 948 |
+
"learning_rate": 2.109380712144476e-06,
|
| 949 |
+
"loss": 0.0,
|
| 950 |
+
"step": 65500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 2.895371792059662,
|
| 954 |
+
"grad_norm": 3.334460052428767e-05,
|
| 955 |
+
"learning_rate": 1.7438034656722967e-06,
|
| 956 |
+
"loss": 0.0,
|
| 957 |
+
"step": 66000
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 2.917306426847993,
|
| 961 |
+
"grad_norm": 4.7563084081048146e-05,
|
| 962 |
+
"learning_rate": 1.3782262192001172e-06,
|
| 963 |
+
"loss": 0.0,
|
| 964 |
+
"step": 66500
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 2.9392410616363236,
|
| 968 |
+
"grad_norm": 1.908742342493497e-05,
|
| 969 |
+
"learning_rate": 1.0126489727279376e-06,
|
| 970 |
+
"loss": 0.0,
|
| 971 |
+
"step": 67000
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 2.9611756964246547,
|
| 975 |
+
"grad_norm": 3.217206904082559e-05,
|
| 976 |
+
"learning_rate": 6.470717262557578e-07,
|
| 977 |
+
"loss": 0.0,
|
| 978 |
+
"step": 67500
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 2.9831103312129854,
|
| 982 |
+
"grad_norm": 4.605944195645861e-05,
|
| 983 |
+
"learning_rate": 2.814944797835783e-07,
|
| 984 |
+
"loss": 0.0,
|
| 985 |
+
"step": 68000
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.0,
|
| 989 |
+
"eval_accuracy": 0.7633416105001773,
|
| 990 |
+
"eval_f1": 0.8263822744093812,
|
| 991 |
+
"eval_loss": 2.0727663040161133,
|
| 992 |
+
"eval_precision": 0.9621029413546957,
|
| 993 |
+
"eval_recall": 0.7242190921033426,
|
| 994 |
+
"eval_runtime": 717.116,
|
| 995 |
+
"eval_samples_per_second": 196.551,
|
| 996 |
+
"eval_steps_per_second": 24.569,
|
| 997 |
+
"step": 68385
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 3.0,
|
| 1001 |
+
"step": 68385,
|
| 1002 |
+
"total_flos": 4.2393620250444464e+19,
|
| 1003 |
+
"train_loss": 0.011377156182601716,
|
| 1004 |
+
"train_runtime": 7007.9208,
|
| 1005 |
+
"train_samples_per_second": 78.065,
|
| 1006 |
+
"train_steps_per_second": 9.758
|
| 1007 |
+
}
|
| 1008 |
+
],
|
| 1009 |
+
"logging_steps": 500,
|
| 1010 |
+
"max_steps": 68385,
|
| 1011 |
+
"num_input_tokens_seen": 0,
|
| 1012 |
+
"num_train_epochs": 3,
|
| 1013 |
+
"save_steps": 500,
|
| 1014 |
+
"stateful_callbacks": {
|
| 1015 |
+
"TrainerControl": {
|
| 1016 |
+
"args": {
|
| 1017 |
+
"should_epoch_stop": false,
|
| 1018 |
+
"should_evaluate": false,
|
| 1019 |
+
"should_log": false,
|
| 1020 |
+
"should_save": true,
|
| 1021 |
+
"should_training_stop": true
|
| 1022 |
+
},
|
| 1023 |
+
"attributes": {}
|
| 1024 |
+
}
|
| 1025 |
+
},
|
| 1026 |
+
"total_flos": 4.2393620250444464e+19,
|
| 1027 |
+
"train_batch_size": 8,
|
| 1028 |
+
"trial_name": null,
|
| 1029 |
+
"trial_params": null
|
| 1030 |
+
}
|