Model save
Browse files- all_results.json +9 -0
- generation_config.json +12 -0
- train_results.json +9 -0
- trainer_state.json +2172 -0
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.052201553022480446,
|
| 5 |
+
"train_runtime": 2376.2971,
|
| 6 |
+
"train_samples": 10510,
|
| 7 |
+
"train_samples_per_second": 4.423,
|
| 8 |
+
"train_steps_per_second": 0.553
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128009
|
| 7 |
+
],
|
| 8 |
+
"max_length": 4096,
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.53.3"
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.052201553022480446,
|
| 5 |
+
"train_runtime": 2376.2971,
|
| 6 |
+
"train_samples": 10510,
|
| 7 |
+
"train_samples_per_second": 4.423,
|
| 8 |
+
"train_steps_per_second": 0.553
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2172 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1314,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00076103500761035,
|
| 14 |
+
"grad_norm": 303.2175309709565,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"logits/chosen": 0.1357421875,
|
| 17 |
+
"logits/rejected": -0.1416015625,
|
| 18 |
+
"logps/chosen": -536.0,
|
| 19 |
+
"logps/rejected": -372.0,
|
| 20 |
+
"loss": 0.6924,
|
| 21 |
+
"nll_loss": 1.3515625,
|
| 22 |
+
"rewards/accuracies": 0.0,
|
| 23 |
+
"rewards/chosen": 0.0,
|
| 24 |
+
"rewards/margins": 0.0,
|
| 25 |
+
"rewards/rejected": 0.0,
|
| 26 |
+
"step": 1
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.0076103500761035,
|
| 30 |
+
"grad_norm": 368.601857167257,
|
| 31 |
+
"learning_rate": 3.4090909090909086e-08,
|
| 32 |
+
"logits/chosen": 0.09814453125,
|
| 33 |
+
"logits/rejected": -0.0525716133415699,
|
| 34 |
+
"logps/chosen": -575.3333129882812,
|
| 35 |
+
"logps/rejected": -460.0,
|
| 36 |
+
"loss": 0.7304,
|
| 37 |
+
"nll_loss": 1.4830728769302368,
|
| 38 |
+
"rewards/accuracies": 0.2222222238779068,
|
| 39 |
+
"rewards/chosen": 0.03680419921875,
|
| 40 |
+
"rewards/margins": -0.01531304232776165,
|
| 41 |
+
"rewards/rejected": 0.0521647147834301,
|
| 42 |
+
"step": 10
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.015220700152207,
|
| 46 |
+
"grad_norm": 290.84566236586966,
|
| 47 |
+
"learning_rate": 7.196969696969697e-08,
|
| 48 |
+
"logits/chosen": 0.14541015028953552,
|
| 49 |
+
"logits/rejected": -0.04328613355755806,
|
| 50 |
+
"logps/chosen": -559.5999755859375,
|
| 51 |
+
"logps/rejected": -378.6000061035156,
|
| 52 |
+
"loss": 0.6982,
|
| 53 |
+
"nll_loss": 1.278906226158142,
|
| 54 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 55 |
+
"rewards/chosen": -0.01192626915872097,
|
| 56 |
+
"rewards/margins": -6.10351571594947e-06,
|
| 57 |
+
"rewards/rejected": -0.011883544735610485,
|
| 58 |
+
"step": 20
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0228310502283105,
|
| 62 |
+
"grad_norm": 264.33469587660574,
|
| 63 |
+
"learning_rate": 1.0984848484848484e-07,
|
| 64 |
+
"logits/chosen": 0.11530456691980362,
|
| 65 |
+
"logits/rejected": -0.01654052734375,
|
| 66 |
+
"logps/chosen": -497.5,
|
| 67 |
+
"logps/rejected": -370.29998779296875,
|
| 68 |
+
"loss": 0.6754,
|
| 69 |
+
"nll_loss": 1.200781226158142,
|
| 70 |
+
"rewards/accuracies": 0.3125,
|
| 71 |
+
"rewards/chosen": 0.02252044714987278,
|
| 72 |
+
"rewards/margins": 0.03815612941980362,
|
| 73 |
+
"rewards/rejected": -0.01563110388815403,
|
| 74 |
+
"step": 30
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.030441400304414,
|
| 78 |
+
"grad_norm": 302.0148081017884,
|
| 79 |
+
"learning_rate": 1.4772727272727272e-07,
|
| 80 |
+
"logits/chosen": 0.08310546725988388,
|
| 81 |
+
"logits/rejected": -0.06760253757238388,
|
| 82 |
+
"logps/chosen": -622.7999877929688,
|
| 83 |
+
"logps/rejected": -496.5,
|
| 84 |
+
"loss": 0.632,
|
| 85 |
+
"nll_loss": 1.357812523841858,
|
| 86 |
+
"rewards/accuracies": 0.574999988079071,
|
| 87 |
+
"rewards/chosen": -0.008761596865952015,
|
| 88 |
+
"rewards/margins": 0.15755920112133026,
|
| 89 |
+
"rewards/rejected": -0.16655273735523224,
|
| 90 |
+
"step": 40
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.0380517503805175,
|
| 94 |
+
"grad_norm": 207.75823997721494,
|
| 95 |
+
"learning_rate": 1.856060606060606e-07,
|
| 96 |
+
"logits/chosen": 0.09158935397863388,
|
| 97 |
+
"logits/rejected": -0.10623779147863388,
|
| 98 |
+
"logps/chosen": -635.4000244140625,
|
| 99 |
+
"logps/rejected": -458.0,
|
| 100 |
+
"loss": 0.5591,
|
| 101 |
+
"nll_loss": 1.40234375,
|
| 102 |
+
"rewards/accuracies": 0.875,
|
| 103 |
+
"rewards/chosen": -0.04698486253619194,
|
| 104 |
+
"rewards/margins": 0.39501953125,
|
| 105 |
+
"rewards/rejected": -0.4413085877895355,
|
| 106 |
+
"step": 50
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.045662100456621,
|
| 110 |
+
"grad_norm": 232.18657547260776,
|
| 111 |
+
"learning_rate": 2.2348484848484846e-07,
|
| 112 |
+
"logits/chosen": 0.03912963718175888,
|
| 113 |
+
"logits/rejected": -0.18083496391773224,
|
| 114 |
+
"logps/chosen": -584.5999755859375,
|
| 115 |
+
"logps/rejected": -446.8999938964844,
|
| 116 |
+
"loss": 0.4174,
|
| 117 |
+
"nll_loss": 1.3078124523162842,
|
| 118 |
+
"rewards/accuracies": 0.9375,
|
| 119 |
+
"rewards/chosen": -0.10451660305261612,
|
| 120 |
+
"rewards/margins": 0.818359375,
|
| 121 |
+
"rewards/rejected": -0.923632800579071,
|
| 122 |
+
"step": 60
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0532724505327245,
|
| 126 |
+
"grad_norm": 124.85469490515551,
|
| 127 |
+
"learning_rate": 2.6136363636363634e-07,
|
| 128 |
+
"logits/chosen": -0.03662109375,
|
| 129 |
+
"logits/rejected": -0.15817871689796448,
|
| 130 |
+
"logps/chosen": -658.0,
|
| 131 |
+
"logps/rejected": -474.0,
|
| 132 |
+
"loss": 0.2946,
|
| 133 |
+
"nll_loss": 1.4148437976837158,
|
| 134 |
+
"rewards/accuracies": 0.949999988079071,
|
| 135 |
+
"rewards/chosen": -0.07039184868335724,
|
| 136 |
+
"rewards/margins": 1.586328148841858,
|
| 137 |
+
"rewards/rejected": -1.65625,
|
| 138 |
+
"step": 70
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.060882800608828,
|
| 142 |
+
"grad_norm": 86.0204209830328,
|
| 143 |
+
"learning_rate": 2.9924242424242425e-07,
|
| 144 |
+
"logits/chosen": 0.02304687537252903,
|
| 145 |
+
"logits/rejected": -0.06480713188648224,
|
| 146 |
+
"logps/chosen": -507.0,
|
| 147 |
+
"logps/rejected": -455.0,
|
| 148 |
+
"loss": 0.218,
|
| 149 |
+
"nll_loss": 1.357031226158142,
|
| 150 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 151 |
+
"rewards/chosen": -0.25971680879592896,
|
| 152 |
+
"rewards/margins": 2.0703125,
|
| 153 |
+
"rewards/rejected": -2.3257813453674316,
|
| 154 |
+
"step": 80
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.0684931506849315,
|
| 158 |
+
"grad_norm": 76.83932328557562,
|
| 159 |
+
"learning_rate": 3.371212121212121e-07,
|
| 160 |
+
"logits/chosen": -0.0015747069846838713,
|
| 161 |
+
"logits/rejected": -0.03731689602136612,
|
| 162 |
+
"logps/chosen": -519.2000122070312,
|
| 163 |
+
"logps/rejected": -402.20001220703125,
|
| 164 |
+
"loss": 0.2342,
|
| 165 |
+
"nll_loss": 1.3230469226837158,
|
| 166 |
+
"rewards/accuracies": 0.949999988079071,
|
| 167 |
+
"rewards/chosen": -0.35136717557907104,
|
| 168 |
+
"rewards/margins": 2.703125,
|
| 169 |
+
"rewards/rejected": -3.0562500953674316,
|
| 170 |
+
"step": 90
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.076103500761035,
|
| 174 |
+
"grad_norm": 49.92360549661787,
|
| 175 |
+
"learning_rate": 3.75e-07,
|
| 176 |
+
"logits/chosen": 0.10744629055261612,
|
| 177 |
+
"logits/rejected": -0.13419190049171448,
|
| 178 |
+
"logps/chosen": -489.8999938964844,
|
| 179 |
+
"logps/rejected": -429.5,
|
| 180 |
+
"loss": 0.144,
|
| 181 |
+
"nll_loss": 1.256250023841858,
|
| 182 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 183 |
+
"rewards/chosen": -0.06059570237994194,
|
| 184 |
+
"rewards/margins": 3.467968702316284,
|
| 185 |
+
"rewards/rejected": -3.5296874046325684,
|
| 186 |
+
"step": 100
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.0837138508371385,
|
| 190 |
+
"grad_norm": 58.02766606703985,
|
| 191 |
+
"learning_rate": 4.1287878787878786e-07,
|
| 192 |
+
"logits/chosen": -0.008868408389389515,
|
| 193 |
+
"logits/rejected": -0.18212890625,
|
| 194 |
+
"logps/chosen": -566.0,
|
| 195 |
+
"logps/rejected": -507.0,
|
| 196 |
+
"loss": 0.0769,
|
| 197 |
+
"nll_loss": 1.299218773841858,
|
| 198 |
+
"rewards/accuracies": 1.0,
|
| 199 |
+
"rewards/chosen": 0.1328125,
|
| 200 |
+
"rewards/margins": 4.6328125,
|
| 201 |
+
"rewards/rejected": -4.50390625,
|
| 202 |
+
"step": 110
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.091324200913242,
|
| 206 |
+
"grad_norm": 9.486986111317089,
|
| 207 |
+
"learning_rate": 4.507575757575757e-07,
|
| 208 |
+
"logits/chosen": 0.01968994177877903,
|
| 209 |
+
"logits/rejected": -0.31318360567092896,
|
| 210 |
+
"logps/chosen": -516.0,
|
| 211 |
+
"logps/rejected": -502.79998779296875,
|
| 212 |
+
"loss": 0.0761,
|
| 213 |
+
"nll_loss": 1.310546875,
|
| 214 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 215 |
+
"rewards/chosen": 0.4986572265625,
|
| 216 |
+
"rewards/margins": 5.587500095367432,
|
| 217 |
+
"rewards/rejected": -5.089062690734863,
|
| 218 |
+
"step": 120
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.0989345509893455,
|
| 222 |
+
"grad_norm": 27.721172724343948,
|
| 223 |
+
"learning_rate": 4.886363636363636e-07,
|
| 224 |
+
"logits/chosen": -0.04987793043255806,
|
| 225 |
+
"logits/rejected": -0.140380859375,
|
| 226 |
+
"logps/chosen": -655.0,
|
| 227 |
+
"logps/rejected": -492.79998779296875,
|
| 228 |
+
"loss": 0.3649,
|
| 229 |
+
"nll_loss": 1.3679687976837158,
|
| 230 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 231 |
+
"rewards/chosen": -0.04736328125,
|
| 232 |
+
"rewards/margins": 4.926562309265137,
|
| 233 |
+
"rewards/rejected": -4.984375,
|
| 234 |
+
"step": 130
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.106544901065449,
|
| 238 |
+
"grad_norm": 191.85974880626392,
|
| 239 |
+
"learning_rate": 4.970389170896784e-07,
|
| 240 |
+
"logits/chosen": 0.09293212741613388,
|
| 241 |
+
"logits/rejected": -0.12996216118335724,
|
| 242 |
+
"logps/chosen": -555.7999877929688,
|
| 243 |
+
"logps/rejected": -464.20001220703125,
|
| 244 |
+
"loss": 0.0816,
|
| 245 |
+
"nll_loss": 1.43359375,
|
| 246 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 247 |
+
"rewards/chosen": 0.5732421875,
|
| 248 |
+
"rewards/margins": 6.125,
|
| 249 |
+
"rewards/rejected": -5.543749809265137,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.1141552511415525,
|
| 254 |
+
"grad_norm": 11.2332059562663,
|
| 255 |
+
"learning_rate": 4.928087986463621e-07,
|
| 256 |
+
"logits/chosen": 0.01511230506002903,
|
| 257 |
+
"logits/rejected": -0.29150390625,
|
| 258 |
+
"logps/chosen": -667.5999755859375,
|
| 259 |
+
"logps/rejected": -607.5999755859375,
|
| 260 |
+
"loss": 0.0364,
|
| 261 |
+
"nll_loss": 1.265625,
|
| 262 |
+
"rewards/accuracies": 0.987500011920929,
|
| 263 |
+
"rewards/chosen": 0.02332153357565403,
|
| 264 |
+
"rewards/margins": 8.153124809265137,
|
| 265 |
+
"rewards/rejected": -8.128125190734863,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.121765601217656,
|
| 270 |
+
"grad_norm": 37.66600316105928,
|
| 271 |
+
"learning_rate": 4.885786802030457e-07,
|
| 272 |
+
"logits/chosen": 0.00010986327833961695,
|
| 273 |
+
"logits/rejected": -0.30913084745407104,
|
| 274 |
+
"logps/chosen": -589.0,
|
| 275 |
+
"logps/rejected": -644.5999755859375,
|
| 276 |
+
"loss": 0.0212,
|
| 277 |
+
"nll_loss": 1.4304687976837158,
|
| 278 |
+
"rewards/accuracies": 1.0,
|
| 279 |
+
"rewards/chosen": -0.008593750186264515,
|
| 280 |
+
"rewards/margins": 8.940625190734863,
|
| 281 |
+
"rewards/rejected": -8.959375381469727,
|
| 282 |
+
"step": 160
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.1293759512937595,
|
| 286 |
+
"grad_norm": 6.969455968571805,
|
| 287 |
+
"learning_rate": 4.843485617597293e-07,
|
| 288 |
+
"logits/chosen": 0.06293945014476776,
|
| 289 |
+
"logits/rejected": -0.14820556342601776,
|
| 290 |
+
"logps/chosen": -467.0,
|
| 291 |
+
"logps/rejected": -399.0,
|
| 292 |
+
"loss": 0.0499,
|
| 293 |
+
"nll_loss": 1.4128906726837158,
|
| 294 |
+
"rewards/accuracies": 0.987500011920929,
|
| 295 |
+
"rewards/chosen": -0.35380858182907104,
|
| 296 |
+
"rewards/margins": 7.698437690734863,
|
| 297 |
+
"rewards/rejected": -8.053125381469727,
|
| 298 |
+
"step": 170
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.136986301369863,
|
| 302 |
+
"grad_norm": 60.8398584611724,
|
| 303 |
+
"learning_rate": 4.801184433164128e-07,
|
| 304 |
+
"logits/chosen": 0.02814941480755806,
|
| 305 |
+
"logits/rejected": -0.2801269590854645,
|
| 306 |
+
"logps/chosen": -605.4000244140625,
|
| 307 |
+
"logps/rejected": -575.5999755859375,
|
| 308 |
+
"loss": 0.0526,
|
| 309 |
+
"nll_loss": 1.409765601158142,
|
| 310 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 311 |
+
"rewards/chosen": -0.47587889432907104,
|
| 312 |
+
"rewards/margins": 10.84375,
|
| 313 |
+
"rewards/rejected": -11.324999809265137,
|
| 314 |
+
"step": 180
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.1445966514459665,
|
| 318 |
+
"grad_norm": 3.507659046944847,
|
| 319 |
+
"learning_rate": 4.7588832487309643e-07,
|
| 320 |
+
"logits/chosen": 0.0074462890625,
|
| 321 |
+
"logits/rejected": -0.27363282442092896,
|
| 322 |
+
"logps/chosen": -686.4000244140625,
|
| 323 |
+
"logps/rejected": -565.7999877929688,
|
| 324 |
+
"loss": 0.0314,
|
| 325 |
+
"nll_loss": 1.287500023841858,
|
| 326 |
+
"rewards/accuracies": 0.987500011920929,
|
| 327 |
+
"rewards/chosen": -1.643164038658142,
|
| 328 |
+
"rewards/margins": 10.109375,
|
| 329 |
+
"rewards/rejected": -11.75,
|
| 330 |
+
"step": 190
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.15220700152207,
|
| 334 |
+
"grad_norm": 193.32127281164986,
|
| 335 |
+
"learning_rate": 4.7165820642978e-07,
|
| 336 |
+
"logits/chosen": -0.007580566219985485,
|
| 337 |
+
"logits/rejected": -0.24066162109375,
|
| 338 |
+
"logps/chosen": -667.2000122070312,
|
| 339 |
+
"logps/rejected": -560.7999877929688,
|
| 340 |
+
"loss": 0.1741,
|
| 341 |
+
"nll_loss": 1.292578101158142,
|
| 342 |
+
"rewards/accuracies": 0.9375,
|
| 343 |
+
"rewards/chosen": -0.9162353277206421,
|
| 344 |
+
"rewards/margins": 11.543749809265137,
|
| 345 |
+
"rewards/rejected": -12.462499618530273,
|
| 346 |
+
"step": 200
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.1598173515981735,
|
| 350 |
+
"grad_norm": 1.1938779215691453,
|
| 351 |
+
"learning_rate": 4.674280879864636e-07,
|
| 352 |
+
"logits/chosen": 0.04982910305261612,
|
| 353 |
+
"logits/rejected": -0.3872314393520355,
|
| 354 |
+
"logps/chosen": -511.0,
|
| 355 |
+
"logps/rejected": -586.4000244140625,
|
| 356 |
+
"loss": 0.0423,
|
| 357 |
+
"nll_loss": 1.341406226158142,
|
| 358 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 359 |
+
"rewards/chosen": -1.2195312976837158,
|
| 360 |
+
"rewards/margins": 12.975000381469727,
|
| 361 |
+
"rewards/rejected": -14.199999809265137,
|
| 362 |
+
"step": 210
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.167427701674277,
|
| 366 |
+
"grad_norm": 16.933820639627886,
|
| 367 |
+
"learning_rate": 4.631979695431472e-07,
|
| 368 |
+
"logits/chosen": 0.14785155653953552,
|
| 369 |
+
"logits/rejected": -0.25715333223342896,
|
| 370 |
+
"logps/chosen": -497.3999938964844,
|
| 371 |
+
"logps/rejected": -499.79998779296875,
|
| 372 |
+
"loss": 0.0227,
|
| 373 |
+
"nll_loss": 1.182031273841858,
|
| 374 |
+
"rewards/accuracies": 1.0,
|
| 375 |
+
"rewards/chosen": -0.968945324420929,
|
| 376 |
+
"rewards/margins": 12.368749618530273,
|
| 377 |
+
"rewards/rejected": -13.337499618530273,
|
| 378 |
+
"step": 220
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.1750380517503805,
|
| 382 |
+
"grad_norm": 9.410898919627003,
|
| 383 |
+
"learning_rate": 4.5896785109983075e-07,
|
| 384 |
+
"logits/chosen": 0.06458739936351776,
|
| 385 |
+
"logits/rejected": -0.28864747285842896,
|
| 386 |
+
"logps/chosen": -571.2000122070312,
|
| 387 |
+
"logps/rejected": -523.0,
|
| 388 |
+
"loss": 0.0486,
|
| 389 |
+
"nll_loss": 1.304296851158142,
|
| 390 |
+
"rewards/accuracies": 0.987500011920929,
|
| 391 |
+
"rewards/chosen": -0.633251965045929,
|
| 392 |
+
"rewards/margins": 13.043749809265137,
|
| 393 |
+
"rewards/rejected": -13.681249618530273,
|
| 394 |
+
"step": 230
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.182648401826484,
|
| 398 |
+
"grad_norm": 0.1635849847821337,
|
| 399 |
+
"learning_rate": 4.5473773265651437e-07,
|
| 400 |
+
"logits/chosen": 0.11040039360523224,
|
| 401 |
+
"logits/rejected": -0.329833984375,
|
| 402 |
+
"logps/chosen": -493.20001220703125,
|
| 403 |
+
"logps/rejected": -563.0,
|
| 404 |
+
"loss": 0.0048,
|
| 405 |
+
"nll_loss": 1.19921875,
|
| 406 |
+
"rewards/accuracies": 1.0,
|
| 407 |
+
"rewards/chosen": 0.02285156212747097,
|
| 408 |
+
"rewards/margins": 13.193750381469727,
|
| 409 |
+
"rewards/rejected": -13.175000190734863,
|
| 410 |
+
"step": 240
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.1902587519025875,
|
| 414 |
+
"grad_norm": 10.949527892032492,
|
| 415 |
+
"learning_rate": 4.5050761421319793e-07,
|
| 416 |
+
"logits/chosen": 0.15396729111671448,
|
| 417 |
+
"logits/rejected": -0.292236328125,
|
| 418 |
+
"logps/chosen": -502.3999938964844,
|
| 419 |
+
"logps/rejected": -471.6000061035156,
|
| 420 |
+
"loss": 0.0033,
|
| 421 |
+
"nll_loss": 1.3406250476837158,
|
| 422 |
+
"rewards/accuracies": 1.0,
|
| 423 |
+
"rewards/chosen": 0.45576173067092896,
|
| 424 |
+
"rewards/margins": 14.675000190734863,
|
| 425 |
+
"rewards/rejected": -14.212499618530273,
|
| 426 |
+
"step": 250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.197869101978691,
|
| 430 |
+
"grad_norm": 2.4125036589922657,
|
| 431 |
+
"learning_rate": 4.4627749576988155e-07,
|
| 432 |
+
"logits/chosen": -0.010760498233139515,
|
| 433 |
+
"logits/rejected": -0.387939453125,
|
| 434 |
+
"logps/chosen": -512.7999877929688,
|
| 435 |
+
"logps/rejected": -590.4000244140625,
|
| 436 |
+
"loss": 0.0249,
|
| 437 |
+
"nll_loss": 1.3800780773162842,
|
| 438 |
+
"rewards/accuracies": 0.987500011920929,
|
| 439 |
+
"rewards/chosen": -1.225195288658142,
|
| 440 |
+
"rewards/margins": 15.28125,
|
| 441 |
+
"rewards/rejected": -16.487499237060547,
|
| 442 |
+
"step": 260
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.2054794520547945,
|
| 446 |
+
"grad_norm": 1.624318783089076,
|
| 447 |
+
"learning_rate": 4.420473773265651e-07,
|
| 448 |
+
"logits/chosen": 0.0046142577193677425,
|
| 449 |
+
"logits/rejected": -0.34760743379592896,
|
| 450 |
+
"logps/chosen": -759.4000244140625,
|
| 451 |
+
"logps/rejected": -594.4000244140625,
|
| 452 |
+
"loss": 0.024,
|
| 453 |
+
"nll_loss": 1.3757812976837158,
|
| 454 |
+
"rewards/accuracies": 0.987500011920929,
|
| 455 |
+
"rewards/chosen": -1.7131836414337158,
|
| 456 |
+
"rewards/margins": 13.300000190734863,
|
| 457 |
+
"rewards/rejected": -14.987500190734863,
|
| 458 |
+
"step": 270
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.213089802130898,
|
| 462 |
+
"grad_norm": 0.03610075868390092,
|
| 463 |
+
"learning_rate": 4.3781725888324873e-07,
|
| 464 |
+
"logits/chosen": -0.02946777269244194,
|
| 465 |
+
"logits/rejected": -0.2095947265625,
|
| 466 |
+
"logps/chosen": -504.0,
|
| 467 |
+
"logps/rejected": -592.5999755859375,
|
| 468 |
+
"loss": 0.0122,
|
| 469 |
+
"nll_loss": 1.328125,
|
| 470 |
+
"rewards/accuracies": 1.0,
|
| 471 |
+
"rewards/chosen": -2.0609374046325684,
|
| 472 |
+
"rewards/margins": 13.762499809265137,
|
| 473 |
+
"rewards/rejected": -15.793749809265137,
|
| 474 |
+
"step": 280
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.2207001522070015,
|
| 478 |
+
"grad_norm": 19.136550803392385,
|
| 479 |
+
"learning_rate": 4.335871404399323e-07,
|
| 480 |
+
"logits/chosen": 0.07322998344898224,
|
| 481 |
+
"logits/rejected": -0.39726561307907104,
|
| 482 |
+
"logps/chosen": -512.0,
|
| 483 |
+
"logps/rejected": -639.5999755859375,
|
| 484 |
+
"loss": 0.0216,
|
| 485 |
+
"nll_loss": 1.3984375,
|
| 486 |
+
"rewards/accuracies": 0.987500011920929,
|
| 487 |
+
"rewards/chosen": -1.4064452648162842,
|
| 488 |
+
"rewards/margins": 14.287500381469727,
|
| 489 |
+
"rewards/rejected": -15.693750381469727,
|
| 490 |
+
"step": 290
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.228310502283105,
|
| 494 |
+
"grad_norm": 0.4497174950007498,
|
| 495 |
+
"learning_rate": 4.2935702199661586e-07,
|
| 496 |
+
"logits/chosen": 0.115234375,
|
| 497 |
+
"logits/rejected": -0.3787597715854645,
|
| 498 |
+
"logps/chosen": -490.6000061035156,
|
| 499 |
+
"logps/rejected": -584.5999755859375,
|
| 500 |
+
"loss": 0.0102,
|
| 501 |
+
"nll_loss": 1.2195312976837158,
|
| 502 |
+
"rewards/accuracies": 1.0,
|
| 503 |
+
"rewards/chosen": -0.11577148735523224,
|
| 504 |
+
"rewards/margins": 15.6875,
|
| 505 |
+
"rewards/rejected": -15.806249618530273,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.2359208523592085,
|
| 510 |
+
"grad_norm": 10.781205925954207,
|
| 511 |
+
"learning_rate": 4.251269035532995e-07,
|
| 512 |
+
"logits/chosen": 0.10917969048023224,
|
| 513 |
+
"logits/rejected": -0.390380859375,
|
| 514 |
+
"logps/chosen": -519.7999877929688,
|
| 515 |
+
"logps/rejected": -509.3999938964844,
|
| 516 |
+
"loss": 0.0154,
|
| 517 |
+
"nll_loss": 1.3796875476837158,
|
| 518 |
+
"rewards/accuracies": 1.0,
|
| 519 |
+
"rewards/chosen": 0.30937498807907104,
|
| 520 |
+
"rewards/margins": 13.143750190734863,
|
| 521 |
+
"rewards/rejected": -12.831250190734863,
|
| 522 |
+
"step": 310
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.243531202435312,
|
| 526 |
+
"grad_norm": 0.09232446693354569,
|
| 527 |
+
"learning_rate": 4.2089678510998305e-07,
|
| 528 |
+
"logits/chosen": 0.063690185546875,
|
| 529 |
+
"logits/rejected": -0.40312498807907104,
|
| 530 |
+
"logps/chosen": -637.2000122070312,
|
| 531 |
+
"logps/rejected": -632.2000122070312,
|
| 532 |
+
"loss": 0.002,
|
| 533 |
+
"nll_loss": 1.5554687976837158,
|
| 534 |
+
"rewards/accuracies": 1.0,
|
| 535 |
+
"rewards/chosen": 0.2099609375,
|
| 536 |
+
"rewards/margins": 16.40625,
|
| 537 |
+
"rewards/rejected": -16.225000381469727,
|
| 538 |
+
"step": 320
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.2511415525114155,
|
| 542 |
+
"grad_norm": 2.1956385644079557,
|
| 543 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 544 |
+
"logits/chosen": 0.09484253078699112,
|
| 545 |
+
"logits/rejected": -0.4134765565395355,
|
| 546 |
+
"logps/chosen": -451.6000061035156,
|
| 547 |
+
"logps/rejected": -508.79998779296875,
|
| 548 |
+
"loss": 0.0241,
|
| 549 |
+
"nll_loss": 1.285546898841858,
|
| 550 |
+
"rewards/accuracies": 0.987500011920929,
|
| 551 |
+
"rewards/chosen": -1.09765625,
|
| 552 |
+
"rewards/margins": 15.831250190734863,
|
| 553 |
+
"rewards/rejected": -16.943750381469727,
|
| 554 |
+
"step": 330
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.258751902587519,
|
| 558 |
+
"grad_norm": 0.6192792844136195,
|
| 559 |
+
"learning_rate": 4.1243654822335023e-07,
|
| 560 |
+
"logits/chosen": 0.03125,
|
| 561 |
+
"logits/rejected": -0.37730711698532104,
|
| 562 |
+
"logps/chosen": -510.3999938964844,
|
| 563 |
+
"logps/rejected": -518.4000244140625,
|
| 564 |
+
"loss": 0.0082,
|
| 565 |
+
"nll_loss": 1.2136719226837158,
|
| 566 |
+
"rewards/accuracies": 1.0,
|
| 567 |
+
"rewards/chosen": -0.24785156548023224,
|
| 568 |
+
"rewards/margins": 15.975000381469727,
|
| 569 |
+
"rewards/rejected": -16.237499237060547,
|
| 570 |
+
"step": 340
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.2663622526636225,
|
| 574 |
+
"grad_norm": 0.08449334028664242,
|
| 575 |
+
"learning_rate": 4.0820642978003385e-07,
|
| 576 |
+
"logits/chosen": 0.09104003757238388,
|
| 577 |
+
"logits/rejected": -0.3393310606479645,
|
| 578 |
+
"logps/chosen": -715.4000244140625,
|
| 579 |
+
"logps/rejected": -643.7999877929688,
|
| 580 |
+
"loss": 0.0462,
|
| 581 |
+
"nll_loss": 1.3984375,
|
| 582 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 583 |
+
"rewards/chosen": -0.2917724549770355,
|
| 584 |
+
"rewards/margins": 17.450000762939453,
|
| 585 |
+
"rewards/rejected": -17.737499237060547,
|
| 586 |
+
"step": 350
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.273972602739726,
|
| 590 |
+
"grad_norm": 0.05520971780153057,
|
| 591 |
+
"learning_rate": 4.039763113367174e-07,
|
| 592 |
+
"logits/chosen": 0.003955078311264515,
|
| 593 |
+
"logits/rejected": -0.4921875,
|
| 594 |
+
"logps/chosen": -622.5999755859375,
|
| 595 |
+
"logps/rejected": -616.4000244140625,
|
| 596 |
+
"loss": 0.002,
|
| 597 |
+
"nll_loss": 1.2628905773162842,
|
| 598 |
+
"rewards/accuracies": 1.0,
|
| 599 |
+
"rewards/chosen": -0.24228516221046448,
|
| 600 |
+
"rewards/margins": 16.981250762939453,
|
| 601 |
+
"rewards/rejected": -17.21875,
|
| 602 |
+
"step": 360
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.2815829528158295,
|
| 606 |
+
"grad_norm": 0.017612797398325684,
|
| 607 |
+
"learning_rate": 3.99746192893401e-07,
|
| 608 |
+
"logits/chosen": 0.15422363579273224,
|
| 609 |
+
"logits/rejected": -0.34794920682907104,
|
| 610 |
+
"logps/chosen": -646.2000122070312,
|
| 611 |
+
"logps/rejected": -551.7999877929688,
|
| 612 |
+
"loss": 0.0023,
|
| 613 |
+
"nll_loss": 1.19921875,
|
| 614 |
+
"rewards/accuracies": 1.0,
|
| 615 |
+
"rewards/chosen": 1.68359375,
|
| 616 |
+
"rewards/margins": 17.524999618530273,
|
| 617 |
+
"rewards/rejected": -15.837499618530273,
|
| 618 |
+
"step": 370
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.289193302891933,
|
| 622 |
+
"grad_norm": 1.1857663083986685,
|
| 623 |
+
"learning_rate": 3.9551607445008455e-07,
|
| 624 |
+
"logits/chosen": 0.06464843451976776,
|
| 625 |
+
"logits/rejected": -0.39179688692092896,
|
| 626 |
+
"logps/chosen": -589.4000244140625,
|
| 627 |
+
"logps/rejected": -574.2000122070312,
|
| 628 |
+
"loss": 0.0027,
|
| 629 |
+
"nll_loss": 1.2585937976837158,
|
| 630 |
+
"rewards/accuracies": 1.0,
|
| 631 |
+
"rewards/chosen": 0.4532714784145355,
|
| 632 |
+
"rewards/margins": 18.387500762939453,
|
| 633 |
+
"rewards/rejected": -17.956249237060547,
|
| 634 |
+
"step": 380
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.2968036529680365,
|
| 638 |
+
"grad_norm": 0.3144677528152258,
|
| 639 |
+
"learning_rate": 3.9128595600676816e-07,
|
| 640 |
+
"logits/chosen": 0.07275390625,
|
| 641 |
+
"logits/rejected": -0.38837891817092896,
|
| 642 |
+
"logps/chosen": -570.7999877929688,
|
| 643 |
+
"logps/rejected": -620.0,
|
| 644 |
+
"loss": 0.002,
|
| 645 |
+
"nll_loss": 1.288671851158142,
|
| 646 |
+
"rewards/accuracies": 1.0,
|
| 647 |
+
"rewards/chosen": 0.03593749925494194,
|
| 648 |
+
"rewards/margins": 18.381250381469727,
|
| 649 |
+
"rewards/rejected": -18.34375,
|
| 650 |
+
"step": 390
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.30441400304414,
|
| 654 |
+
"grad_norm": 8.00145866519447,
|
| 655 |
+
"learning_rate": 3.870558375634518e-07,
|
| 656 |
+
"logits/chosen": 0.05874023586511612,
|
| 657 |
+
"logits/rejected": -0.3757568299770355,
|
| 658 |
+
"logps/chosen": -622.2000122070312,
|
| 659 |
+
"logps/rejected": -608.0,
|
| 660 |
+
"loss": 0.0044,
|
| 661 |
+
"nll_loss": 1.279687523841858,
|
| 662 |
+
"rewards/accuracies": 1.0,
|
| 663 |
+
"rewards/chosen": -0.05029296875,
|
| 664 |
+
"rewards/margins": 18.881250381469727,
|
| 665 |
+
"rewards/rejected": -18.9375,
|
| 666 |
+
"step": 400
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.3120243531202435,
|
| 670 |
+
"grad_norm": 0.02899554688728597,
|
| 671 |
+
"learning_rate": 3.8282571912013535e-07,
|
| 672 |
+
"logits/chosen": 0.11918945610523224,
|
| 673 |
+
"logits/rejected": -0.43144530057907104,
|
| 674 |
+
"logps/chosen": -621.7999877929688,
|
| 675 |
+
"logps/rejected": -600.4000244140625,
|
| 676 |
+
"loss": 0.0025,
|
| 677 |
+
"nll_loss": 1.2492187023162842,
|
| 678 |
+
"rewards/accuracies": 1.0,
|
| 679 |
+
"rewards/chosen": -0.79296875,
|
| 680 |
+
"rewards/margins": 19.837499618530273,
|
| 681 |
+
"rewards/rejected": -20.606250762939453,
|
| 682 |
+
"step": 410
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.319634703196347,
|
| 686 |
+
"grad_norm": 1.4324728498112853,
|
| 687 |
+
"learning_rate": 3.7859560067681897e-07,
|
| 688 |
+
"logits/chosen": 0.11809082329273224,
|
| 689 |
+
"logits/rejected": -0.3350585997104645,
|
| 690 |
+
"logps/chosen": -574.7999877929688,
|
| 691 |
+
"logps/rejected": -569.7999877929688,
|
| 692 |
+
"loss": 0.0024,
|
| 693 |
+
"nll_loss": 1.163671851158142,
|
| 694 |
+
"rewards/accuracies": 1.0,
|
| 695 |
+
"rewards/chosen": -0.234130859375,
|
| 696 |
+
"rewards/margins": 20.043750762939453,
|
| 697 |
+
"rewards/rejected": -20.274999618530273,
|
| 698 |
+
"step": 420
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.3272450532724505,
|
| 702 |
+
"grad_norm": 83.90065360414611,
|
| 703 |
+
"learning_rate": 3.7436548223350253e-07,
|
| 704 |
+
"logits/chosen": 0.06676025688648224,
|
| 705 |
+
"logits/rejected": -0.47773438692092896,
|
| 706 |
+
"logps/chosen": -645.0,
|
| 707 |
+
"logps/rejected": -627.2000122070312,
|
| 708 |
+
"loss": 0.0078,
|
| 709 |
+
"nll_loss": 1.335546851158142,
|
| 710 |
+
"rewards/accuracies": 1.0,
|
| 711 |
+
"rewards/chosen": 0.05810546875,
|
| 712 |
+
"rewards/margins": 21.206249237060547,
|
| 713 |
+
"rewards/rejected": -21.162500381469727,
|
| 714 |
+
"step": 430
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.334855403348554,
|
| 718 |
+
"grad_norm": 2.461542845657294,
|
| 719 |
+
"learning_rate": 3.701353637901861e-07,
|
| 720 |
+
"logits/chosen": 0.12839356064796448,
|
| 721 |
+
"logits/rejected": -0.4951171875,
|
| 722 |
+
"logps/chosen": -557.7999877929688,
|
| 723 |
+
"logps/rejected": -610.4000244140625,
|
| 724 |
+
"loss": 0.0211,
|
| 725 |
+
"nll_loss": 1.2726562023162842,
|
| 726 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 727 |
+
"rewards/chosen": -0.33623045682907104,
|
| 728 |
+
"rewards/margins": 20.799999237060547,
|
| 729 |
+
"rewards/rejected": -21.143749237060547,
|
| 730 |
+
"step": 440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.3424657534246575,
|
| 734 |
+
"grad_norm": 3.4782361709714933,
|
| 735 |
+
"learning_rate": 3.6590524534686966e-07,
|
| 736 |
+
"logits/chosen": 0.08744506537914276,
|
| 737 |
+
"logits/rejected": -0.35893553495407104,
|
| 738 |
+
"logps/chosen": -479.0,
|
| 739 |
+
"logps/rejected": -553.4000244140625,
|
| 740 |
+
"loss": 0.0063,
|
| 741 |
+
"nll_loss": 1.3468749523162842,
|
| 742 |
+
"rewards/accuracies": 1.0,
|
| 743 |
+
"rewards/chosen": -0.5337890386581421,
|
| 744 |
+
"rewards/margins": 18.975000381469727,
|
| 745 |
+
"rewards/rejected": -19.506250381469727,
|
| 746 |
+
"step": 450
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.350076103500761,
|
| 750 |
+
"grad_norm": 7.299829329882369,
|
| 751 |
+
"learning_rate": 3.616751269035533e-07,
|
| 752 |
+
"logits/chosen": 0.103759765625,
|
| 753 |
+
"logits/rejected": -0.3382812440395355,
|
| 754 |
+
"logps/chosen": -540.5999755859375,
|
| 755 |
+
"logps/rejected": -605.7999877929688,
|
| 756 |
+
"loss": 0.0114,
|
| 757 |
+
"nll_loss": 1.2082030773162842,
|
| 758 |
+
"rewards/accuracies": 1.0,
|
| 759 |
+
"rewards/chosen": 1.527246117591858,
|
| 760 |
+
"rewards/margins": 19.806249618530273,
|
| 761 |
+
"rewards/rejected": -18.293750762939453,
|
| 762 |
+
"step": 460
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.3576864535768645,
|
| 766 |
+
"grad_norm": 0.04936824246428895,
|
| 767 |
+
"learning_rate": 3.574450084602369e-07,
|
| 768 |
+
"logits/chosen": 0.009326172061264515,
|
| 769 |
+
"logits/rejected": -0.4322265684604645,
|
| 770 |
+
"logps/chosen": -656.5999755859375,
|
| 771 |
+
"logps/rejected": -630.7999877929688,
|
| 772 |
+
"loss": 0.1617,
|
| 773 |
+
"nll_loss": 1.392187476158142,
|
| 774 |
+
"rewards/accuracies": 0.987500011920929,
|
| 775 |
+
"rewards/chosen": 0.644726574420929,
|
| 776 |
+
"rewards/margins": 21.206249237060547,
|
| 777 |
+
"rewards/rejected": -20.556249618530273,
|
| 778 |
+
"step": 470
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.365296803652968,
|
| 782 |
+
"grad_norm": 0.03612475204041786,
|
| 783 |
+
"learning_rate": 3.5321489001692046e-07,
|
| 784 |
+
"logits/chosen": 0.10490722954273224,
|
| 785 |
+
"logits/rejected": -0.46660155057907104,
|
| 786 |
+
"logps/chosen": -539.0,
|
| 787 |
+
"logps/rejected": -592.5999755859375,
|
| 788 |
+
"loss": 0.0108,
|
| 789 |
+
"nll_loss": 1.284765601158142,
|
| 790 |
+
"rewards/accuracies": 1.0,
|
| 791 |
+
"rewards/chosen": 0.5726562738418579,
|
| 792 |
+
"rewards/margins": 21.325000762939453,
|
| 793 |
+
"rewards/rejected": -20.762500762939453,
|
| 794 |
+
"step": 480
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.3729071537290715,
|
| 798 |
+
"grad_norm": 7.151199652416472,
|
| 799 |
+
"learning_rate": 3.489847715736041e-07,
|
| 800 |
+
"logits/chosen": 0.15537109971046448,
|
| 801 |
+
"logits/rejected": -0.4066406190395355,
|
| 802 |
+
"logps/chosen": -491.70001220703125,
|
| 803 |
+
"logps/rejected": -558.2000122070312,
|
| 804 |
+
"loss": 0.003,
|
| 805 |
+
"nll_loss": 1.3078124523162842,
|
| 806 |
+
"rewards/accuracies": 1.0,
|
| 807 |
+
"rewards/chosen": 0.071533203125,
|
| 808 |
+
"rewards/margins": 19.600000381469727,
|
| 809 |
+
"rewards/rejected": -19.524999618530273,
|
| 810 |
+
"step": 490
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.380517503805175,
|
| 814 |
+
"grad_norm": 0.01609195103759266,
|
| 815 |
+
"learning_rate": 3.447546531302876e-07,
|
| 816 |
+
"logits/chosen": 0.1488037109375,
|
| 817 |
+
"logits/rejected": -0.3904785215854645,
|
| 818 |
+
"logps/chosen": -482.3999938964844,
|
| 819 |
+
"logps/rejected": -668.2000122070312,
|
| 820 |
+
"loss": 0.127,
|
| 821 |
+
"nll_loss": 1.2683594226837158,
|
| 822 |
+
"rewards/accuracies": 0.987500011920929,
|
| 823 |
+
"rewards/chosen": -1.641992211341858,
|
| 824 |
+
"rewards/margins": 18.799999237060547,
|
| 825 |
+
"rewards/rejected": -20.412500381469727,
|
| 826 |
+
"step": 500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.3881278538812785,
|
| 830 |
+
"grad_norm": 0.4256994501063543,
|
| 831 |
+
"learning_rate": 3.405245346869712e-07,
|
| 832 |
+
"logits/chosen": 0.12789306044578552,
|
| 833 |
+
"logits/rejected": -0.4141601622104645,
|
| 834 |
+
"logps/chosen": -639.4000244140625,
|
| 835 |
+
"logps/rejected": -642.4000244140625,
|
| 836 |
+
"loss": 0.0035,
|
| 837 |
+
"nll_loss": 1.3367187976837158,
|
| 838 |
+
"rewards/accuracies": 1.0,
|
| 839 |
+
"rewards/chosen": 1.034570336341858,
|
| 840 |
+
"rewards/margins": 19.975000381469727,
|
| 841 |
+
"rewards/rejected": -18.943750381469727,
|
| 842 |
+
"step": 510
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.395738203957382,
|
| 846 |
+
"grad_norm": 1.6580894257634202,
|
| 847 |
+
"learning_rate": 3.362944162436548e-07,
|
| 848 |
+
"logits/chosen": 0.3023437559604645,
|
| 849 |
+
"logits/rejected": -0.39238280057907104,
|
| 850 |
+
"logps/chosen": -514.2000122070312,
|
| 851 |
+
"logps/rejected": -526.2000122070312,
|
| 852 |
+
"loss": 0.0165,
|
| 853 |
+
"nll_loss": 1.3429687023162842,
|
| 854 |
+
"rewards/accuracies": 0.987500011920929,
|
| 855 |
+
"rewards/chosen": 1.346044898033142,
|
| 856 |
+
"rewards/margins": 16.962499618530273,
|
| 857 |
+
"rewards/rejected": -15.643750190734863,
|
| 858 |
+
"step": 520
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.4033485540334855,
|
| 862 |
+
"grad_norm": 0.3626855176795643,
|
| 863 |
+
"learning_rate": 3.320642978003384e-07,
|
| 864 |
+
"logits/chosen": 0.2386474609375,
|
| 865 |
+
"logits/rejected": -0.24819335341453552,
|
| 866 |
+
"logps/chosen": -566.9000244140625,
|
| 867 |
+
"logps/rejected": -600.4000244140625,
|
| 868 |
+
"loss": 0.0054,
|
| 869 |
+
"nll_loss": 1.373046875,
|
| 870 |
+
"rewards/accuracies": 1.0,
|
| 871 |
+
"rewards/chosen": 1.1706206798553467,
|
| 872 |
+
"rewards/margins": 18.3125,
|
| 873 |
+
"rewards/rejected": -17.118749618530273,
|
| 874 |
+
"step": 530
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.410958904109589,
|
| 878 |
+
"grad_norm": 1.6989362396908643,
|
| 879 |
+
"learning_rate": 3.2783417935702196e-07,
|
| 880 |
+
"logits/chosen": 0.22885742783546448,
|
| 881 |
+
"logits/rejected": -0.3630127012729645,
|
| 882 |
+
"logps/chosen": -439.3999938964844,
|
| 883 |
+
"logps/rejected": -524.5999755859375,
|
| 884 |
+
"loss": 0.013,
|
| 885 |
+
"nll_loss": 1.192968726158142,
|
| 886 |
+
"rewards/accuracies": 1.0,
|
| 887 |
+
"rewards/chosen": 0.7841796875,
|
| 888 |
+
"rewards/margins": 17.862499237060547,
|
| 889 |
+
"rewards/rejected": -17.075000762939453,
|
| 890 |
+
"step": 540
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.4185692541856925,
|
| 894 |
+
"grad_norm": 0.10460634430017504,
|
| 895 |
+
"learning_rate": 3.236040609137056e-07,
|
| 896 |
+
"logits/chosen": 0.23752442002296448,
|
| 897 |
+
"logits/rejected": -0.3759765625,
|
| 898 |
+
"logps/chosen": -479.20001220703125,
|
| 899 |
+
"logps/rejected": -587.0,
|
| 900 |
+
"loss": 0.0028,
|
| 901 |
+
"nll_loss": 1.346093773841858,
|
| 902 |
+
"rewards/accuracies": 1.0,
|
| 903 |
+
"rewards/chosen": 2.2855467796325684,
|
| 904 |
+
"rewards/margins": 20.6875,
|
| 905 |
+
"rewards/rejected": -18.381250381469727,
|
| 906 |
+
"step": 550
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.426179604261796,
|
| 910 |
+
"grad_norm": 0.032351935939549575,
|
| 911 |
+
"learning_rate": 3.193739424703892e-07,
|
| 912 |
+
"logits/chosen": 0.12680664658546448,
|
| 913 |
+
"logits/rejected": -0.36494141817092896,
|
| 914 |
+
"logps/chosen": -557.2000122070312,
|
| 915 |
+
"logps/rejected": -624.5999755859375,
|
| 916 |
+
"loss": 0.0055,
|
| 917 |
+
"nll_loss": 1.223046898841858,
|
| 918 |
+
"rewards/accuracies": 1.0,
|
| 919 |
+
"rewards/chosen": 0.3065429627895355,
|
| 920 |
+
"rewards/margins": 20.549999237060547,
|
| 921 |
+
"rewards/rejected": -20.237499237060547,
|
| 922 |
+
"step": 560
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.4337899543378995,
|
| 926 |
+
"grad_norm": 0.2622584796542904,
|
| 927 |
+
"learning_rate": 3.151438240270727e-07,
|
| 928 |
+
"logits/chosen": 0.04736328125,
|
| 929 |
+
"logits/rejected": -0.40087890625,
|
| 930 |
+
"logps/chosen": -561.0,
|
| 931 |
+
"logps/rejected": -774.0,
|
| 932 |
+
"loss": 0.0023,
|
| 933 |
+
"nll_loss": 1.2335937023162842,
|
| 934 |
+
"rewards/accuracies": 1.0,
|
| 935 |
+
"rewards/chosen": -1.651953101158142,
|
| 936 |
+
"rewards/margins": 23.549999237060547,
|
| 937 |
+
"rewards/rejected": -25.225000381469727,
|
| 938 |
+
"step": 570
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.441400304414003,
|
| 942 |
+
"grad_norm": 0.25779336035907713,
|
| 943 |
+
"learning_rate": 3.1091370558375633e-07,
|
| 944 |
+
"logits/chosen": 0.17866210639476776,
|
| 945 |
+
"logits/rejected": -0.5062500238418579,
|
| 946 |
+
"logps/chosen": -628.7999877929688,
|
| 947 |
+
"logps/rejected": -634.7999877929688,
|
| 948 |
+
"loss": 0.0042,
|
| 949 |
+
"nll_loss": 1.2957031726837158,
|
| 950 |
+
"rewards/accuracies": 1.0,
|
| 951 |
+
"rewards/chosen": -0.9549804925918579,
|
| 952 |
+
"rewards/margins": 23.575000762939453,
|
| 953 |
+
"rewards/rejected": -24.53125,
|
| 954 |
+
"step": 580
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.4490106544901065,
|
| 958 |
+
"grad_norm": 0.11822977784077095,
|
| 959 |
+
"learning_rate": 3.066835871404399e-07,
|
| 960 |
+
"logits/chosen": 0.11000976711511612,
|
| 961 |
+
"logits/rejected": -0.48027342557907104,
|
| 962 |
+
"logps/chosen": -501.20001220703125,
|
| 963 |
+
"logps/rejected": -692.7999877929688,
|
| 964 |
+
"loss": 0.0343,
|
| 965 |
+
"nll_loss": 1.326562523841858,
|
| 966 |
+
"rewards/accuracies": 0.987500011920929,
|
| 967 |
+
"rewards/chosen": -0.61181640625,
|
| 968 |
+
"rewards/margins": 24.106250762939453,
|
| 969 |
+
"rewards/rejected": -24.712499618530273,
|
| 970 |
+
"step": 590
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.45662100456621,
|
| 974 |
+
"grad_norm": 0.02882989474463555,
|
| 975 |
+
"learning_rate": 3.024534686971235e-07,
|
| 976 |
+
"logits/chosen": 0.03955078125,
|
| 977 |
+
"logits/rejected": -0.4144531190395355,
|
| 978 |
+
"logps/chosen": -494.6000061035156,
|
| 979 |
+
"logps/rejected": -598.4000244140625,
|
| 980 |
+
"loss": 0.0099,
|
| 981 |
+
"nll_loss": 1.211328148841858,
|
| 982 |
+
"rewards/accuracies": 0.987500011920929,
|
| 983 |
+
"rewards/chosen": -0.780468761920929,
|
| 984 |
+
"rewards/margins": 23.524999618530273,
|
| 985 |
+
"rewards/rejected": -24.287500381469727,
|
| 986 |
+
"step": 600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.4642313546423135,
|
| 990 |
+
"grad_norm": 0.034483519735511114,
|
| 991 |
+
"learning_rate": 2.982233502538071e-07,
|
| 992 |
+
"logits/chosen": 0.05483398586511612,
|
| 993 |
+
"logits/rejected": -0.366668701171875,
|
| 994 |
+
"logps/chosen": -612.2000122070312,
|
| 995 |
+
"logps/rejected": -622.2000122070312,
|
| 996 |
+
"loss": 0.0174,
|
| 997 |
+
"nll_loss": 1.28515625,
|
| 998 |
+
"rewards/accuracies": 0.987500011920929,
|
| 999 |
+
"rewards/chosen": -1.4785277843475342,
|
| 1000 |
+
"rewards/margins": 23.712499618530273,
|
| 1001 |
+
"rewards/rejected": -25.174999237060547,
|
| 1002 |
+
"step": 610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.471841704718417,
|
| 1006 |
+
"grad_norm": 2.5644101799929846,
|
| 1007 |
+
"learning_rate": 2.939932318104907e-07,
|
| 1008 |
+
"logits/chosen": 0.11363525688648224,
|
| 1009 |
+
"logits/rejected": -0.40190428495407104,
|
| 1010 |
+
"logps/chosen": -418.0,
|
| 1011 |
+
"logps/rejected": -549.5999755859375,
|
| 1012 |
+
"loss": 0.0109,
|
| 1013 |
+
"nll_loss": 1.171875,
|
| 1014 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1015 |
+
"rewards/chosen": -1.073632836341858,
|
| 1016 |
+
"rewards/margins": 20.625,
|
| 1017 |
+
"rewards/rejected": -21.6875,
|
| 1018 |
+
"step": 620
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.4794520547945205,
|
| 1022 |
+
"grad_norm": 2.6373510806137754,
|
| 1023 |
+
"learning_rate": 2.897631133671743e-07,
|
| 1024 |
+
"logits/chosen": 0.2061714231967926,
|
| 1025 |
+
"logits/rejected": -0.4150390625,
|
| 1026 |
+
"logps/chosen": -542.5999755859375,
|
| 1027 |
+
"logps/rejected": -563.7999877929688,
|
| 1028 |
+
"loss": 0.0023,
|
| 1029 |
+
"nll_loss": 1.240234375,
|
| 1030 |
+
"rewards/accuracies": 1.0,
|
| 1031 |
+
"rewards/chosen": 0.03828125074505806,
|
| 1032 |
+
"rewards/margins": 19.537500381469727,
|
| 1033 |
+
"rewards/rejected": -19.481250762939453,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.487062404870624,
|
| 1038 |
+
"grad_norm": 0.050167363326375274,
|
| 1039 |
+
"learning_rate": 2.8553299492385783e-07,
|
| 1040 |
+
"logits/chosen": -0.01143798790872097,
|
| 1041 |
+
"logits/rejected": -0.4168945252895355,
|
| 1042 |
+
"logps/chosen": -618.5999755859375,
|
| 1043 |
+
"logps/rejected": -646.7999877929688,
|
| 1044 |
+
"loss": 0.0128,
|
| 1045 |
+
"nll_loss": 1.254296898841858,
|
| 1046 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1047 |
+
"rewards/chosen": -0.3956054747104645,
|
| 1048 |
+
"rewards/margins": 20.131250381469727,
|
| 1049 |
+
"rewards/rejected": -20.524999618530273,
|
| 1050 |
+
"step": 640
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.4946727549467275,
|
| 1054 |
+
"grad_norm": 0.43797209086876404,
|
| 1055 |
+
"learning_rate": 2.8130287648054145e-07,
|
| 1056 |
+
"logits/chosen": 0.18203124403953552,
|
| 1057 |
+
"logits/rejected": -0.3395019471645355,
|
| 1058 |
+
"logps/chosen": -506.5,
|
| 1059 |
+
"logps/rejected": -598.7999877929688,
|
| 1060 |
+
"loss": 0.006,
|
| 1061 |
+
"nll_loss": 1.189453125,
|
| 1062 |
+
"rewards/accuracies": 1.0,
|
| 1063 |
+
"rewards/chosen": 0.10546875,
|
| 1064 |
+
"rewards/margins": 19.006250381469727,
|
| 1065 |
+
"rewards/rejected": -18.887500762939453,
|
| 1066 |
+
"step": 650
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.502283105022831,
|
| 1070 |
+
"grad_norm": 0.3407013367264488,
|
| 1071 |
+
"learning_rate": 2.77072758037225e-07,
|
| 1072 |
+
"logits/chosen": 0.25898438692092896,
|
| 1073 |
+
"logits/rejected": -0.41328126192092896,
|
| 1074 |
+
"logps/chosen": -508.6000061035156,
|
| 1075 |
+
"logps/rejected": -587.5999755859375,
|
| 1076 |
+
"loss": 0.0017,
|
| 1077 |
+
"nll_loss": 1.240625023841858,
|
| 1078 |
+
"rewards/accuracies": 1.0,
|
| 1079 |
+
"rewards/chosen": 1.7921874523162842,
|
| 1080 |
+
"rewards/margins": 20.112499237060547,
|
| 1081 |
+
"rewards/rejected": -18.293750762939453,
|
| 1082 |
+
"step": 660
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.5098934550989346,
|
| 1086 |
+
"grad_norm": 0.058867537956072004,
|
| 1087 |
+
"learning_rate": 2.7284263959390863e-07,
|
| 1088 |
+
"logits/chosen": 0.03278503566980362,
|
| 1089 |
+
"logits/rejected": -0.47998046875,
|
| 1090 |
+
"logps/chosen": -616.9000244140625,
|
| 1091 |
+
"logps/rejected": -652.0,
|
| 1092 |
+
"loss": 0.033,
|
| 1093 |
+
"nll_loss": 1.4207031726837158,
|
| 1094 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1095 |
+
"rewards/chosen": 0.4398437440395355,
|
| 1096 |
+
"rewards/margins": 19.725000381469727,
|
| 1097 |
+
"rewards/rejected": -19.325000762939453,
|
| 1098 |
+
"step": 670
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.517503805175038,
|
| 1102 |
+
"grad_norm": 0.04388996236374909,
|
| 1103 |
+
"learning_rate": 2.686125211505922e-07,
|
| 1104 |
+
"logits/chosen": 0.07414551079273224,
|
| 1105 |
+
"logits/rejected": -0.4593749940395355,
|
| 1106 |
+
"logps/chosen": -629.4000244140625,
|
| 1107 |
+
"logps/rejected": -614.5999755859375,
|
| 1108 |
+
"loss": 0.0022,
|
| 1109 |
+
"nll_loss": 1.5242187976837158,
|
| 1110 |
+
"rewards/accuracies": 1.0,
|
| 1111 |
+
"rewards/chosen": 2.062304735183716,
|
| 1112 |
+
"rewards/margins": 19.9375,
|
| 1113 |
+
"rewards/rejected": -17.881250381469727,
|
| 1114 |
+
"step": 680
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.5251141552511416,
|
| 1118 |
+
"grad_norm": 0.021018531124894033,
|
| 1119 |
+
"learning_rate": 2.643824027072758e-07,
|
| 1120 |
+
"logits/chosen": 0.15976563096046448,
|
| 1121 |
+
"logits/rejected": -0.43769532442092896,
|
| 1122 |
+
"logps/chosen": -534.7999877929688,
|
| 1123 |
+
"logps/rejected": -596.7999877929688,
|
| 1124 |
+
"loss": 0.008,
|
| 1125 |
+
"nll_loss": 1.373437523841858,
|
| 1126 |
+
"rewards/accuracies": 1.0,
|
| 1127 |
+
"rewards/chosen": 1.443750023841858,
|
| 1128 |
+
"rewards/margins": 19.8125,
|
| 1129 |
+
"rewards/rejected": -18.387500762939453,
|
| 1130 |
+
"step": 690
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.532724505327245,
|
| 1134 |
+
"grad_norm": 834.7732657475113,
|
| 1135 |
+
"learning_rate": 2.601522842639594e-07,
|
| 1136 |
+
"logits/chosen": 0.16179199516773224,
|
| 1137 |
+
"logits/rejected": -0.41254884004592896,
|
| 1138 |
+
"logps/chosen": -592.9000244140625,
|
| 1139 |
+
"logps/rejected": -707.0,
|
| 1140 |
+
"loss": 0.0429,
|
| 1141 |
+
"nll_loss": 1.241796851158142,
|
| 1142 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1143 |
+
"rewards/chosen": 1.0242187976837158,
|
| 1144 |
+
"rewards/margins": 19.362499237060547,
|
| 1145 |
+
"rewards/rejected": -18.350000381469727,
|
| 1146 |
+
"step": 700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5403348554033486,
|
| 1150 |
+
"grad_norm": 0.01714945296670409,
|
| 1151 |
+
"learning_rate": 2.5592216582064294e-07,
|
| 1152 |
+
"logits/chosen": 0.216552734375,
|
| 1153 |
+
"logits/rejected": -0.5347656011581421,
|
| 1154 |
+
"logps/chosen": -429.79998779296875,
|
| 1155 |
+
"logps/rejected": -547.7999877929688,
|
| 1156 |
+
"loss": 0.0037,
|
| 1157 |
+
"nll_loss": 1.252343773841858,
|
| 1158 |
+
"rewards/accuracies": 1.0,
|
| 1159 |
+
"rewards/chosen": 1.5441405773162842,
|
| 1160 |
+
"rewards/margins": 20.337499618530273,
|
| 1161 |
+
"rewards/rejected": -18.8125,
|
| 1162 |
+
"step": 710
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.547945205479452,
|
| 1166 |
+
"grad_norm": 0.01990415245870054,
|
| 1167 |
+
"learning_rate": 2.5169204737732656e-07,
|
| 1168 |
+
"logits/chosen": 0.16127929091453552,
|
| 1169 |
+
"logits/rejected": -0.3584960997104645,
|
| 1170 |
+
"logps/chosen": -730.2000122070312,
|
| 1171 |
+
"logps/rejected": -689.7999877929688,
|
| 1172 |
+
"loss": 0.0022,
|
| 1173 |
+
"nll_loss": 1.258203148841858,
|
| 1174 |
+
"rewards/accuracies": 1.0,
|
| 1175 |
+
"rewards/chosen": 2.3626952171325684,
|
| 1176 |
+
"rewards/margins": 21.106250762939453,
|
| 1177 |
+
"rewards/rejected": -18.762500762939453,
|
| 1178 |
+
"step": 720
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.5555555555555556,
|
| 1182 |
+
"grad_norm": 71.95741498413015,
|
| 1183 |
+
"learning_rate": 2.4746192893401013e-07,
|
| 1184 |
+
"logits/chosen": 0.14338378608226776,
|
| 1185 |
+
"logits/rejected": -0.30546873807907104,
|
| 1186 |
+
"logps/chosen": -547.2000122070312,
|
| 1187 |
+
"logps/rejected": -565.5999755859375,
|
| 1188 |
+
"loss": 0.0043,
|
| 1189 |
+
"nll_loss": 1.2351562976837158,
|
| 1190 |
+
"rewards/accuracies": 1.0,
|
| 1191 |
+
"rewards/chosen": 1.099023461341858,
|
| 1192 |
+
"rewards/margins": 19.15625,
|
| 1193 |
+
"rewards/rejected": -18.03125,
|
| 1194 |
+
"step": 730
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.563165905631659,
|
| 1198 |
+
"grad_norm": 0.054312022053468544,
|
| 1199 |
+
"learning_rate": 2.432318104906937e-07,
|
| 1200 |
+
"logits/chosen": 0.12090454250574112,
|
| 1201 |
+
"logits/rejected": -0.4435058534145355,
|
| 1202 |
+
"logps/chosen": -529.5999755859375,
|
| 1203 |
+
"logps/rejected": -671.2000122070312,
|
| 1204 |
+
"loss": 0.0418,
|
| 1205 |
+
"nll_loss": 1.1882812976837158,
|
| 1206 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1207 |
+
"rewards/chosen": 1.5811278820037842,
|
| 1208 |
+
"rewards/margins": 20.137500762939453,
|
| 1209 |
+
"rewards/rejected": -18.5625,
|
| 1210 |
+
"step": 740
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 0.5707762557077626,
|
| 1214 |
+
"grad_norm": 2.3258639545400035,
|
| 1215 |
+
"learning_rate": 2.390016920473773e-07,
|
| 1216 |
+
"logits/chosen": 0.23714599013328552,
|
| 1217 |
+
"logits/rejected": -0.3446899354457855,
|
| 1218 |
+
"logps/chosen": -566.5999755859375,
|
| 1219 |
+
"logps/rejected": -625.5999755859375,
|
| 1220 |
+
"loss": 0.0015,
|
| 1221 |
+
"nll_loss": 1.150390625,
|
| 1222 |
+
"rewards/accuracies": 1.0,
|
| 1223 |
+
"rewards/chosen": 1.6384766101837158,
|
| 1224 |
+
"rewards/margins": 19.231250762939453,
|
| 1225 |
+
"rewards/rejected": -17.59375,
|
| 1226 |
+
"step": 750
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.578386605783866,
|
| 1230 |
+
"grad_norm": 0.01897121637282194,
|
| 1231 |
+
"learning_rate": 2.347715736040609e-07,
|
| 1232 |
+
"logits/chosen": 0.07750244438648224,
|
| 1233 |
+
"logits/rejected": -0.45783692598342896,
|
| 1234 |
+
"logps/chosen": -403.6000061035156,
|
| 1235 |
+
"logps/rejected": -605.7999877929688,
|
| 1236 |
+
"loss": 0.0016,
|
| 1237 |
+
"nll_loss": 1.304296851158142,
|
| 1238 |
+
"rewards/accuracies": 1.0,
|
| 1239 |
+
"rewards/chosen": 0.553759753704071,
|
| 1240 |
+
"rewards/margins": 20.450000762939453,
|
| 1241 |
+
"rewards/rejected": -19.893749237060547,
|
| 1242 |
+
"step": 760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.5859969558599696,
|
| 1246 |
+
"grad_norm": 2.5177771348346765,
|
| 1247 |
+
"learning_rate": 2.305414551607445e-07,
|
| 1248 |
+
"logits/chosen": 0.17706298828125,
|
| 1249 |
+
"logits/rejected": -0.4447265565395355,
|
| 1250 |
+
"logps/chosen": -510.6000061035156,
|
| 1251 |
+
"logps/rejected": -648.2000122070312,
|
| 1252 |
+
"loss": 0.0063,
|
| 1253 |
+
"nll_loss": 1.101171851158142,
|
| 1254 |
+
"rewards/accuracies": 1.0,
|
| 1255 |
+
"rewards/chosen": -0.14189453423023224,
|
| 1256 |
+
"rewards/margins": 19.912500381469727,
|
| 1257 |
+
"rewards/rejected": -20.075000762939453,
|
| 1258 |
+
"step": 770
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.593607305936073,
|
| 1262 |
+
"grad_norm": 0.03363001591054618,
|
| 1263 |
+
"learning_rate": 2.263113367174281e-07,
|
| 1264 |
+
"logits/chosen": 0.22104492783546448,
|
| 1265 |
+
"logits/rejected": -0.47509765625,
|
| 1266 |
+
"logps/chosen": -446.20001220703125,
|
| 1267 |
+
"logps/rejected": -597.5999755859375,
|
| 1268 |
+
"loss": 0.0017,
|
| 1269 |
+
"nll_loss": 1.345312476158142,
|
| 1270 |
+
"rewards/accuracies": 1.0,
|
| 1271 |
+
"rewards/chosen": 1.215917944908142,
|
| 1272 |
+
"rewards/margins": 23.112499237060547,
|
| 1273 |
+
"rewards/rejected": -21.924999237060547,
|
| 1274 |
+
"step": 780
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.6012176560121766,
|
| 1278 |
+
"grad_norm": 71.55734725284898,
|
| 1279 |
+
"learning_rate": 2.2208121827411165e-07,
|
| 1280 |
+
"logits/chosen": 0.17466430366039276,
|
| 1281 |
+
"logits/rejected": -0.45488280057907104,
|
| 1282 |
+
"logps/chosen": -607.5999755859375,
|
| 1283 |
+
"logps/rejected": -631.0,
|
| 1284 |
+
"loss": 0.0063,
|
| 1285 |
+
"nll_loss": 1.263281226158142,
|
| 1286 |
+
"rewards/accuracies": 1.0,
|
| 1287 |
+
"rewards/chosen": 1.437109351158142,
|
| 1288 |
+
"rewards/margins": 22.125,
|
| 1289 |
+
"rewards/rejected": -20.725000381469727,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.60882800608828,
|
| 1294 |
+
"grad_norm": 0.019373194860030203,
|
| 1295 |
+
"learning_rate": 2.1785109983079524e-07,
|
| 1296 |
+
"logits/chosen": 0.10065917670726776,
|
| 1297 |
+
"logits/rejected": -0.382568359375,
|
| 1298 |
+
"logps/chosen": -575.5999755859375,
|
| 1299 |
+
"logps/rejected": -706.0,
|
| 1300 |
+
"loss": 0.0211,
|
| 1301 |
+
"nll_loss": 1.1941406726837158,
|
| 1302 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1303 |
+
"rewards/chosen": 0.6693359613418579,
|
| 1304 |
+
"rewards/margins": 22.462499618530273,
|
| 1305 |
+
"rewards/rejected": -21.774999618530273,
|
| 1306 |
+
"step": 800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.6164383561643836,
|
| 1310 |
+
"grad_norm": 0.26561558287367,
|
| 1311 |
+
"learning_rate": 2.1362098138747884e-07,
|
| 1312 |
+
"logits/chosen": 0.22358398139476776,
|
| 1313 |
+
"logits/rejected": -0.38427734375,
|
| 1314 |
+
"logps/chosen": -551.0,
|
| 1315 |
+
"logps/rejected": -575.4000244140625,
|
| 1316 |
+
"loss": 0.0615,
|
| 1317 |
+
"nll_loss": 1.1300780773162842,
|
| 1318 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1319 |
+
"rewards/chosen": 2.546875,
|
| 1320 |
+
"rewards/margins": 20.100000381469727,
|
| 1321 |
+
"rewards/rejected": -17.568750381469727,
|
| 1322 |
+
"step": 810
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.624048706240487,
|
| 1326 |
+
"grad_norm": 0.006969235140622949,
|
| 1327 |
+
"learning_rate": 2.0939086294416243e-07,
|
| 1328 |
+
"logits/chosen": 0.15556640923023224,
|
| 1329 |
+
"logits/rejected": -0.4107421934604645,
|
| 1330 |
+
"logps/chosen": -549.5999755859375,
|
| 1331 |
+
"logps/rejected": -530.5999755859375,
|
| 1332 |
+
"loss": 0.0031,
|
| 1333 |
+
"nll_loss": 1.3093750476837158,
|
| 1334 |
+
"rewards/accuracies": 1.0,
|
| 1335 |
+
"rewards/chosen": 1.6941406726837158,
|
| 1336 |
+
"rewards/margins": 18.075000762939453,
|
| 1337 |
+
"rewards/rejected": -16.393749237060547,
|
| 1338 |
+
"step": 820
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 0.6316590563165906,
|
| 1342 |
+
"grad_norm": 0.015068256562013079,
|
| 1343 |
+
"learning_rate": 2.0516074450084602e-07,
|
| 1344 |
+
"logits/chosen": 0.14608153700828552,
|
| 1345 |
+
"logits/rejected": -0.5054687261581421,
|
| 1346 |
+
"logps/chosen": -599.0,
|
| 1347 |
+
"logps/rejected": -602.2000122070312,
|
| 1348 |
+
"loss": 0.0013,
|
| 1349 |
+
"nll_loss": 1.1984374523162842,
|
| 1350 |
+
"rewards/accuracies": 1.0,
|
| 1351 |
+
"rewards/chosen": 2.768359422683716,
|
| 1352 |
+
"rewards/margins": 19.625,
|
| 1353 |
+
"rewards/rejected": -16.837499618530273,
|
| 1354 |
+
"step": 830
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.639269406392694,
|
| 1358 |
+
"grad_norm": 0.010671724338771326,
|
| 1359 |
+
"learning_rate": 2.009306260575296e-07,
|
| 1360 |
+
"logits/chosen": 0.18085937201976776,
|
| 1361 |
+
"logits/rejected": -0.3910156190395355,
|
| 1362 |
+
"logps/chosen": -504.0,
|
| 1363 |
+
"logps/rejected": -537.4000244140625,
|
| 1364 |
+
"loss": 0.0185,
|
| 1365 |
+
"nll_loss": 1.1261718273162842,
|
| 1366 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1367 |
+
"rewards/chosen": 2.4957032203674316,
|
| 1368 |
+
"rewards/margins": 19.368749618530273,
|
| 1369 |
+
"rewards/rejected": -16.875,
|
| 1370 |
+
"step": 840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.6468797564687976,
|
| 1374 |
+
"grad_norm": 4.142424859345109,
|
| 1375 |
+
"learning_rate": 1.967005076142132e-07,
|
| 1376 |
+
"logits/chosen": 0.118408203125,
|
| 1377 |
+
"logits/rejected": -0.515429675579071,
|
| 1378 |
+
"logps/chosen": -447.3999938964844,
|
| 1379 |
+
"logps/rejected": -592.2000122070312,
|
| 1380 |
+
"loss": 0.0033,
|
| 1381 |
+
"nll_loss": 1.2218749523162842,
|
| 1382 |
+
"rewards/accuracies": 1.0,
|
| 1383 |
+
"rewards/chosen": 1.3917968273162842,
|
| 1384 |
+
"rewards/margins": 18.462499618530273,
|
| 1385 |
+
"rewards/rejected": -17.0625,
|
| 1386 |
+
"step": 850
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 0.654490106544901,
|
| 1390 |
+
"grad_norm": 6.748885525534139,
|
| 1391 |
+
"learning_rate": 1.9247038917089677e-07,
|
| 1392 |
+
"logits/chosen": 0.13216552138328552,
|
| 1393 |
+
"logits/rejected": -0.39873045682907104,
|
| 1394 |
+
"logps/chosen": -553.4000244140625,
|
| 1395 |
+
"logps/rejected": -589.0,
|
| 1396 |
+
"loss": 0.0028,
|
| 1397 |
+
"nll_loss": 1.221093773841858,
|
| 1398 |
+
"rewards/accuracies": 1.0,
|
| 1399 |
+
"rewards/chosen": 1.2517821788787842,
|
| 1400 |
+
"rewards/margins": 18.393749237060547,
|
| 1401 |
+
"rewards/rejected": -17.174999237060547,
|
| 1402 |
+
"step": 860
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.6621004566210046,
|
| 1406 |
+
"grad_norm": 0.14405173888676007,
|
| 1407 |
+
"learning_rate": 1.8824027072758036e-07,
|
| 1408 |
+
"logits/chosen": 0.11458740383386612,
|
| 1409 |
+
"logits/rejected": -0.35358887910842896,
|
| 1410 |
+
"logps/chosen": -537.5999755859375,
|
| 1411 |
+
"logps/rejected": -524.0,
|
| 1412 |
+
"loss": 0.0036,
|
| 1413 |
+
"nll_loss": 1.1101562976837158,
|
| 1414 |
+
"rewards/accuracies": 1.0,
|
| 1415 |
+
"rewards/chosen": 2.094531297683716,
|
| 1416 |
+
"rewards/margins": 18.700000762939453,
|
| 1417 |
+
"rewards/rejected": -16.587499618530273,
|
| 1418 |
+
"step": 870
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.669710806697108,
|
| 1422 |
+
"grad_norm": 0.023715943386875705,
|
| 1423 |
+
"learning_rate": 1.8401015228426395e-07,
|
| 1424 |
+
"logits/chosen": 0.1767578125,
|
| 1425 |
+
"logits/rejected": -0.4921875,
|
| 1426 |
+
"logps/chosen": -493.3999938964844,
|
| 1427 |
+
"logps/rejected": -587.2000122070312,
|
| 1428 |
+
"loss": 0.0018,
|
| 1429 |
+
"nll_loss": 1.103124976158142,
|
| 1430 |
+
"rewards/accuracies": 1.0,
|
| 1431 |
+
"rewards/chosen": 1.36083984375,
|
| 1432 |
+
"rewards/margins": 20.612499237060547,
|
| 1433 |
+
"rewards/rejected": -19.262500762939453,
|
| 1434 |
+
"step": 880
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.6773211567732116,
|
| 1438 |
+
"grad_norm": 1.7724922766017863,
|
| 1439 |
+
"learning_rate": 1.7978003384094754e-07,
|
| 1440 |
+
"logits/chosen": 0.1239013671875,
|
| 1441 |
+
"logits/rejected": -0.4574218690395355,
|
| 1442 |
+
"logps/chosen": -593.5999755859375,
|
| 1443 |
+
"logps/rejected": -584.2000122070312,
|
| 1444 |
+
"loss": 0.0085,
|
| 1445 |
+
"nll_loss": 1.1789062023162842,
|
| 1446 |
+
"rewards/accuracies": 1.0,
|
| 1447 |
+
"rewards/chosen": 0.84326171875,
|
| 1448 |
+
"rewards/margins": 20.087499618530273,
|
| 1449 |
+
"rewards/rejected": -19.262500762939453,
|
| 1450 |
+
"step": 890
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.684931506849315,
|
| 1454 |
+
"grad_norm": 0.0621351147314125,
|
| 1455 |
+
"learning_rate": 1.755499153976311e-07,
|
| 1456 |
+
"logits/chosen": 0.18120117485523224,
|
| 1457 |
+
"logits/rejected": -0.43937987089157104,
|
| 1458 |
+
"logps/chosen": -639.7999877929688,
|
| 1459 |
+
"logps/rejected": -665.5999755859375,
|
| 1460 |
+
"loss": 0.009,
|
| 1461 |
+
"nll_loss": 1.3125,
|
| 1462 |
+
"rewards/accuracies": 1.0,
|
| 1463 |
+
"rewards/chosen": 1.5368163585662842,
|
| 1464 |
+
"rewards/margins": 21.887500762939453,
|
| 1465 |
+
"rewards/rejected": -20.337499618530273,
|
| 1466 |
+
"step": 900
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.6925418569254186,
|
| 1470 |
+
"grad_norm": 2.0647975965834875,
|
| 1471 |
+
"learning_rate": 1.713197969543147e-07,
|
| 1472 |
+
"logits/chosen": 0.22529296576976776,
|
| 1473 |
+
"logits/rejected": -0.4823242127895355,
|
| 1474 |
+
"logps/chosen": -465.3999938964844,
|
| 1475 |
+
"logps/rejected": -666.4000244140625,
|
| 1476 |
+
"loss": 0.0039,
|
| 1477 |
+
"nll_loss": 0.9964843988418579,
|
| 1478 |
+
"rewards/accuracies": 1.0,
|
| 1479 |
+
"rewards/chosen": 1.2707030773162842,
|
| 1480 |
+
"rewards/margins": 21.8125,
|
| 1481 |
+
"rewards/rejected": -20.537500381469727,
|
| 1482 |
+
"step": 910
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.700152207001522,
|
| 1486 |
+
"grad_norm": 0.01827808812673078,
|
| 1487 |
+
"learning_rate": 1.6708967851099832e-07,
|
| 1488 |
+
"logits/chosen": 0.1510009765625,
|
| 1489 |
+
"logits/rejected": -0.5501953363418579,
|
| 1490 |
+
"logps/chosen": -475.0,
|
| 1491 |
+
"logps/rejected": -580.2000122070312,
|
| 1492 |
+
"loss": 0.0109,
|
| 1493 |
+
"nll_loss": 1.261328101158142,
|
| 1494 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1495 |
+
"rewards/chosen": 2.5765137672424316,
|
| 1496 |
+
"rewards/margins": 20.4375,
|
| 1497 |
+
"rewards/rejected": -17.875,
|
| 1498 |
+
"step": 920
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 0.7077625570776256,
|
| 1502 |
+
"grad_norm": 0.0473030684728447,
|
| 1503 |
+
"learning_rate": 1.6285956006768189e-07,
|
| 1504 |
+
"logits/chosen": 0.10871581733226776,
|
| 1505 |
+
"logits/rejected": -0.545214831829071,
|
| 1506 |
+
"logps/chosen": -624.0,
|
| 1507 |
+
"logps/rejected": -598.2000122070312,
|
| 1508 |
+
"loss": 0.0018,
|
| 1509 |
+
"nll_loss": 1.4874999523162842,
|
| 1510 |
+
"rewards/accuracies": 1.0,
|
| 1511 |
+
"rewards/chosen": 2.554003953933716,
|
| 1512 |
+
"rewards/margins": 21.587499618530273,
|
| 1513 |
+
"rewards/rejected": -19.037500381469727,
|
| 1514 |
+
"step": 930
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.715372907153729,
|
| 1518 |
+
"grad_norm": 0.03084654301861264,
|
| 1519 |
+
"learning_rate": 1.5862944162436548e-07,
|
| 1520 |
+
"logits/chosen": 0.11713866889476776,
|
| 1521 |
+
"logits/rejected": -0.4892578125,
|
| 1522 |
+
"logps/chosen": -564.4000244140625,
|
| 1523 |
+
"logps/rejected": -675.5999755859375,
|
| 1524 |
+
"loss": 0.0014,
|
| 1525 |
+
"nll_loss": 1.1964843273162842,
|
| 1526 |
+
"rewards/accuracies": 1.0,
|
| 1527 |
+
"rewards/chosen": 1.4977538585662842,
|
| 1528 |
+
"rewards/margins": 23.1875,
|
| 1529 |
+
"rewards/rejected": -21.6875,
|
| 1530 |
+
"step": 940
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.7229832572298326,
|
| 1534 |
+
"grad_norm": 0.13129050590623412,
|
| 1535 |
+
"learning_rate": 1.5439932318104907e-07,
|
| 1536 |
+
"logits/chosen": 0.17340087890625,
|
| 1537 |
+
"logits/rejected": -0.4446777403354645,
|
| 1538 |
+
"logps/chosen": -552.5999755859375,
|
| 1539 |
+
"logps/rejected": -611.0,
|
| 1540 |
+
"loss": 0.0019,
|
| 1541 |
+
"nll_loss": 1.2511718273162842,
|
| 1542 |
+
"rewards/accuracies": 1.0,
|
| 1543 |
+
"rewards/chosen": 1.9181640148162842,
|
| 1544 |
+
"rewards/margins": 22.612499237060547,
|
| 1545 |
+
"rewards/rejected": -20.693750381469727,
|
| 1546 |
+
"step": 950
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 0.730593607305936,
|
| 1550 |
+
"grad_norm": 0.12209233572971019,
|
| 1551 |
+
"learning_rate": 1.5016920473773263e-07,
|
| 1552 |
+
"logits/chosen": 0.07487793266773224,
|
| 1553 |
+
"logits/rejected": -0.37958985567092896,
|
| 1554 |
+
"logps/chosen": -553.5999755859375,
|
| 1555 |
+
"logps/rejected": -524.0,
|
| 1556 |
+
"loss": 0.0135,
|
| 1557 |
+
"nll_loss": 1.306640625,
|
| 1558 |
+
"rewards/accuracies": 1.0,
|
| 1559 |
+
"rewards/chosen": 0.573437511920929,
|
| 1560 |
+
"rewards/margins": 18.46875,
|
| 1561 |
+
"rewards/rejected": -17.912500381469727,
|
| 1562 |
+
"step": 960
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.7382039573820396,
|
| 1566 |
+
"grad_norm": 0.01928538493363527,
|
| 1567 |
+
"learning_rate": 1.4593908629441623e-07,
|
| 1568 |
+
"logits/chosen": 0.13522949814796448,
|
| 1569 |
+
"logits/rejected": -0.512011706829071,
|
| 1570 |
+
"logps/chosen": -415.6000061035156,
|
| 1571 |
+
"logps/rejected": -602.4000244140625,
|
| 1572 |
+
"loss": 0.0018,
|
| 1573 |
+
"nll_loss": 1.1726562976837158,
|
| 1574 |
+
"rewards/accuracies": 1.0,
|
| 1575 |
+
"rewards/chosen": 1.518164038658142,
|
| 1576 |
+
"rewards/margins": 21.549999237060547,
|
| 1577 |
+
"rewards/rejected": -20.024999618530273,
|
| 1578 |
+
"step": 970
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.745814307458143,
|
| 1582 |
+
"grad_norm": 0.08309182356914309,
|
| 1583 |
+
"learning_rate": 1.4170896785109982e-07,
|
| 1584 |
+
"logits/chosen": 0.12703856825828552,
|
| 1585 |
+
"logits/rejected": -0.44999998807907104,
|
| 1586 |
+
"logps/chosen": -445.79998779296875,
|
| 1587 |
+
"logps/rejected": -566.5999755859375,
|
| 1588 |
+
"loss": 0.0151,
|
| 1589 |
+
"nll_loss": 1.248437523841858,
|
| 1590 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1591 |
+
"rewards/chosen": 0.9463866949081421,
|
| 1592 |
+
"rewards/margins": 20.5625,
|
| 1593 |
+
"rewards/rejected": -19.649999618530273,
|
| 1594 |
+
"step": 980
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.7534246575342466,
|
| 1598 |
+
"grad_norm": 0.010718640909118582,
|
| 1599 |
+
"learning_rate": 1.374788494077834e-07,
|
| 1600 |
+
"logits/chosen": -0.0050598145462572575,
|
| 1601 |
+
"logits/rejected": -0.566210925579071,
|
| 1602 |
+
"logps/chosen": -565.2000122070312,
|
| 1603 |
+
"logps/rejected": -642.0,
|
| 1604 |
+
"loss": 0.0016,
|
| 1605 |
+
"nll_loss": 1.3722655773162842,
|
| 1606 |
+
"rewards/accuracies": 1.0,
|
| 1607 |
+
"rewards/chosen": 1.545507788658142,
|
| 1608 |
+
"rewards/margins": 23.412500381469727,
|
| 1609 |
+
"rewards/rejected": -21.887500762939453,
|
| 1610 |
+
"step": 990
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.76103500761035,
|
| 1614 |
+
"grad_norm": 1.7875696700411792,
|
| 1615 |
+
"learning_rate": 1.33248730964467e-07,
|
| 1616 |
+
"logits/chosen": 0.15089721977710724,
|
| 1617 |
+
"logits/rejected": -0.573046863079071,
|
| 1618 |
+
"logps/chosen": -597.0,
|
| 1619 |
+
"logps/rejected": -627.7999877929688,
|
| 1620 |
+
"loss": 0.0025,
|
| 1621 |
+
"nll_loss": 1.303125023841858,
|
| 1622 |
+
"rewards/accuracies": 1.0,
|
| 1623 |
+
"rewards/chosen": 2.196484327316284,
|
| 1624 |
+
"rewards/margins": 23.674999237060547,
|
| 1625 |
+
"rewards/rejected": -21.456249237060547,
|
| 1626 |
+
"step": 1000
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.7686453576864536,
|
| 1630 |
+
"grad_norm": 0.28944428189253263,
|
| 1631 |
+
"learning_rate": 1.290186125211506e-07,
|
| 1632 |
+
"logits/chosen": 0.09410400688648224,
|
| 1633 |
+
"logits/rejected": -0.49995118379592896,
|
| 1634 |
+
"logps/chosen": -417.0,
|
| 1635 |
+
"logps/rejected": -561.7999877929688,
|
| 1636 |
+
"loss": 0.0017,
|
| 1637 |
+
"nll_loss": 1.210546851158142,
|
| 1638 |
+
"rewards/accuracies": 1.0,
|
| 1639 |
+
"rewards/chosen": 1.556640625,
|
| 1640 |
+
"rewards/margins": 21.712499618530273,
|
| 1641 |
+
"rewards/rejected": -20.137500762939453,
|
| 1642 |
+
"step": 1010
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.776255707762557,
|
| 1646 |
+
"grad_norm": 0.5677920114993124,
|
| 1647 |
+
"learning_rate": 1.2478849407783419e-07,
|
| 1648 |
+
"logits/chosen": 0.07459716498851776,
|
| 1649 |
+
"logits/rejected": -0.5376952886581421,
|
| 1650 |
+
"logps/chosen": -557.4000244140625,
|
| 1651 |
+
"logps/rejected": -737.5999755859375,
|
| 1652 |
+
"loss": 0.0014,
|
| 1653 |
+
"nll_loss": 1.2921874523162842,
|
| 1654 |
+
"rewards/accuracies": 1.0,
|
| 1655 |
+
"rewards/chosen": 1.4765625,
|
| 1656 |
+
"rewards/margins": 24.337499618530273,
|
| 1657 |
+
"rewards/rejected": -22.850000381469727,
|
| 1658 |
+
"step": 1020
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.7838660578386606,
|
| 1662 |
+
"grad_norm": 0.026811795664542712,
|
| 1663 |
+
"learning_rate": 1.2055837563451775e-07,
|
| 1664 |
+
"logits/chosen": 0.02998046949505806,
|
| 1665 |
+
"logits/rejected": -0.4181152284145355,
|
| 1666 |
+
"logps/chosen": -617.0,
|
| 1667 |
+
"logps/rejected": -614.7999877929688,
|
| 1668 |
+
"loss": 0.0022,
|
| 1669 |
+
"nll_loss": 1.2687499523162842,
|
| 1670 |
+
"rewards/accuracies": 1.0,
|
| 1671 |
+
"rewards/chosen": 1.533593773841858,
|
| 1672 |
+
"rewards/margins": 22.262500762939453,
|
| 1673 |
+
"rewards/rejected": -20.700000762939453,
|
| 1674 |
+
"step": 1030
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.791476407914764,
|
| 1678 |
+
"grad_norm": 0.06837563913443363,
|
| 1679 |
+
"learning_rate": 1.1632825719120134e-07,
|
| 1680 |
+
"logits/chosen": 0.17073974013328552,
|
| 1681 |
+
"logits/rejected": -0.5516601800918579,
|
| 1682 |
+
"logps/chosen": -534.7999877929688,
|
| 1683 |
+
"logps/rejected": -592.2000122070312,
|
| 1684 |
+
"loss": 0.0013,
|
| 1685 |
+
"nll_loss": 1.191796898841858,
|
| 1686 |
+
"rewards/accuracies": 1.0,
|
| 1687 |
+
"rewards/chosen": 1.787109375,
|
| 1688 |
+
"rewards/margins": 22.862499237060547,
|
| 1689 |
+
"rewards/rejected": -21.087499618530273,
|
| 1690 |
+
"step": 1040
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.7990867579908676,
|
| 1694 |
+
"grad_norm": 0.028405006733601978,
|
| 1695 |
+
"learning_rate": 1.1209813874788493e-07,
|
| 1696 |
+
"logits/chosen": 0.07956542819738388,
|
| 1697 |
+
"logits/rejected": -0.6153320074081421,
|
| 1698 |
+
"logps/chosen": -433.6000061035156,
|
| 1699 |
+
"logps/rejected": -694.7999877929688,
|
| 1700 |
+
"loss": 0.0278,
|
| 1701 |
+
"nll_loss": 1.44921875,
|
| 1702 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1703 |
+
"rewards/chosen": 1.0939452648162842,
|
| 1704 |
+
"rewards/margins": 23.524999618530273,
|
| 1705 |
+
"rewards/rejected": -22.4375,
|
| 1706 |
+
"step": 1050
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.806697108066971,
|
| 1710 |
+
"grad_norm": 0.04442154053079314,
|
| 1711 |
+
"learning_rate": 1.0786802030456853e-07,
|
| 1712 |
+
"logits/chosen": 0.11202392727136612,
|
| 1713 |
+
"logits/rejected": -0.4344238340854645,
|
| 1714 |
+
"logps/chosen": -552.7999877929688,
|
| 1715 |
+
"logps/rejected": -608.5999755859375,
|
| 1716 |
+
"loss": 0.0012,
|
| 1717 |
+
"nll_loss": 1.119140625,
|
| 1718 |
+
"rewards/accuracies": 1.0,
|
| 1719 |
+
"rewards/chosen": 1.293554663658142,
|
| 1720 |
+
"rewards/margins": 22.506250381469727,
|
| 1721 |
+
"rewards/rejected": -21.1875,
|
| 1722 |
+
"step": 1060
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.8143074581430746,
|
| 1726 |
+
"grad_norm": 0.05791145700448629,
|
| 1727 |
+
"learning_rate": 1.036379018612521e-07,
|
| 1728 |
+
"logits/chosen": 0.05427246168255806,
|
| 1729 |
+
"logits/rejected": -0.3960937559604645,
|
| 1730 |
+
"logps/chosen": -592.5999755859375,
|
| 1731 |
+
"logps/rejected": -604.5999755859375,
|
| 1732 |
+
"loss": 0.0015,
|
| 1733 |
+
"nll_loss": 1.204687476158142,
|
| 1734 |
+
"rewards/accuracies": 1.0,
|
| 1735 |
+
"rewards/chosen": 1.5237305164337158,
|
| 1736 |
+
"rewards/margins": 22.375,
|
| 1737 |
+
"rewards/rejected": -20.837499618530273,
|
| 1738 |
+
"step": 1070
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.821917808219178,
|
| 1742 |
+
"grad_norm": 0.060214111600052515,
|
| 1743 |
+
"learning_rate": 9.94077834179357e-08,
|
| 1744 |
+
"logits/chosen": 0.03056640550494194,
|
| 1745 |
+
"logits/rejected": -0.5289062261581421,
|
| 1746 |
+
"logps/chosen": -603.2000122070312,
|
| 1747 |
+
"logps/rejected": -631.0,
|
| 1748 |
+
"loss": 0.0015,
|
| 1749 |
+
"nll_loss": 1.353515625,
|
| 1750 |
+
"rewards/accuracies": 1.0,
|
| 1751 |
+
"rewards/chosen": 1.174218773841858,
|
| 1752 |
+
"rewards/margins": 22.125,
|
| 1753 |
+
"rewards/rejected": -20.962499618530273,
|
| 1754 |
+
"step": 1080
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.8295281582952816,
|
| 1758 |
+
"grad_norm": 3.7603786344107384,
|
| 1759 |
+
"learning_rate": 9.517766497461928e-08,
|
| 1760 |
+
"logits/chosen": 0.09332275390625,
|
| 1761 |
+
"logits/rejected": -0.64453125,
|
| 1762 |
+
"logps/chosen": -379.29998779296875,
|
| 1763 |
+
"logps/rejected": -650.0,
|
| 1764 |
+
"loss": 0.0019,
|
| 1765 |
+
"nll_loss": 1.244140625,
|
| 1766 |
+
"rewards/accuracies": 1.0,
|
| 1767 |
+
"rewards/chosen": 1.859375,
|
| 1768 |
+
"rewards/margins": 25.350000381469727,
|
| 1769 |
+
"rewards/rejected": -23.462499618530273,
|
| 1770 |
+
"step": 1090
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.837138508371385,
|
| 1774 |
+
"grad_norm": 0.01433450651830501,
|
| 1775 |
+
"learning_rate": 9.094754653130288e-08,
|
| 1776 |
+
"logits/chosen": 0.10407714545726776,
|
| 1777 |
+
"logits/rejected": -0.525390625,
|
| 1778 |
+
"logps/chosen": -522.2000122070312,
|
| 1779 |
+
"logps/rejected": -695.5999755859375,
|
| 1780 |
+
"loss": 0.0019,
|
| 1781 |
+
"nll_loss": 1.1882812976837158,
|
| 1782 |
+
"rewards/accuracies": 1.0,
|
| 1783 |
+
"rewards/chosen": 1.301171898841858,
|
| 1784 |
+
"rewards/margins": 23.399999618530273,
|
| 1785 |
+
"rewards/rejected": -22.112499237060547,
|
| 1786 |
+
"step": 1100
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.8447488584474886,
|
| 1790 |
+
"grad_norm": 0.12337194483982274,
|
| 1791 |
+
"learning_rate": 8.671742808798646e-08,
|
| 1792 |
+
"logits/chosen": -0.0038330077659338713,
|
| 1793 |
+
"logits/rejected": -0.553906261920929,
|
| 1794 |
+
"logps/chosen": -665.7999877929688,
|
| 1795 |
+
"logps/rejected": -687.5999755859375,
|
| 1796 |
+
"loss": 0.006,
|
| 1797 |
+
"nll_loss": 1.2785155773162842,
|
| 1798 |
+
"rewards/accuracies": 1.0,
|
| 1799 |
+
"rewards/chosen": 1.5480468273162842,
|
| 1800 |
+
"rewards/margins": 23.75,
|
| 1801 |
+
"rewards/rejected": -22.212499618530273,
|
| 1802 |
+
"step": 1110
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.852359208523592,
|
| 1806 |
+
"grad_norm": 0.03365097008505571,
|
| 1807 |
+
"learning_rate": 8.248730964467004e-08,
|
| 1808 |
+
"logits/chosen": 0.20329590141773224,
|
| 1809 |
+
"logits/rejected": -0.568554699420929,
|
| 1810 |
+
"logps/chosen": -417.8999938964844,
|
| 1811 |
+
"logps/rejected": -588.2000122070312,
|
| 1812 |
+
"loss": 0.0076,
|
| 1813 |
+
"nll_loss": 1.183984398841858,
|
| 1814 |
+
"rewards/accuracies": 1.0,
|
| 1815 |
+
"rewards/chosen": 2.1611328125,
|
| 1816 |
+
"rewards/margins": 22.981250762939453,
|
| 1817 |
+
"rewards/rejected": -20.806249618530273,
|
| 1818 |
+
"step": 1120
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.8599695585996956,
|
| 1822 |
+
"grad_norm": 0.02467706843751577,
|
| 1823 |
+
"learning_rate": 7.825719120135363e-08,
|
| 1824 |
+
"logits/chosen": 0.10390625149011612,
|
| 1825 |
+
"logits/rejected": -0.537402331829071,
|
| 1826 |
+
"logps/chosen": -503.6000061035156,
|
| 1827 |
+
"logps/rejected": -546.4000244140625,
|
| 1828 |
+
"loss": 0.0013,
|
| 1829 |
+
"nll_loss": 1.300390601158142,
|
| 1830 |
+
"rewards/accuracies": 1.0,
|
| 1831 |
+
"rewards/chosen": 3.0023436546325684,
|
| 1832 |
+
"rewards/margins": 21.649999618530273,
|
| 1833 |
+
"rewards/rejected": -18.681249618530273,
|
| 1834 |
+
"step": 1130
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.867579908675799,
|
| 1838 |
+
"grad_norm": 0.05802796855087668,
|
| 1839 |
+
"learning_rate": 7.402707275803722e-08,
|
| 1840 |
+
"logits/chosen": 0.17983397841453552,
|
| 1841 |
+
"logits/rejected": -0.38789063692092896,
|
| 1842 |
+
"logps/chosen": -527.0,
|
| 1843 |
+
"logps/rejected": -623.7999877929688,
|
| 1844 |
+
"loss": 0.0012,
|
| 1845 |
+
"nll_loss": 1.175390601158142,
|
| 1846 |
+
"rewards/accuracies": 1.0,
|
| 1847 |
+
"rewards/chosen": 2.4507813453674316,
|
| 1848 |
+
"rewards/margins": 24.012500762939453,
|
| 1849 |
+
"rewards/rejected": -21.587499618530273,
|
| 1850 |
+
"step": 1140
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.8751902587519026,
|
| 1854 |
+
"grad_norm": 0.09053688426526861,
|
| 1855 |
+
"learning_rate": 6.979695431472081e-08,
|
| 1856 |
+
"logits/chosen": 0.015576171688735485,
|
| 1857 |
+
"logits/rejected": -0.5982421636581421,
|
| 1858 |
+
"logps/chosen": -693.4000244140625,
|
| 1859 |
+
"logps/rejected": -703.2000122070312,
|
| 1860 |
+
"loss": 0.0014,
|
| 1861 |
+
"nll_loss": 1.3367187976837158,
|
| 1862 |
+
"rewards/accuracies": 1.0,
|
| 1863 |
+
"rewards/chosen": 0.52734375,
|
| 1864 |
+
"rewards/margins": 22.649999618530273,
|
| 1865 |
+
"rewards/rejected": -22.125,
|
| 1866 |
+
"step": 1150
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.882800608828006,
|
| 1870 |
+
"grad_norm": 0.01860993955675537,
|
| 1871 |
+
"learning_rate": 6.556683587140439e-08,
|
| 1872 |
+
"logits/chosen": 0.06380615383386612,
|
| 1873 |
+
"logits/rejected": -0.5582031011581421,
|
| 1874 |
+
"logps/chosen": -528.0,
|
| 1875 |
+
"logps/rejected": -633.4000244140625,
|
| 1876 |
+
"loss": 0.0117,
|
| 1877 |
+
"nll_loss": 1.3117187023162842,
|
| 1878 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1879 |
+
"rewards/chosen": 0.88134765625,
|
| 1880 |
+
"rewards/margins": 21.0625,
|
| 1881 |
+
"rewards/rejected": -20.162500381469727,
|
| 1882 |
+
"step": 1160
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.8904109589041096,
|
| 1886 |
+
"grad_norm": 4.598194993393434,
|
| 1887 |
+
"learning_rate": 6.133671742808798e-08,
|
| 1888 |
+
"logits/chosen": 0.132568359375,
|
| 1889 |
+
"logits/rejected": -0.49199217557907104,
|
| 1890 |
+
"logps/chosen": -521.4000244140625,
|
| 1891 |
+
"logps/rejected": -592.2000122070312,
|
| 1892 |
+
"loss": 0.0034,
|
| 1893 |
+
"nll_loss": 1.120703101158142,
|
| 1894 |
+
"rewards/accuracies": 1.0,
|
| 1895 |
+
"rewards/chosen": 1.017578125,
|
| 1896 |
+
"rewards/margins": 22.506250381469727,
|
| 1897 |
+
"rewards/rejected": -21.475000381469727,
|
| 1898 |
+
"step": 1170
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.898021308980213,
|
| 1902 |
+
"grad_norm": 0.022759871757378192,
|
| 1903 |
+
"learning_rate": 5.710659898477157e-08,
|
| 1904 |
+
"logits/chosen": -0.04151611402630806,
|
| 1905 |
+
"logits/rejected": -0.538281261920929,
|
| 1906 |
+
"logps/chosen": -616.2999877929688,
|
| 1907 |
+
"logps/rejected": -637.7999877929688,
|
| 1908 |
+
"loss": 0.0015,
|
| 1909 |
+
"nll_loss": 1.326171875,
|
| 1910 |
+
"rewards/accuracies": 1.0,
|
| 1911 |
+
"rewards/chosen": 1.3427734375,
|
| 1912 |
+
"rewards/margins": 21.59375,
|
| 1913 |
+
"rewards/rejected": -20.212499618530273,
|
| 1914 |
+
"step": 1180
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.9056316590563166,
|
| 1918 |
+
"grad_norm": 0.01135329070278885,
|
| 1919 |
+
"learning_rate": 5.287648054145516e-08,
|
| 1920 |
+
"logits/chosen": 0.1397705078125,
|
| 1921 |
+
"logits/rejected": -0.5125976800918579,
|
| 1922 |
+
"logps/chosen": -493.20001220703125,
|
| 1923 |
+
"logps/rejected": -665.4000244140625,
|
| 1924 |
+
"loss": 0.0108,
|
| 1925 |
+
"nll_loss": 1.087890625,
|
| 1926 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1927 |
+
"rewards/chosen": 2.242382764816284,
|
| 1928 |
+
"rewards/margins": 22.5625,
|
| 1929 |
+
"rewards/rejected": -20.350000381469727,
|
| 1930 |
+
"step": 1190
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.91324200913242,
|
| 1934 |
+
"grad_norm": 0.015766670753422946,
|
| 1935 |
+
"learning_rate": 4.8646362098138747e-08,
|
| 1936 |
+
"logits/chosen": 0.14870604872703552,
|
| 1937 |
+
"logits/rejected": -0.54931640625,
|
| 1938 |
+
"logps/chosen": -416.79998779296875,
|
| 1939 |
+
"logps/rejected": -672.4000244140625,
|
| 1940 |
+
"loss": 0.0014,
|
| 1941 |
+
"nll_loss": 1.298828125,
|
| 1942 |
+
"rewards/accuracies": 1.0,
|
| 1943 |
+
"rewards/chosen": 1.400781273841858,
|
| 1944 |
+
"rewards/margins": 22.618749618530273,
|
| 1945 |
+
"rewards/rejected": -21.243749618530273,
|
| 1946 |
+
"step": 1200
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.9208523592085236,
|
| 1950 |
+
"grad_norm": 0.17525846513779506,
|
| 1951 |
+
"learning_rate": 4.441624365482233e-08,
|
| 1952 |
+
"logits/chosen": 0.14744873344898224,
|
| 1953 |
+
"logits/rejected": -0.523632824420929,
|
| 1954 |
+
"logps/chosen": -414.0,
|
| 1955 |
+
"logps/rejected": -531.7999877929688,
|
| 1956 |
+
"loss": 0.0015,
|
| 1957 |
+
"nll_loss": 1.117578148841858,
|
| 1958 |
+
"rewards/accuracies": 1.0,
|
| 1959 |
+
"rewards/chosen": 1.55078125,
|
| 1960 |
+
"rewards/margins": 20.600000381469727,
|
| 1961 |
+
"rewards/rejected": -19.037500381469727,
|
| 1962 |
+
"step": 1210
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.928462709284627,
|
| 1966 |
+
"grad_norm": 1.6497974711537087,
|
| 1967 |
+
"learning_rate": 4.018612521150592e-08,
|
| 1968 |
+
"logits/chosen": 0.07460937649011612,
|
| 1969 |
+
"logits/rejected": -0.5708984136581421,
|
| 1970 |
+
"logps/chosen": -445.79998779296875,
|
| 1971 |
+
"logps/rejected": -633.2000122070312,
|
| 1972 |
+
"loss": 0.0031,
|
| 1973 |
+
"nll_loss": 1.2960937023162842,
|
| 1974 |
+
"rewards/accuracies": 1.0,
|
| 1975 |
+
"rewards/chosen": 1.904296875,
|
| 1976 |
+
"rewards/margins": 23.487499237060547,
|
| 1977 |
+
"rewards/rejected": -21.575000762939453,
|
| 1978 |
+
"step": 1220
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.9360730593607306,
|
| 1982 |
+
"grad_norm": 0.01936074104953005,
|
| 1983 |
+
"learning_rate": 3.595600676818951e-08,
|
| 1984 |
+
"logits/chosen": 0.18209227919578552,
|
| 1985 |
+
"logits/rejected": -0.5876709222793579,
|
| 1986 |
+
"logps/chosen": -497.3999938964844,
|
| 1987 |
+
"logps/rejected": -682.4000244140625,
|
| 1988 |
+
"loss": 0.0012,
|
| 1989 |
+
"nll_loss": 1.1589844226837158,
|
| 1990 |
+
"rewards/accuracies": 1.0,
|
| 1991 |
+
"rewards/chosen": 1.68505859375,
|
| 1992 |
+
"rewards/margins": 24.0,
|
| 1993 |
+
"rewards/rejected": -22.299999237060547,
|
| 1994 |
+
"step": 1230
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.943683409436834,
|
| 1998 |
+
"grad_norm": 0.02913464666159855,
|
| 1999 |
+
"learning_rate": 3.1725888324873094e-08,
|
| 2000 |
+
"logits/chosen": 0.13217774033546448,
|
| 2001 |
+
"logits/rejected": -0.531054675579071,
|
| 2002 |
+
"logps/chosen": -554.0,
|
| 2003 |
+
"logps/rejected": -580.2000122070312,
|
| 2004 |
+
"loss": 0.0023,
|
| 2005 |
+
"nll_loss": 1.1726562976837158,
|
| 2006 |
+
"rewards/accuracies": 1.0,
|
| 2007 |
+
"rewards/chosen": 2.253124952316284,
|
| 2008 |
+
"rewards/margins": 22.337499618530273,
|
| 2009 |
+
"rewards/rejected": -20.0625,
|
| 2010 |
+
"step": 1240
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 0.9512937595129376,
|
| 2014 |
+
"grad_norm": 64.54202193399442,
|
| 2015 |
+
"learning_rate": 2.7495769881556683e-08,
|
| 2016 |
+
"logits/chosen": 0.06240234524011612,
|
| 2017 |
+
"logits/rejected": -0.5318359136581421,
|
| 2018 |
+
"logps/chosen": -558.2000122070312,
|
| 2019 |
+
"logps/rejected": -560.5999755859375,
|
| 2020 |
+
"loss": 0.0084,
|
| 2021 |
+
"nll_loss": 1.2644531726837158,
|
| 2022 |
+
"rewards/accuracies": 1.0,
|
| 2023 |
+
"rewards/chosen": 2.30078125,
|
| 2024 |
+
"rewards/margins": 22.625,
|
| 2025 |
+
"rewards/rejected": -20.325000762939453,
|
| 2026 |
+
"step": 1250
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.958904109589041,
|
| 2030 |
+
"grad_norm": 0.05824637331901012,
|
| 2031 |
+
"learning_rate": 2.326565143824027e-08,
|
| 2032 |
+
"logits/chosen": 0.03659667819738388,
|
| 2033 |
+
"logits/rejected": -0.5044921636581421,
|
| 2034 |
+
"logps/chosen": -543.4000244140625,
|
| 2035 |
+
"logps/rejected": -586.0,
|
| 2036 |
+
"loss": 0.0014,
|
| 2037 |
+
"nll_loss": 1.3367187976837158,
|
| 2038 |
+
"rewards/accuracies": 1.0,
|
| 2039 |
+
"rewards/chosen": 1.899999976158142,
|
| 2040 |
+
"rewards/margins": 22.299999237060547,
|
| 2041 |
+
"rewards/rejected": -20.412500381469727,
|
| 2042 |
+
"step": 1260
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.9665144596651446,
|
| 2046 |
+
"grad_norm": 4.3621795274936375,
|
| 2047 |
+
"learning_rate": 1.903553299492386e-08,
|
| 2048 |
+
"logits/chosen": 0.269287109375,
|
| 2049 |
+
"logits/rejected": -0.5249999761581421,
|
| 2050 |
+
"logps/chosen": -365.0,
|
| 2051 |
+
"logps/rejected": -547.5999755859375,
|
| 2052 |
+
"loss": 0.0021,
|
| 2053 |
+
"nll_loss": 1.131250023841858,
|
| 2054 |
+
"rewards/accuracies": 1.0,
|
| 2055 |
+
"rewards/chosen": 2.3441405296325684,
|
| 2056 |
+
"rewards/margins": 22.424999237060547,
|
| 2057 |
+
"rewards/rejected": -20.087499618530273,
|
| 2058 |
+
"step": 1270
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.974124809741248,
|
| 2062 |
+
"grad_norm": 3.450773236527342,
|
| 2063 |
+
"learning_rate": 1.4805414551607443e-08,
|
| 2064 |
+
"logits/chosen": 0.15256042778491974,
|
| 2065 |
+
"logits/rejected": -0.47431641817092896,
|
| 2066 |
+
"logps/chosen": -528.0,
|
| 2067 |
+
"logps/rejected": -670.0,
|
| 2068 |
+
"loss": 0.0014,
|
| 2069 |
+
"nll_loss": 1.1640625,
|
| 2070 |
+
"rewards/accuracies": 1.0,
|
| 2071 |
+
"rewards/chosen": 2.569140672683716,
|
| 2072 |
+
"rewards/margins": 23.262500762939453,
|
| 2073 |
+
"rewards/rejected": -20.71875,
|
| 2074 |
+
"step": 1280
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.9817351598173516,
|
| 2078 |
+
"grad_norm": 1.5387980781657402,
|
| 2079 |
+
"learning_rate": 1.0575296108291032e-08,
|
| 2080 |
+
"logits/chosen": 0.06939087063074112,
|
| 2081 |
+
"logits/rejected": -0.6625000238418579,
|
| 2082 |
+
"logps/chosen": -564.7999877929688,
|
| 2083 |
+
"logps/rejected": -745.7999877929688,
|
| 2084 |
+
"loss": 0.0094,
|
| 2085 |
+
"nll_loss": 1.283203125,
|
| 2086 |
+
"rewards/accuracies": 1.0,
|
| 2087 |
+
"rewards/chosen": 1.0632812976837158,
|
| 2088 |
+
"rewards/margins": 24.325000762939453,
|
| 2089 |
+
"rewards/rejected": -23.262500762939453,
|
| 2090 |
+
"step": 1290
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.989345509893455,
|
| 2094 |
+
"grad_norm": 0.014999223766331345,
|
| 2095 |
+
"learning_rate": 6.345177664974619e-09,
|
| 2096 |
+
"logits/chosen": 0.20089110732078552,
|
| 2097 |
+
"logits/rejected": -0.40087890625,
|
| 2098 |
+
"logps/chosen": -491.79998779296875,
|
| 2099 |
+
"logps/rejected": -533.2000122070312,
|
| 2100 |
+
"loss": 0.0014,
|
| 2101 |
+
"nll_loss": 1.0988280773162842,
|
| 2102 |
+
"rewards/accuracies": 1.0,
|
| 2103 |
+
"rewards/chosen": 2.5972657203674316,
|
| 2104 |
+
"rewards/margins": 20.149999618530273,
|
| 2105 |
+
"rewards/rejected": -17.556249618530273,
|
| 2106 |
+
"step": 1300
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 0.9969558599695586,
|
| 2110 |
+
"grad_norm": 0.03237073978467657,
|
| 2111 |
+
"learning_rate": 2.115059221658206e-09,
|
| 2112 |
+
"logits/chosen": 0.11948852241039276,
|
| 2113 |
+
"logits/rejected": -0.51483154296875,
|
| 2114 |
+
"logps/chosen": -498.20001220703125,
|
| 2115 |
+
"logps/rejected": -538.0,
|
| 2116 |
+
"loss": 0.002,
|
| 2117 |
+
"nll_loss": 1.239843726158142,
|
| 2118 |
+
"rewards/accuracies": 1.0,
|
| 2119 |
+
"rewards/chosen": 2.21844482421875,
|
| 2120 |
+
"rewards/margins": 19.618749618530273,
|
| 2121 |
+
"rewards/rejected": -17.393749237060547,
|
| 2122 |
+
"step": 1310
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.0,
|
| 2126 |
+
"eval_logits/chosen": 0.16595585644245148,
|
| 2127 |
+
"eval_logits/rejected": -0.46630859375,
|
| 2128 |
+
"eval_logps/chosen": -414.4166564941406,
|
| 2129 |
+
"eval_logps/rejected": -455.5,
|
| 2130 |
+
"eval_loss": 0.002019643783569336,
|
| 2131 |
+
"eval_nll_loss": 1.1422525644302368,
|
| 2132 |
+
"eval_rewards/accuracies": 1.0,
|
| 2133 |
+
"eval_rewards/chosen": 3.3219401836395264,
|
| 2134 |
+
"eval_rewards/margins": 19.63020896911621,
|
| 2135 |
+
"eval_rewards/rejected": -16.296875,
|
| 2136 |
+
"eval_runtime": 7.7598,
|
| 2137 |
+
"eval_samples_per_second": 12.371,
|
| 2138 |
+
"eval_steps_per_second": 1.546,
|
| 2139 |
+
"step": 1314
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 1.0,
|
| 2143 |
+
"step": 1314,
|
| 2144 |
+
"total_flos": 0.0,
|
| 2145 |
+
"train_loss": 0.052201553022480446,
|
| 2146 |
+
"train_runtime": 2376.2971,
|
| 2147 |
+
"train_samples_per_second": 4.423,
|
| 2148 |
+
"train_steps_per_second": 0.553
|
| 2149 |
+
}
|
| 2150 |
+
],
|
| 2151 |
+
"logging_steps": 10,
|
| 2152 |
+
"max_steps": 1314,
|
| 2153 |
+
"num_input_tokens_seen": 0,
|
| 2154 |
+
"num_train_epochs": 1,
|
| 2155 |
+
"save_steps": 500,
|
| 2156 |
+
"stateful_callbacks": {
|
| 2157 |
+
"TrainerControl": {
|
| 2158 |
+
"args": {
|
| 2159 |
+
"should_epoch_stop": false,
|
| 2160 |
+
"should_evaluate": false,
|
| 2161 |
+
"should_log": false,
|
| 2162 |
+
"should_save": true,
|
| 2163 |
+
"should_training_stop": true
|
| 2164 |
+
},
|
| 2165 |
+
"attributes": {}
|
| 2166 |
+
}
|
| 2167 |
+
},
|
| 2168 |
+
"total_flos": 0.0,
|
| 2169 |
+
"train_batch_size": 2,
|
| 2170 |
+
"trial_name": null,
|
| 2171 |
+
"trial_params": null
|
| 2172 |
+
}
|