Model save
Browse files- all_results.json +9 -0
- generation_config.json +14 -0
- train_results.json +9 -0
- trainer_state.json +2156 -0
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.4647196615989924,
|
| 5 |
+
"train_runtime": 4672.0978,
|
| 6 |
+
"train_samples": 10466,
|
| 7 |
+
"train_samples_per_second": 2.24,
|
| 8 |
+
"train_steps_per_second": 0.28
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.05,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.53.3"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.4647196615989924,
|
| 5 |
+
"train_runtime": 4672.0978,
|
| 6 |
+
"train_samples": 10466,
|
| 7 |
+
"train_samples_per_second": 2.24,
|
| 8 |
+
"train_steps_per_second": 0.28
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2156 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": 1309,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0007639419404125286,
|
| 14 |
+
"grad_norm": 135.31482947366968,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"logits/chosen": -0.1904296875,
|
| 17 |
+
"logits/rejected": -0.51171875,
|
| 18 |
+
"logps/chosen": -462.0,
|
| 19 |
+
"logps/rejected": -376.0,
|
| 20 |
+
"loss": 0.6914,
|
| 21 |
+
"nll_loss": 1.1796875,
|
| 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": 0.007639419404125287,
|
| 30 |
+
"grad_norm": 134.88438652307744,
|
| 31 |
+
"learning_rate": 3.435114503816794e-08,
|
| 32 |
+
"logits/chosen": -0.113525390625,
|
| 33 |
+
"logits/rejected": -0.2179904580116272,
|
| 34 |
+
"logps/chosen": -471.1111145019531,
|
| 35 |
+
"logps/rejected": -452.6666564941406,
|
| 36 |
+
"loss": 0.6802,
|
| 37 |
+
"nll_loss": 0.9861111044883728,
|
| 38 |
+
"rewards/accuracies": 0.2916666567325592,
|
| 39 |
+
"rewards/chosen": 0.015292697586119175,
|
| 40 |
+
"rewards/margins": 0.0403103306889534,
|
| 41 |
+
"rewards/rejected": -0.0250244140625,
|
| 42 |
+
"step": 10
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.015278838808250574,
|
| 46 |
+
"grad_norm": 117.40180735452381,
|
| 47 |
+
"learning_rate": 7.251908396946564e-08,
|
| 48 |
+
"logits/chosen": -0.269775390625,
|
| 49 |
+
"logits/rejected": -0.26899415254592896,
|
| 50 |
+
"logps/chosen": -482.3999938964844,
|
| 51 |
+
"logps/rejected": -484.79998779296875,
|
| 52 |
+
"loss": 0.6959,
|
| 53 |
+
"nll_loss": 0.942187488079071,
|
| 54 |
+
"rewards/accuracies": 0.26249998807907104,
|
| 55 |
+
"rewards/chosen": -0.020050048828125,
|
| 56 |
+
"rewards/margins": -0.0012695312034338713,
|
| 57 |
+
"rewards/rejected": -0.01875000074505806,
|
| 58 |
+
"step": 20
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.02291825821237586,
|
| 62 |
+
"grad_norm": 162.31090600401154,
|
| 63 |
+
"learning_rate": 1.1068702290076336e-07,
|
| 64 |
+
"logits/chosen": -0.33186036348342896,
|
| 65 |
+
"logits/rejected": -0.19320067763328552,
|
| 66 |
+
"logps/chosen": -588.7999877929688,
|
| 67 |
+
"logps/rejected": -558.7999877929688,
|
| 68 |
+
"loss": 0.6941,
|
| 69 |
+
"nll_loss": 0.955859363079071,
|
| 70 |
+
"rewards/accuracies": 0.26249998807907104,
|
| 71 |
+
"rewards/chosen": 0.01497802697122097,
|
| 72 |
+
"rewards/margins": 0.010040283203125,
|
| 73 |
+
"rewards/rejected": 0.0049682618118822575,
|
| 74 |
+
"step": 30
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.030557677616501147,
|
| 78 |
+
"grad_norm": 128.81685925882783,
|
| 79 |
+
"learning_rate": 1.4885496183206107e-07,
|
| 80 |
+
"logits/chosen": -0.2830566465854645,
|
| 81 |
+
"logits/rejected": -0.288330078125,
|
| 82 |
+
"logps/chosen": -446.20001220703125,
|
| 83 |
+
"logps/rejected": -442.20001220703125,
|
| 84 |
+
"loss": 0.674,
|
| 85 |
+
"nll_loss": 1.08984375,
|
| 86 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 87 |
+
"rewards/chosen": 0.03880615159869194,
|
| 88 |
+
"rewards/margins": 0.04369506984949112,
|
| 89 |
+
"rewards/rejected": -0.0050048828125,
|
| 90 |
+
"step": 40
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.03819709702062643,
|
| 94 |
+
"grad_norm": 124.70939134917927,
|
| 95 |
+
"learning_rate": 1.8702290076335877e-07,
|
| 96 |
+
"logits/chosen": -0.12124023586511612,
|
| 97 |
+
"logits/rejected": -0.07585449516773224,
|
| 98 |
+
"logps/chosen": -501.6000061035156,
|
| 99 |
+
"logps/rejected": -485.0,
|
| 100 |
+
"loss": 0.6767,
|
| 101 |
+
"nll_loss": 0.9253906011581421,
|
| 102 |
+
"rewards/accuracies": 0.375,
|
| 103 |
+
"rewards/chosen": 0.07005615532398224,
|
| 104 |
+
"rewards/margins": 0.04642333835363388,
|
| 105 |
+
"rewards/rejected": 0.02379760704934597,
|
| 106 |
+
"step": 50
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.04583651642475172,
|
| 110 |
+
"grad_norm": 226.10945049239828,
|
| 111 |
+
"learning_rate": 2.2519083969465648e-07,
|
| 112 |
+
"logits/chosen": -0.13395996391773224,
|
| 113 |
+
"logits/rejected": -0.1141357421875,
|
| 114 |
+
"logps/chosen": -608.2000122070312,
|
| 115 |
+
"logps/rejected": -578.0,
|
| 116 |
+
"loss": 0.6779,
|
| 117 |
+
"nll_loss": 1.0046875476837158,
|
| 118 |
+
"rewards/accuracies": 0.4375,
|
| 119 |
+
"rewards/chosen": 0.11510620266199112,
|
| 120 |
+
"rewards/margins": 0.067718505859375,
|
| 121 |
+
"rewards/rejected": 0.04757080227136612,
|
| 122 |
+
"step": 60
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.053475935828877004,
|
| 126 |
+
"grad_norm": 125.83102595124332,
|
| 127 |
+
"learning_rate": 2.633587786259542e-07,
|
| 128 |
+
"logits/chosen": -0.14421387016773224,
|
| 129 |
+
"logits/rejected": -0.2423095703125,
|
| 130 |
+
"logps/chosen": -514.5999755859375,
|
| 131 |
+
"logps/rejected": -461.20001220703125,
|
| 132 |
+
"loss": 0.6211,
|
| 133 |
+
"nll_loss": 0.9957031011581421,
|
| 134 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 135 |
+
"rewards/chosen": 0.33037108182907104,
|
| 136 |
+
"rewards/margins": 0.19284668564796448,
|
| 137 |
+
"rewards/rejected": 0.13779297471046448,
|
| 138 |
+
"step": 70
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.061115355233002294,
|
| 142 |
+
"grad_norm": 102.17014243801489,
|
| 143 |
+
"learning_rate": 3.015267175572519e-07,
|
| 144 |
+
"logits/chosen": -0.0196533203125,
|
| 145 |
+
"logits/rejected": -0.01951904222369194,
|
| 146 |
+
"logps/chosen": -467.79998779296875,
|
| 147 |
+
"logps/rejected": -433.6000061035156,
|
| 148 |
+
"loss": 0.5739,
|
| 149 |
+
"nll_loss": 0.918749988079071,
|
| 150 |
+
"rewards/accuracies": 0.637499988079071,
|
| 151 |
+
"rewards/chosen": 0.32421875,
|
| 152 |
+
"rewards/margins": 0.3267578184604645,
|
| 153 |
+
"rewards/rejected": -0.0025634765625,
|
| 154 |
+
"step": 80
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.06875477463712758,
|
| 158 |
+
"grad_norm": 116.393294226921,
|
| 159 |
+
"learning_rate": 3.396946564885496e-07,
|
| 160 |
+
"logits/chosen": -0.07778320461511612,
|
| 161 |
+
"logits/rejected": -0.06748046725988388,
|
| 162 |
+
"logps/chosen": -685.5999755859375,
|
| 163 |
+
"logps/rejected": -568.5999755859375,
|
| 164 |
+
"loss": 0.6071,
|
| 165 |
+
"nll_loss": 0.9535156488418579,
|
| 166 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 167 |
+
"rewards/chosen": 0.2954345643520355,
|
| 168 |
+
"rewards/margins": 0.2642578184604645,
|
| 169 |
+
"rewards/rejected": 0.03144531324505806,
|
| 170 |
+
"step": 90
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.07639419404125286,
|
| 174 |
+
"grad_norm": 98.72713078984985,
|
| 175 |
+
"learning_rate": 3.7786259541984735e-07,
|
| 176 |
+
"logits/chosen": -0.2978515625,
|
| 177 |
+
"logits/rejected": -0.25141602754592896,
|
| 178 |
+
"logps/chosen": -519.7999877929688,
|
| 179 |
+
"logps/rejected": -522.4000244140625,
|
| 180 |
+
"loss": 0.5738,
|
| 181 |
+
"nll_loss": 1.067968726158142,
|
| 182 |
+
"rewards/accuracies": 0.625,
|
| 183 |
+
"rewards/chosen": 0.14719848334789276,
|
| 184 |
+
"rewards/margins": 0.42902833223342896,
|
| 185 |
+
"rewards/rejected": -0.28105467557907104,
|
| 186 |
+
"step": 100
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.08403361344537816,
|
| 190 |
+
"grad_norm": 124.37966570806609,
|
| 191 |
+
"learning_rate": 4.1603053435114506e-07,
|
| 192 |
+
"logits/chosen": -0.123291015625,
|
| 193 |
+
"logits/rejected": -0.130828857421875,
|
| 194 |
+
"logps/chosen": -548.0,
|
| 195 |
+
"logps/rejected": -491.6000061035156,
|
| 196 |
+
"loss": 0.5854,
|
| 197 |
+
"nll_loss": 0.877734363079071,
|
| 198 |
+
"rewards/accuracies": 0.550000011920929,
|
| 199 |
+
"rewards/chosen": 0.23984375596046448,
|
| 200 |
+
"rewards/margins": 0.43937987089157104,
|
| 201 |
+
"rewards/rejected": -0.19887694716453552,
|
| 202 |
+
"step": 110
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.09167303284950344,
|
| 206 |
+
"grad_norm": 96.92927310238335,
|
| 207 |
+
"learning_rate": 4.541984732824427e-07,
|
| 208 |
+
"logits/chosen": -0.15711669623851776,
|
| 209 |
+
"logits/rejected": -0.014111327938735485,
|
| 210 |
+
"logps/chosen": -496.79998779296875,
|
| 211 |
+
"logps/rejected": -501.20001220703125,
|
| 212 |
+
"loss": 0.5269,
|
| 213 |
+
"nll_loss": 0.9039062261581421,
|
| 214 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 215 |
+
"rewards/chosen": 0.15468749403953552,
|
| 216 |
+
"rewards/margins": 0.699902355670929,
|
| 217 |
+
"rewards/rejected": -0.544921875,
|
| 218 |
+
"step": 120
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.09931245225362872,
|
| 222 |
+
"grad_norm": 105.44903151573537,
|
| 223 |
+
"learning_rate": 4.923664122137405e-07,
|
| 224 |
+
"logits/chosen": 0.05002441257238388,
|
| 225 |
+
"logits/rejected": -0.05361328274011612,
|
| 226 |
+
"logps/chosen": -578.4000244140625,
|
| 227 |
+
"logps/rejected": -516.5999755859375,
|
| 228 |
+
"loss": 0.4919,
|
| 229 |
+
"nll_loss": 0.947265625,
|
| 230 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 231 |
+
"rewards/chosen": 0.523144543170929,
|
| 232 |
+
"rewards/margins": 0.8398681879043579,
|
| 233 |
+
"rewards/rejected": -0.3170837461948395,
|
| 234 |
+
"step": 130
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.10695187165775401,
|
| 238 |
+
"grad_norm": 67.21759101521735,
|
| 239 |
+
"learning_rate": 4.966044142614601e-07,
|
| 240 |
+
"logits/chosen": -0.07595214992761612,
|
| 241 |
+
"logits/rejected": 0.0031250000465661287,
|
| 242 |
+
"logps/chosen": -436.6000061035156,
|
| 243 |
+
"logps/rejected": -423.79998779296875,
|
| 244 |
+
"loss": 0.4388,
|
| 245 |
+
"nll_loss": 1.072265625,
|
| 246 |
+
"rewards/accuracies": 0.800000011920929,
|
| 247 |
+
"rewards/chosen": 0.4901367127895355,
|
| 248 |
+
"rewards/margins": 1.0460937023162842,
|
| 249 |
+
"rewards/rejected": -0.5567382574081421,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.11459129106187929,
|
| 254 |
+
"grad_norm": 68.2687002753991,
|
| 255 |
+
"learning_rate": 4.923599320882851e-07,
|
| 256 |
+
"logits/chosen": -0.11578369140625,
|
| 257 |
+
"logits/rejected": -0.09633789211511612,
|
| 258 |
+
"logps/chosen": -482.3999938964844,
|
| 259 |
+
"logps/rejected": -514.4000244140625,
|
| 260 |
+
"loss": 0.4481,
|
| 261 |
+
"nll_loss": 1.000390648841858,
|
| 262 |
+
"rewards/accuracies": 0.762499988079071,
|
| 263 |
+
"rewards/chosen": 0.5595703125,
|
| 264 |
+
"rewards/margins": 1.3738281726837158,
|
| 265 |
+
"rewards/rejected": -0.8130248785018921,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.12223071046600459,
|
| 270 |
+
"grad_norm": 127.87576870884502,
|
| 271 |
+
"learning_rate": 4.881154499151103e-07,
|
| 272 |
+
"logits/chosen": 0.094970703125,
|
| 273 |
+
"logits/rejected": 0.12006835639476776,
|
| 274 |
+
"logps/chosen": -520.5999755859375,
|
| 275 |
+
"logps/rejected": -488.6000061035156,
|
| 276 |
+
"loss": 0.4244,
|
| 277 |
+
"nll_loss": 1.0402343273162842,
|
| 278 |
+
"rewards/accuracies": 0.762499988079071,
|
| 279 |
+
"rewards/chosen": 0.3745971620082855,
|
| 280 |
+
"rewards/margins": 1.32421875,
|
| 281 |
+
"rewards/rejected": -0.950451672077179,
|
| 282 |
+
"step": 160
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.12987012987012986,
|
| 286 |
+
"grad_norm": 85.075639185204,
|
| 287 |
+
"learning_rate": 4.838709677419355e-07,
|
| 288 |
+
"logits/chosen": -0.03000488318502903,
|
| 289 |
+
"logits/rejected": 0.00927734375,
|
| 290 |
+
"logps/chosen": -564.4000244140625,
|
| 291 |
+
"logps/rejected": -466.0,
|
| 292 |
+
"loss": 0.5527,
|
| 293 |
+
"nll_loss": 0.9906250238418579,
|
| 294 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 295 |
+
"rewards/chosen": 0.9181152582168579,
|
| 296 |
+
"rewards/margins": 0.9400390386581421,
|
| 297 |
+
"rewards/rejected": -0.0205078125,
|
| 298 |
+
"step": 170
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.13750954927425516,
|
| 302 |
+
"grad_norm": 72.56682651323408,
|
| 303 |
+
"learning_rate": 4.796264855687606e-07,
|
| 304 |
+
"logits/chosen": 0.15263672173023224,
|
| 305 |
+
"logits/rejected": 0.12807616591453552,
|
| 306 |
+
"logps/chosen": -481.20001220703125,
|
| 307 |
+
"logps/rejected": -463.3999938964844,
|
| 308 |
+
"loss": 0.5139,
|
| 309 |
+
"nll_loss": 0.9906250238418579,
|
| 310 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 311 |
+
"rewards/chosen": 0.3233398497104645,
|
| 312 |
+
"rewards/margins": 1.0710937976837158,
|
| 313 |
+
"rewards/rejected": -0.748730480670929,
|
| 314 |
+
"step": 180
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.14514896867838045,
|
| 318 |
+
"grad_norm": 66.8084913798019,
|
| 319 |
+
"learning_rate": 4.753820033955857e-07,
|
| 320 |
+
"logits/chosen": 0.10141601413488388,
|
| 321 |
+
"logits/rejected": 0.028076171875,
|
| 322 |
+
"logps/chosen": -562.2000122070312,
|
| 323 |
+
"logps/rejected": -514.7999877929688,
|
| 324 |
+
"loss": 0.4805,
|
| 325 |
+
"nll_loss": 1.060937523841858,
|
| 326 |
+
"rewards/accuracies": 0.737500011920929,
|
| 327 |
+
"rewards/chosen": 0.14283446967601776,
|
| 328 |
+
"rewards/margins": 1.2859375476837158,
|
| 329 |
+
"rewards/rejected": -1.1462891101837158,
|
| 330 |
+
"step": 190
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.15278838808250572,
|
| 334 |
+
"grad_norm": 93.37492304215273,
|
| 335 |
+
"learning_rate": 4.7113752122241087e-07,
|
| 336 |
+
"logits/chosen": 0.13945312798023224,
|
| 337 |
+
"logits/rejected": 0.12504883110523224,
|
| 338 |
+
"logps/chosen": -555.0,
|
| 339 |
+
"logps/rejected": -517.0,
|
| 340 |
+
"loss": 0.5766,
|
| 341 |
+
"nll_loss": 0.900390625,
|
| 342 |
+
"rewards/accuracies": 0.637499988079071,
|
| 343 |
+
"rewards/chosen": 0.2847656309604645,
|
| 344 |
+
"rewards/margins": 0.858593761920929,
|
| 345 |
+
"rewards/rejected": -0.5726073980331421,
|
| 346 |
+
"step": 200
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.16042780748663102,
|
| 350 |
+
"grad_norm": 122.1908698464538,
|
| 351 |
+
"learning_rate": 4.66893039049236e-07,
|
| 352 |
+
"logits/chosen": 0.11003418266773224,
|
| 353 |
+
"logits/rejected": 0.07954101264476776,
|
| 354 |
+
"logps/chosen": -510.0,
|
| 355 |
+
"logps/rejected": -481.79998779296875,
|
| 356 |
+
"loss": 0.436,
|
| 357 |
+
"nll_loss": 0.966015636920929,
|
| 358 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 359 |
+
"rewards/chosen": -0.05595703050494194,
|
| 360 |
+
"rewards/margins": 1.149023413658142,
|
| 361 |
+
"rewards/rejected": -1.20556640625,
|
| 362 |
+
"step": 210
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.16806722689075632,
|
| 366 |
+
"grad_norm": 119.50281555399438,
|
| 367 |
+
"learning_rate": 4.6264855687606106e-07,
|
| 368 |
+
"logits/chosen": 0.03498535230755806,
|
| 369 |
+
"logits/rejected": 0.06562499701976776,
|
| 370 |
+
"logps/chosen": -458.20001220703125,
|
| 371 |
+
"logps/rejected": -433.3999938964844,
|
| 372 |
+
"loss": 0.4885,
|
| 373 |
+
"nll_loss": 1.021093726158142,
|
| 374 |
+
"rewards/accuracies": 0.75,
|
| 375 |
+
"rewards/chosen": -0.17633056640625,
|
| 376 |
+
"rewards/margins": 1.134374976158142,
|
| 377 |
+
"rewards/rejected": -1.310156226158142,
|
| 378 |
+
"step": 220
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.17570664629488159,
|
| 382 |
+
"grad_norm": 54.05921608969606,
|
| 383 |
+
"learning_rate": 4.5840407470288624e-07,
|
| 384 |
+
"logits/chosen": 0.14101561903953552,
|
| 385 |
+
"logits/rejected": 0.1561279296875,
|
| 386 |
+
"logps/chosen": -496.20001220703125,
|
| 387 |
+
"logps/rejected": -491.79998779296875,
|
| 388 |
+
"loss": 0.3993,
|
| 389 |
+
"nll_loss": 0.958984375,
|
| 390 |
+
"rewards/accuracies": 0.737500011920929,
|
| 391 |
+
"rewards/chosen": -0.15058593451976776,
|
| 392 |
+
"rewards/margins": 1.658593773841858,
|
| 393 |
+
"rewards/rejected": -1.8125,
|
| 394 |
+
"step": 230
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.18334606569900688,
|
| 398 |
+
"grad_norm": 122.95753964854363,
|
| 399 |
+
"learning_rate": 4.5415959252971136e-07,
|
| 400 |
+
"logits/chosen": 0.31416016817092896,
|
| 401 |
+
"logits/rejected": 0.24771729111671448,
|
| 402 |
+
"logps/chosen": -451.3999938964844,
|
| 403 |
+
"logps/rejected": -461.6000061035156,
|
| 404 |
+
"loss": 0.5026,
|
| 405 |
+
"nll_loss": 1.0085937976837158,
|
| 406 |
+
"rewards/accuracies": 0.675000011920929,
|
| 407 |
+
"rewards/chosen": -0.4728759825229645,
|
| 408 |
+
"rewards/margins": 1.413671851158142,
|
| 409 |
+
"rewards/rejected": -1.8835937976837158,
|
| 410 |
+
"step": 240
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.19098548510313215,
|
| 414 |
+
"grad_norm": 70.39494400187786,
|
| 415 |
+
"learning_rate": 4.499151103565365e-07,
|
| 416 |
+
"logits/chosen": 0.16347655653953552,
|
| 417 |
+
"logits/rejected": 0.20424804091453552,
|
| 418 |
+
"logps/chosen": -635.2000122070312,
|
| 419 |
+
"logps/rejected": -493.79998779296875,
|
| 420 |
+
"loss": 0.4695,
|
| 421 |
+
"nll_loss": 0.934374988079071,
|
| 422 |
+
"rewards/accuracies": 0.737500011920929,
|
| 423 |
+
"rewards/chosen": 0.5228027105331421,
|
| 424 |
+
"rewards/margins": 1.4542968273162842,
|
| 425 |
+
"rewards/rejected": -0.932080090045929,
|
| 426 |
+
"step": 250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.19862490450725745,
|
| 430 |
+
"grad_norm": 98.89422904437872,
|
| 431 |
+
"learning_rate": 4.456706281833616e-07,
|
| 432 |
+
"logits/chosen": 0.2513671815395355,
|
| 433 |
+
"logits/rejected": 0.1383056640625,
|
| 434 |
+
"logps/chosen": -476.20001220703125,
|
| 435 |
+
"logps/rejected": -469.3999938964844,
|
| 436 |
+
"loss": 0.5171,
|
| 437 |
+
"nll_loss": 1.030859351158142,
|
| 438 |
+
"rewards/accuracies": 0.675000011920929,
|
| 439 |
+
"rewards/chosen": 0.675097644329071,
|
| 440 |
+
"rewards/margins": 1.3078124523162842,
|
| 441 |
+
"rewards/rejected": -0.6328125,
|
| 442 |
+
"step": 260
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.20626432391138275,
|
| 446 |
+
"grad_norm": 87.2798705625053,
|
| 447 |
+
"learning_rate": 4.4142614601018673e-07,
|
| 448 |
+
"logits/chosen": 0.13022461533546448,
|
| 449 |
+
"logits/rejected": 0.320068359375,
|
| 450 |
+
"logps/chosen": -515.0,
|
| 451 |
+
"logps/rejected": -514.2000122070312,
|
| 452 |
+
"loss": 0.4791,
|
| 453 |
+
"nll_loss": 0.948046863079071,
|
| 454 |
+
"rewards/accuracies": 0.6875,
|
| 455 |
+
"rewards/chosen": 0.964648425579071,
|
| 456 |
+
"rewards/margins": 1.5105469226837158,
|
| 457 |
+
"rewards/rejected": -0.5445312261581421,
|
| 458 |
+
"step": 270
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.21390374331550802,
|
| 462 |
+
"grad_norm": 108.15959014981838,
|
| 463 |
+
"learning_rate": 4.3718166383701186e-07,
|
| 464 |
+
"logits/chosen": -0.01950683631002903,
|
| 465 |
+
"logits/rejected": 0.17734375596046448,
|
| 466 |
+
"logps/chosen": -593.7999877929688,
|
| 467 |
+
"logps/rejected": -553.2000122070312,
|
| 468 |
+
"loss": 0.4719,
|
| 469 |
+
"nll_loss": 1.0437500476837158,
|
| 470 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 471 |
+
"rewards/chosen": 1.063085913658142,
|
| 472 |
+
"rewards/margins": 1.41015625,
|
| 473 |
+
"rewards/rejected": -0.34550780057907104,
|
| 474 |
+
"step": 280
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.2215431627196333,
|
| 478 |
+
"grad_norm": 84.53969769192312,
|
| 479 |
+
"learning_rate": 4.32937181663837e-07,
|
| 480 |
+
"logits/chosen": 0.29316407442092896,
|
| 481 |
+
"logits/rejected": 0.2685302793979645,
|
| 482 |
+
"logps/chosen": -584.2000122070312,
|
| 483 |
+
"logps/rejected": -543.0,
|
| 484 |
+
"loss": 0.4121,
|
| 485 |
+
"nll_loss": 0.9039062261581421,
|
| 486 |
+
"rewards/accuracies": 0.75,
|
| 487 |
+
"rewards/chosen": 0.831738293170929,
|
| 488 |
+
"rewards/margins": 1.84765625,
|
| 489 |
+
"rewards/rejected": -1.015625,
|
| 490 |
+
"step": 290
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.22918258212375858,
|
| 494 |
+
"grad_norm": 178.4258829118625,
|
| 495 |
+
"learning_rate": 4.286926994906621e-07,
|
| 496 |
+
"logits/chosen": 0.11018066108226776,
|
| 497 |
+
"logits/rejected": 0.03170166164636612,
|
| 498 |
+
"logps/chosen": -530.2000122070312,
|
| 499 |
+
"logps/rejected": -490.79998779296875,
|
| 500 |
+
"loss": 0.4702,
|
| 501 |
+
"nll_loss": 0.9750000238418579,
|
| 502 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 503 |
+
"rewards/chosen": 0.4922851622104645,
|
| 504 |
+
"rewards/margins": 1.6189453601837158,
|
| 505 |
+
"rewards/rejected": -1.127587914466858,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.23682200152788388,
|
| 510 |
+
"grad_norm": 177.58002775235644,
|
| 511 |
+
"learning_rate": 4.244482173174873e-07,
|
| 512 |
+
"logits/chosen": 0.19103392958641052,
|
| 513 |
+
"logits/rejected": 0.210693359375,
|
| 514 |
+
"logps/chosen": -510.6000061035156,
|
| 515 |
+
"logps/rejected": -456.3999938964844,
|
| 516 |
+
"loss": 0.3855,
|
| 517 |
+
"nll_loss": 0.900390625,
|
| 518 |
+
"rewards/accuracies": 0.800000011920929,
|
| 519 |
+
"rewards/chosen": 0.91552734375,
|
| 520 |
+
"rewards/margins": 1.4249999523162842,
|
| 521 |
+
"rewards/rejected": -0.509326159954071,
|
| 522 |
+
"step": 310
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.24446142093200918,
|
| 526 |
+
"grad_norm": 87.23485706050842,
|
| 527 |
+
"learning_rate": 4.202037351443124e-07,
|
| 528 |
+
"logits/chosen": 0.07420654594898224,
|
| 529 |
+
"logits/rejected": 0.21796874701976776,
|
| 530 |
+
"logps/chosen": -473.6000061035156,
|
| 531 |
+
"logps/rejected": -471.6000061035156,
|
| 532 |
+
"loss": 0.455,
|
| 533 |
+
"nll_loss": 1.019921898841858,
|
| 534 |
+
"rewards/accuracies": 0.737500011920929,
|
| 535 |
+
"rewards/chosen": 0.5599609613418579,
|
| 536 |
+
"rewards/margins": 1.7609374523162842,
|
| 537 |
+
"rewards/rejected": -1.200781226158142,
|
| 538 |
+
"step": 320
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.25210084033613445,
|
| 542 |
+
"grad_norm": 118.39889100233657,
|
| 543 |
+
"learning_rate": 4.159592529711375e-07,
|
| 544 |
+
"logits/chosen": 0.15974120795726776,
|
| 545 |
+
"logits/rejected": 0.20943602919578552,
|
| 546 |
+
"logps/chosen": -520.2000122070312,
|
| 547 |
+
"logps/rejected": -501.0,
|
| 548 |
+
"loss": 0.4228,
|
| 549 |
+
"nll_loss": 0.9609375,
|
| 550 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 551 |
+
"rewards/chosen": 0.89111328125,
|
| 552 |
+
"rewards/margins": 1.538671851158142,
|
| 553 |
+
"rewards/rejected": -0.647900402545929,
|
| 554 |
+
"step": 330
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.2597402597402597,
|
| 558 |
+
"grad_norm": 55.527233131819315,
|
| 559 |
+
"learning_rate": 4.1171477079796265e-07,
|
| 560 |
+
"logits/chosen": 0.18583984673023224,
|
| 561 |
+
"logits/rejected": 0.23032227158546448,
|
| 562 |
+
"logps/chosen": -536.5999755859375,
|
| 563 |
+
"logps/rejected": -527.5999755859375,
|
| 564 |
+
"loss": 0.4781,
|
| 565 |
+
"nll_loss": 0.936718761920929,
|
| 566 |
+
"rewards/accuracies": 0.75,
|
| 567 |
+
"rewards/chosen": -0.18813475966453552,
|
| 568 |
+
"rewards/margins": 1.195898413658142,
|
| 569 |
+
"rewards/rejected": -1.384765625,
|
| 570 |
+
"step": 340
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.26737967914438504,
|
| 574 |
+
"grad_norm": 55.046301267314455,
|
| 575 |
+
"learning_rate": 4.074702886247878e-07,
|
| 576 |
+
"logits/chosen": 0.0034667968284338713,
|
| 577 |
+
"logits/rejected": 0.0003173828008584678,
|
| 578 |
+
"logps/chosen": -525.7999877929688,
|
| 579 |
+
"logps/rejected": -488.0,
|
| 580 |
+
"loss": 0.4043,
|
| 581 |
+
"nll_loss": 0.985546886920929,
|
| 582 |
+
"rewards/accuracies": 0.800000011920929,
|
| 583 |
+
"rewards/chosen": 0.538769543170929,
|
| 584 |
+
"rewards/margins": 1.618749976158142,
|
| 585 |
+
"rewards/rejected": -1.0803711414337158,
|
| 586 |
+
"step": 350
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.2750190985485103,
|
| 590 |
+
"grad_norm": 74.35779638281001,
|
| 591 |
+
"learning_rate": 4.0322580645161285e-07,
|
| 592 |
+
"logits/chosen": 0.15576171875,
|
| 593 |
+
"logits/rejected": 0.2554931640625,
|
| 594 |
+
"logps/chosen": -586.7999877929688,
|
| 595 |
+
"logps/rejected": -572.4000244140625,
|
| 596 |
+
"loss": 0.5316,
|
| 597 |
+
"nll_loss": 1.003515601158142,
|
| 598 |
+
"rewards/accuracies": 0.675000011920929,
|
| 599 |
+
"rewards/chosen": -0.10053710639476776,
|
| 600 |
+
"rewards/margins": 1.283203125,
|
| 601 |
+
"rewards/rejected": -1.385351538658142,
|
| 602 |
+
"step": 360
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.2826585179526356,
|
| 606 |
+
"grad_norm": 80.55889148977015,
|
| 607 |
+
"learning_rate": 3.98981324278438e-07,
|
| 608 |
+
"logits/chosen": -0.086669921875,
|
| 609 |
+
"logits/rejected": 0.0487060546875,
|
| 610 |
+
"logps/chosen": -523.2000122070312,
|
| 611 |
+
"logps/rejected": -488.20001220703125,
|
| 612 |
+
"loss": 0.4494,
|
| 613 |
+
"nll_loss": 1.0398437976837158,
|
| 614 |
+
"rewards/accuracies": 0.737500011920929,
|
| 615 |
+
"rewards/chosen": -0.313232421875,
|
| 616 |
+
"rewards/margins": 1.748437523841858,
|
| 617 |
+
"rewards/rejected": -2.060839891433716,
|
| 618 |
+
"step": 370
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.2902979373567609,
|
| 622 |
+
"grad_norm": 79.33557174780084,
|
| 623 |
+
"learning_rate": 3.9473684210526315e-07,
|
| 624 |
+
"logits/chosen": 0.01704101637005806,
|
| 625 |
+
"logits/rejected": 0.013476562686264515,
|
| 626 |
+
"logps/chosen": -593.5999755859375,
|
| 627 |
+
"logps/rejected": -577.2000122070312,
|
| 628 |
+
"loss": 0.5144,
|
| 629 |
+
"nll_loss": 1.001953125,
|
| 630 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 631 |
+
"rewards/chosen": 0.20126953721046448,
|
| 632 |
+
"rewards/margins": 1.30078125,
|
| 633 |
+
"rewards/rejected": -1.0986328125,
|
| 634 |
+
"step": 380
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.2979373567608862,
|
| 638 |
+
"grad_norm": 150.3385845904889,
|
| 639 |
+
"learning_rate": 3.9049235993208827e-07,
|
| 640 |
+
"logits/chosen": 0.10019531100988388,
|
| 641 |
+
"logits/rejected": -0.04719238355755806,
|
| 642 |
+
"logps/chosen": -523.7999877929688,
|
| 643 |
+
"logps/rejected": -526.5999755859375,
|
| 644 |
+
"loss": 0.5132,
|
| 645 |
+
"nll_loss": 1.1007812023162842,
|
| 646 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 647 |
+
"rewards/chosen": -0.0283203125,
|
| 648 |
+
"rewards/margins": 1.380517601966858,
|
| 649 |
+
"rewards/rejected": -1.409277319908142,
|
| 650 |
+
"step": 390
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.30557677616501144,
|
| 654 |
+
"grad_norm": 42.55557637305296,
|
| 655 |
+
"learning_rate": 3.862478777589134e-07,
|
| 656 |
+
"logits/chosen": 0.07124023139476776,
|
| 657 |
+
"logits/rejected": 0.06547851860523224,
|
| 658 |
+
"logps/chosen": -547.2000122070312,
|
| 659 |
+
"logps/rejected": -544.7999877929688,
|
| 660 |
+
"loss": 0.3939,
|
| 661 |
+
"nll_loss": 1.0046875476837158,
|
| 662 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 663 |
+
"rewards/chosen": 0.634228527545929,
|
| 664 |
+
"rewards/margins": 1.944921851158142,
|
| 665 |
+
"rewards/rejected": -1.310546875,
|
| 666 |
+
"step": 400
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.31321619556913677,
|
| 670 |
+
"grad_norm": 65.88455118968501,
|
| 671 |
+
"learning_rate": 3.820033955857385e-07,
|
| 672 |
+
"logits/chosen": 0.09038086235523224,
|
| 673 |
+
"logits/rejected": 0.073974609375,
|
| 674 |
+
"logps/chosen": -521.5999755859375,
|
| 675 |
+
"logps/rejected": -469.3999938964844,
|
| 676 |
+
"loss": 0.3976,
|
| 677 |
+
"nll_loss": 0.9765625,
|
| 678 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 679 |
+
"rewards/chosen": -0.05693359300494194,
|
| 680 |
+
"rewards/margins": 1.83203125,
|
| 681 |
+
"rewards/rejected": -1.890039086341858,
|
| 682 |
+
"step": 410
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.32085561497326204,
|
| 686 |
+
"grad_norm": 90.26286302852353,
|
| 687 |
+
"learning_rate": 3.7775891341256364e-07,
|
| 688 |
+
"logits/chosen": 0.09355469048023224,
|
| 689 |
+
"logits/rejected": 0.1094970703125,
|
| 690 |
+
"logps/chosen": -601.5999755859375,
|
| 691 |
+
"logps/rejected": -540.7999877929688,
|
| 692 |
+
"loss": 0.4069,
|
| 693 |
+
"nll_loss": 0.950390636920929,
|
| 694 |
+
"rewards/accuracies": 0.6875,
|
| 695 |
+
"rewards/chosen": -0.05307617038488388,
|
| 696 |
+
"rewards/margins": 1.935156226158142,
|
| 697 |
+
"rewards/rejected": -1.989843726158142,
|
| 698 |
+
"step": 420
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.3284950343773873,
|
| 702 |
+
"grad_norm": 79.4961646277846,
|
| 703 |
+
"learning_rate": 3.735144312393888e-07,
|
| 704 |
+
"logits/chosen": 0.02392578125,
|
| 705 |
+
"logits/rejected": 0.11666259914636612,
|
| 706 |
+
"logps/chosen": -563.2000122070312,
|
| 707 |
+
"logps/rejected": -509.6000061035156,
|
| 708 |
+
"loss": 0.5029,
|
| 709 |
+
"nll_loss": 0.9609375,
|
| 710 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 711 |
+
"rewards/chosen": -1.2892577648162842,
|
| 712 |
+
"rewards/margins": 1.270898461341858,
|
| 713 |
+
"rewards/rejected": -2.5609374046325684,
|
| 714 |
+
"step": 430
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.33613445378151263,
|
| 718 |
+
"grad_norm": 47.27348830374442,
|
| 719 |
+
"learning_rate": 3.692699490662139e-07,
|
| 720 |
+
"logits/chosen": 0.3138671815395355,
|
| 721 |
+
"logits/rejected": 0.2645019590854645,
|
| 722 |
+
"logps/chosen": -562.2000122070312,
|
| 723 |
+
"logps/rejected": -525.7999877929688,
|
| 724 |
+
"loss": 0.5141,
|
| 725 |
+
"nll_loss": 1.0089843273162842,
|
| 726 |
+
"rewards/accuracies": 0.675000011920929,
|
| 727 |
+
"rewards/chosen": -1.4290039539337158,
|
| 728 |
+
"rewards/margins": 1.462890625,
|
| 729 |
+
"rewards/rejected": -2.892578125,
|
| 730 |
+
"step": 440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.3437738731856379,
|
| 734 |
+
"grad_norm": 136.72527131588762,
|
| 735 |
+
"learning_rate": 3.65025466893039e-07,
|
| 736 |
+
"logits/chosen": 0.07827148586511612,
|
| 737 |
+
"logits/rejected": 0.13486328721046448,
|
| 738 |
+
"logps/chosen": -564.4000244140625,
|
| 739 |
+
"logps/rejected": -536.7999877929688,
|
| 740 |
+
"loss": 0.488,
|
| 741 |
+
"nll_loss": 1.119531273841858,
|
| 742 |
+
"rewards/accuracies": 0.75,
|
| 743 |
+
"rewards/chosen": -0.4256835877895355,
|
| 744 |
+
"rewards/margins": 1.5109374523162842,
|
| 745 |
+
"rewards/rejected": -1.9357421398162842,
|
| 746 |
+
"step": 450
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.35141329258976317,
|
| 750 |
+
"grad_norm": 50.953994445015056,
|
| 751 |
+
"learning_rate": 3.607809847198642e-07,
|
| 752 |
+
"logits/chosen": -0.14997711777687073,
|
| 753 |
+
"logits/rejected": 0.03886718675494194,
|
| 754 |
+
"logps/chosen": -514.0,
|
| 755 |
+
"logps/rejected": -524.4000244140625,
|
| 756 |
+
"loss": 0.4767,
|
| 757 |
+
"nll_loss": 1.031640648841858,
|
| 758 |
+
"rewards/accuracies": 0.762499988079071,
|
| 759 |
+
"rewards/chosen": -0.5672851800918579,
|
| 760 |
+
"rewards/margins": 1.609765648841858,
|
| 761 |
+
"rewards/rejected": -2.1778807640075684,
|
| 762 |
+
"step": 460
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.35905271199388844,
|
| 766 |
+
"grad_norm": 240.3612883527372,
|
| 767 |
+
"learning_rate": 3.5653650254668926e-07,
|
| 768 |
+
"logits/chosen": 0.20814208686351776,
|
| 769 |
+
"logits/rejected": 0.13437500596046448,
|
| 770 |
+
"logps/chosen": -537.0,
|
| 771 |
+
"logps/rejected": -507.0,
|
| 772 |
+
"loss": 0.5926,
|
| 773 |
+
"nll_loss": 0.987500011920929,
|
| 774 |
+
"rewards/accuracies": 0.675000011920929,
|
| 775 |
+
"rewards/chosen": -0.91497802734375,
|
| 776 |
+
"rewards/margins": 1.119531273841858,
|
| 777 |
+
"rewards/rejected": -2.037109375,
|
| 778 |
+
"step": 470
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.36669213139801377,
|
| 782 |
+
"grad_norm": 87.08199892967568,
|
| 783 |
+
"learning_rate": 3.5229202037351443e-07,
|
| 784 |
+
"logits/chosen": -0.12431640923023224,
|
| 785 |
+
"logits/rejected": -0.07424316555261612,
|
| 786 |
+
"logps/chosen": -473.20001220703125,
|
| 787 |
+
"logps/rejected": -490.0,
|
| 788 |
+
"loss": 0.521,
|
| 789 |
+
"nll_loss": 0.983593761920929,
|
| 790 |
+
"rewards/accuracies": 0.6875,
|
| 791 |
+
"rewards/chosen": -0.677001953125,
|
| 792 |
+
"rewards/margins": 1.192968726158142,
|
| 793 |
+
"rewards/rejected": -1.8699219226837158,
|
| 794 |
+
"step": 480
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.37433155080213903,
|
| 798 |
+
"grad_norm": 92.89135301388274,
|
| 799 |
+
"learning_rate": 3.4804753820033956e-07,
|
| 800 |
+
"logits/chosen": -7.324219041038305e-05,
|
| 801 |
+
"logits/rejected": 0.10617675632238388,
|
| 802 |
+
"logps/chosen": -597.4000244140625,
|
| 803 |
+
"logps/rejected": -560.0,
|
| 804 |
+
"loss": 0.3843,
|
| 805 |
+
"nll_loss": 1.0382812023162842,
|
| 806 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 807 |
+
"rewards/chosen": -0.25146484375,
|
| 808 |
+
"rewards/margins": 1.8230469226837158,
|
| 809 |
+
"rewards/rejected": -2.075000047683716,
|
| 810 |
+
"step": 490
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.3819709702062643,
|
| 814 |
+
"grad_norm": 88.06852905492813,
|
| 815 |
+
"learning_rate": 3.438030560271647e-07,
|
| 816 |
+
"logits/chosen": 0.06113281100988388,
|
| 817 |
+
"logits/rejected": 0.12486572563648224,
|
| 818 |
+
"logps/chosen": -531.4000244140625,
|
| 819 |
+
"logps/rejected": -499.20001220703125,
|
| 820 |
+
"loss": 0.506,
|
| 821 |
+
"nll_loss": 1.047265648841858,
|
| 822 |
+
"rewards/accuracies": 0.699999988079071,
|
| 823 |
+
"rewards/chosen": -0.698437511920929,
|
| 824 |
+
"rewards/margins": 1.429296851158142,
|
| 825 |
+
"rewards/rejected": -2.130078077316284,
|
| 826 |
+
"step": 500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.38961038961038963,
|
| 830 |
+
"grad_norm": 127.28421372369633,
|
| 831 |
+
"learning_rate": 3.395585738539898e-07,
|
| 832 |
+
"logits/chosen": 0.06453857570886612,
|
| 833 |
+
"logits/rejected": 0.12724609673023224,
|
| 834 |
+
"logps/chosen": -607.5999755859375,
|
| 835 |
+
"logps/rejected": -596.5999755859375,
|
| 836 |
+
"loss": 0.4705,
|
| 837 |
+
"nll_loss": 0.899218738079071,
|
| 838 |
+
"rewards/accuracies": 0.6875,
|
| 839 |
+
"rewards/chosen": -0.493896484375,
|
| 840 |
+
"rewards/margins": 1.21044921875,
|
| 841 |
+
"rewards/rejected": -1.704687476158142,
|
| 842 |
+
"step": 510
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.3972498090145149,
|
| 846 |
+
"grad_norm": 106.37931014491402,
|
| 847 |
+
"learning_rate": 3.3531409168081493e-07,
|
| 848 |
+
"logits/chosen": 0.074462890625,
|
| 849 |
+
"logits/rejected": -0.02686767652630806,
|
| 850 |
+
"logps/chosen": -612.7999877929688,
|
| 851 |
+
"logps/rejected": -560.2000122070312,
|
| 852 |
+
"loss": 0.5153,
|
| 853 |
+
"nll_loss": 0.954296886920929,
|
| 854 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 855 |
+
"rewards/chosen": -0.33642578125,
|
| 856 |
+
"rewards/margins": 1.134179711341858,
|
| 857 |
+
"rewards/rejected": -1.4684569835662842,
|
| 858 |
+
"step": 520
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.40488922841864017,
|
| 862 |
+
"grad_norm": 92.15914610298877,
|
| 863 |
+
"learning_rate": 3.3106960950764005e-07,
|
| 864 |
+
"logits/chosen": -0.07530517876148224,
|
| 865 |
+
"logits/rejected": -0.0087890625,
|
| 866 |
+
"logps/chosen": -639.7999877929688,
|
| 867 |
+
"logps/rejected": -620.4000244140625,
|
| 868 |
+
"loss": 0.4675,
|
| 869 |
+
"nll_loss": 1.0304687023162842,
|
| 870 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 871 |
+
"rewards/chosen": -0.21469727158546448,
|
| 872 |
+
"rewards/margins": 1.3169677257537842,
|
| 873 |
+
"rewards/rejected": -1.5315430164337158,
|
| 874 |
+
"step": 530
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.4125286478227655,
|
| 878 |
+
"grad_norm": 68.26427654711365,
|
| 879 |
+
"learning_rate": 3.268251273344652e-07,
|
| 880 |
+
"logits/chosen": 0.06784667819738388,
|
| 881 |
+
"logits/rejected": 0.15874023735523224,
|
| 882 |
+
"logps/chosen": -566.2000122070312,
|
| 883 |
+
"logps/rejected": -525.7999877929688,
|
| 884 |
+
"loss": 0.3663,
|
| 885 |
+
"nll_loss": 0.9828125238418579,
|
| 886 |
+
"rewards/accuracies": 0.737500011920929,
|
| 887 |
+
"rewards/chosen": -0.00732421875,
|
| 888 |
+
"rewards/margins": 1.726171851158142,
|
| 889 |
+
"rewards/rejected": -1.734472632408142,
|
| 890 |
+
"step": 540
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.42016806722689076,
|
| 894 |
+
"grad_norm": 79.05578583438974,
|
| 895 |
+
"learning_rate": 3.225806451612903e-07,
|
| 896 |
+
"logits/chosen": -0.0009765625,
|
| 897 |
+
"logits/rejected": 0.02316894568502903,
|
| 898 |
+
"logps/chosen": -535.2000122070312,
|
| 899 |
+
"logps/rejected": -485.6000061035156,
|
| 900 |
+
"loss": 0.3245,
|
| 901 |
+
"nll_loss": 0.958203136920929,
|
| 902 |
+
"rewards/accuracies": 0.800000011920929,
|
| 903 |
+
"rewards/chosen": -1.0714843273162842,
|
| 904 |
+
"rewards/margins": 2.037890672683716,
|
| 905 |
+
"rewards/rejected": -3.1092529296875,
|
| 906 |
+
"step": 550
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.42780748663101603,
|
| 910 |
+
"grad_norm": 72.72051100521483,
|
| 911 |
+
"learning_rate": 3.183361629881154e-07,
|
| 912 |
+
"logits/chosen": 0.33515626192092896,
|
| 913 |
+
"logits/rejected": 0.31201171875,
|
| 914 |
+
"logps/chosen": -551.2000122070312,
|
| 915 |
+
"logps/rejected": -521.2000122070312,
|
| 916 |
+
"loss": 0.5041,
|
| 917 |
+
"nll_loss": 1.116796851158142,
|
| 918 |
+
"rewards/accuracies": 0.675000011920929,
|
| 919 |
+
"rewards/chosen": -1.181640625,
|
| 920 |
+
"rewards/margins": 1.537109375,
|
| 921 |
+
"rewards/rejected": -2.719531297683716,
|
| 922 |
+
"step": 560
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.43544690603514136,
|
| 926 |
+
"grad_norm": 101.18450597632666,
|
| 927 |
+
"learning_rate": 3.140916808149406e-07,
|
| 928 |
+
"logits/chosen": 0.19754639267921448,
|
| 929 |
+
"logits/rejected": 0.2854980528354645,
|
| 930 |
+
"logps/chosen": -634.0,
|
| 931 |
+
"logps/rejected": -620.2000122070312,
|
| 932 |
+
"loss": 0.4623,
|
| 933 |
+
"nll_loss": 1.0085937976837158,
|
| 934 |
+
"rewards/accuracies": 0.6875,
|
| 935 |
+
"rewards/chosen": -1.019921898841858,
|
| 936 |
+
"rewards/margins": 1.576171875,
|
| 937 |
+
"rewards/rejected": -2.5960936546325684,
|
| 938 |
+
"step": 570
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.4430863254392666,
|
| 942 |
+
"grad_norm": 62.30940742785987,
|
| 943 |
+
"learning_rate": 3.0984719864176567e-07,
|
| 944 |
+
"logits/chosen": 0.18598632514476776,
|
| 945 |
+
"logits/rejected": 0.390625,
|
| 946 |
+
"logps/chosen": -529.2000122070312,
|
| 947 |
+
"logps/rejected": -501.3999938964844,
|
| 948 |
+
"loss": 0.3935,
|
| 949 |
+
"nll_loss": 0.9593750238418579,
|
| 950 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 951 |
+
"rewards/chosen": -0.7264648675918579,
|
| 952 |
+
"rewards/margins": 1.8708984851837158,
|
| 953 |
+
"rewards/rejected": -2.5953125953674316,
|
| 954 |
+
"step": 580
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.4507257448433919,
|
| 958 |
+
"grad_norm": 81.31513302420962,
|
| 959 |
+
"learning_rate": 3.056027164685908e-07,
|
| 960 |
+
"logits/chosen": 0.06633301079273224,
|
| 961 |
+
"logits/rejected": 0.10258789360523224,
|
| 962 |
+
"logps/chosen": -542.7999877929688,
|
| 963 |
+
"logps/rejected": -565.4000244140625,
|
| 964 |
+
"loss": 0.5325,
|
| 965 |
+
"nll_loss": 1.1355469226837158,
|
| 966 |
+
"rewards/accuracies": 0.6875,
|
| 967 |
+
"rewards/chosen": -1.4011719226837158,
|
| 968 |
+
"rewards/margins": 1.4412109851837158,
|
| 969 |
+
"rewards/rejected": -2.8414063453674316,
|
| 970 |
+
"step": 590
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.45836516424751717,
|
| 974 |
+
"grad_norm": 104.05872957710817,
|
| 975 |
+
"learning_rate": 3.0135823429541597e-07,
|
| 976 |
+
"logits/chosen": 0.026275634765625,
|
| 977 |
+
"logits/rejected": 0.14370116591453552,
|
| 978 |
+
"logps/chosen": -574.5999755859375,
|
| 979 |
+
"logps/rejected": -551.7999877929688,
|
| 980 |
+
"loss": 0.5553,
|
| 981 |
+
"nll_loss": 0.985546886920929,
|
| 982 |
+
"rewards/accuracies": 0.6875,
|
| 983 |
+
"rewards/chosen": -1.1213867664337158,
|
| 984 |
+
"rewards/margins": 1.3791015148162842,
|
| 985 |
+
"rewards/rejected": -2.4986329078674316,
|
| 986 |
+
"step": 600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.4660045836516425,
|
| 990 |
+
"grad_norm": 90.94069952298143,
|
| 991 |
+
"learning_rate": 2.9711375212224104e-07,
|
| 992 |
+
"logits/chosen": 0.18339844048023224,
|
| 993 |
+
"logits/rejected": 0.17988280951976776,
|
| 994 |
+
"logps/chosen": -555.7999877929688,
|
| 995 |
+
"logps/rejected": -538.4000244140625,
|
| 996 |
+
"loss": 0.5513,
|
| 997 |
+
"nll_loss": 0.989453136920929,
|
| 998 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 999 |
+
"rewards/chosen": -0.832812488079071,
|
| 1000 |
+
"rewards/margins": 1.2374999523162842,
|
| 1001 |
+
"rewards/rejected": -2.067187547683716,
|
| 1002 |
+
"step": 610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.47364400305576776,
|
| 1006 |
+
"grad_norm": 38.710202372779044,
|
| 1007 |
+
"learning_rate": 2.928692699490662e-07,
|
| 1008 |
+
"logits/chosen": 0.09355469048023224,
|
| 1009 |
+
"logits/rejected": 0.11258544772863388,
|
| 1010 |
+
"logps/chosen": -601.5999755859375,
|
| 1011 |
+
"logps/rejected": -556.5999755859375,
|
| 1012 |
+
"loss": 0.3726,
|
| 1013 |
+
"nll_loss": 1.028906226158142,
|
| 1014 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1015 |
+
"rewards/chosen": -0.8633788824081421,
|
| 1016 |
+
"rewards/margins": 1.5730469226837158,
|
| 1017 |
+
"rewards/rejected": -2.438281297683716,
|
| 1018 |
+
"step": 620
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.48128342245989303,
|
| 1022 |
+
"grad_norm": 33.45736397793641,
|
| 1023 |
+
"learning_rate": 2.8862478777589134e-07,
|
| 1024 |
+
"logits/chosen": -0.05854492262005806,
|
| 1025 |
+
"logits/rejected": -0.072296142578125,
|
| 1026 |
+
"logps/chosen": -635.2000122070312,
|
| 1027 |
+
"logps/rejected": -611.2000122070312,
|
| 1028 |
+
"loss": 0.4438,
|
| 1029 |
+
"nll_loss": 1.0750000476837158,
|
| 1030 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1031 |
+
"rewards/chosen": -0.956738293170929,
|
| 1032 |
+
"rewards/margins": 1.6749999523162842,
|
| 1033 |
+
"rewards/rejected": -2.626953125,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.48892284186401835,
|
| 1038 |
+
"grad_norm": 89.25083718246636,
|
| 1039 |
+
"learning_rate": 2.8438030560271646e-07,
|
| 1040 |
+
"logits/chosen": 0.1068115234375,
|
| 1041 |
+
"logits/rejected": 0.04353027418255806,
|
| 1042 |
+
"logps/chosen": -652.7999877929688,
|
| 1043 |
+
"logps/rejected": -652.7999877929688,
|
| 1044 |
+
"loss": 0.4249,
|
| 1045 |
+
"nll_loss": 1.197265625,
|
| 1046 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1047 |
+
"rewards/chosen": -0.766528308391571,
|
| 1048 |
+
"rewards/margins": 1.6980469226837158,
|
| 1049 |
+
"rewards/rejected": -2.4664063453674316,
|
| 1050 |
+
"step": 640
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.4965622612681436,
|
| 1054 |
+
"grad_norm": 74.29229238796317,
|
| 1055 |
+
"learning_rate": 2.801358234295416e-07,
|
| 1056 |
+
"logits/chosen": 0.14731445908546448,
|
| 1057 |
+
"logits/rejected": 0.22744140028953552,
|
| 1058 |
+
"logps/chosen": -451.79998779296875,
|
| 1059 |
+
"logps/rejected": -463.20001220703125,
|
| 1060 |
+
"loss": 0.4997,
|
| 1061 |
+
"nll_loss": 0.9644531011581421,
|
| 1062 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1063 |
+
"rewards/chosen": -1.562109351158142,
|
| 1064 |
+
"rewards/margins": 1.396093726158142,
|
| 1065 |
+
"rewards/rejected": -2.9593749046325684,
|
| 1066 |
+
"step": 650
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.5042016806722689,
|
| 1070 |
+
"grad_norm": 151.21250104416566,
|
| 1071 |
+
"learning_rate": 2.758913412563667e-07,
|
| 1072 |
+
"logits/chosen": 0.13557128608226776,
|
| 1073 |
+
"logits/rejected": 0.2906494140625,
|
| 1074 |
+
"logps/chosen": -526.5999755859375,
|
| 1075 |
+
"logps/rejected": -476.0,
|
| 1076 |
+
"loss": 0.4011,
|
| 1077 |
+
"nll_loss": 1.02734375,
|
| 1078 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1079 |
+
"rewards/chosen": -0.814892590045929,
|
| 1080 |
+
"rewards/margins": 1.8203125,
|
| 1081 |
+
"rewards/rejected": -2.635937452316284,
|
| 1082 |
+
"step": 660
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.5118411000763942,
|
| 1086 |
+
"grad_norm": 91.91804317104803,
|
| 1087 |
+
"learning_rate": 2.7164685908319183e-07,
|
| 1088 |
+
"logits/chosen": 0.20478515326976776,
|
| 1089 |
+
"logits/rejected": 0.33662110567092896,
|
| 1090 |
+
"logps/chosen": -529.0,
|
| 1091 |
+
"logps/rejected": -514.2000122070312,
|
| 1092 |
+
"loss": 0.5713,
|
| 1093 |
+
"nll_loss": 1.0546875,
|
| 1094 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1095 |
+
"rewards/chosen": -0.23017577826976776,
|
| 1096 |
+
"rewards/margins": 1.3191406726837158,
|
| 1097 |
+
"rewards/rejected": -1.549218773841858,
|
| 1098 |
+
"step": 670
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.5194805194805194,
|
| 1102 |
+
"grad_norm": 24.046031163829962,
|
| 1103 |
+
"learning_rate": 2.67402376910017e-07,
|
| 1104 |
+
"logits/chosen": 0.08054199069738388,
|
| 1105 |
+
"logits/rejected": 0.14824219048023224,
|
| 1106 |
+
"logps/chosen": -506.0,
|
| 1107 |
+
"logps/rejected": -516.5999755859375,
|
| 1108 |
+
"loss": 0.4506,
|
| 1109 |
+
"nll_loss": 1.0359375476837158,
|
| 1110 |
+
"rewards/accuracies": 0.75,
|
| 1111 |
+
"rewards/chosen": -0.658935546875,
|
| 1112 |
+
"rewards/margins": 2.053906202316284,
|
| 1113 |
+
"rewards/rejected": -2.713671922683716,
|
| 1114 |
+
"step": 680
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.5271199388846448,
|
| 1118 |
+
"grad_norm": 86.2778829204208,
|
| 1119 |
+
"learning_rate": 2.631578947368421e-07,
|
| 1120 |
+
"logits/chosen": 0.11899413913488388,
|
| 1121 |
+
"logits/rejected": 0.15830078721046448,
|
| 1122 |
+
"logps/chosen": -528.4000244140625,
|
| 1123 |
+
"logps/rejected": -489.6000061035156,
|
| 1124 |
+
"loss": 0.4389,
|
| 1125 |
+
"nll_loss": 1.0031249523162842,
|
| 1126 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1127 |
+
"rewards/chosen": -0.705273449420929,
|
| 1128 |
+
"rewards/margins": 1.63671875,
|
| 1129 |
+
"rewards/rejected": -2.34375,
|
| 1130 |
+
"step": 690
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.5347593582887701,
|
| 1134 |
+
"grad_norm": 52.510026553930075,
|
| 1135 |
+
"learning_rate": 2.589134125636672e-07,
|
| 1136 |
+
"logits/chosen": 0.008349609561264515,
|
| 1137 |
+
"logits/rejected": -0.014941406436264515,
|
| 1138 |
+
"logps/chosen": -612.2000122070312,
|
| 1139 |
+
"logps/rejected": -568.0,
|
| 1140 |
+
"loss": 0.4444,
|
| 1141 |
+
"nll_loss": 1.1515624523162842,
|
| 1142 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1143 |
+
"rewards/chosen": -0.8998047113418579,
|
| 1144 |
+
"rewards/margins": 2.20703125,
|
| 1145 |
+
"rewards/rejected": -3.1070313453674316,
|
| 1146 |
+
"step": 700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5423987776928954,
|
| 1150 |
+
"grad_norm": 168.91459009375725,
|
| 1151 |
+
"learning_rate": 2.546689303904924e-07,
|
| 1152 |
+
"logits/chosen": 0.26335448026657104,
|
| 1153 |
+
"logits/rejected": 0.260498046875,
|
| 1154 |
+
"logps/chosen": -523.5999755859375,
|
| 1155 |
+
"logps/rejected": -479.0,
|
| 1156 |
+
"loss": 0.5232,
|
| 1157 |
+
"nll_loss": 0.9859374761581421,
|
| 1158 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1159 |
+
"rewards/chosen": -0.4925293028354645,
|
| 1160 |
+
"rewards/margins": 1.5232422351837158,
|
| 1161 |
+
"rewards/rejected": -2.0136961936950684,
|
| 1162 |
+
"step": 710
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.5500381970970206,
|
| 1166 |
+
"grad_norm": 42.12866881778182,
|
| 1167 |
+
"learning_rate": 2.5042444821731745e-07,
|
| 1168 |
+
"logits/chosen": 0.216796875,
|
| 1169 |
+
"logits/rejected": 0.07828368991613388,
|
| 1170 |
+
"logps/chosen": -497.3999938964844,
|
| 1171 |
+
"logps/rejected": -565.4000244140625,
|
| 1172 |
+
"loss": 0.49,
|
| 1173 |
+
"nll_loss": 0.9710937738418579,
|
| 1174 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1175 |
+
"rewards/chosen": -0.7671874761581421,
|
| 1176 |
+
"rewards/margins": 1.429101586341858,
|
| 1177 |
+
"rewards/rejected": -2.194531202316284,
|
| 1178 |
+
"step": 720
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.5576776165011459,
|
| 1182 |
+
"grad_norm": 66.13390214135931,
|
| 1183 |
+
"learning_rate": 2.4617996604414257e-07,
|
| 1184 |
+
"logits/chosen": 0.10423584282398224,
|
| 1185 |
+
"logits/rejected": 0.15535888075828552,
|
| 1186 |
+
"logps/chosen": -531.2000122070312,
|
| 1187 |
+
"logps/rejected": -528.0,
|
| 1188 |
+
"loss": 0.3866,
|
| 1189 |
+
"nll_loss": 0.9769531488418579,
|
| 1190 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1191 |
+
"rewards/chosen": -0.10312499850988388,
|
| 1192 |
+
"rewards/margins": 2.0078125,
|
| 1193 |
+
"rewards/rejected": -2.1107420921325684,
|
| 1194 |
+
"step": 730
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.5653170359052712,
|
| 1198 |
+
"grad_norm": 68.88567628228641,
|
| 1199 |
+
"learning_rate": 2.4193548387096775e-07,
|
| 1200 |
+
"logits/chosen": 0.02717285230755806,
|
| 1201 |
+
"logits/rejected": 0.1082763671875,
|
| 1202 |
+
"logps/chosen": -529.2000122070312,
|
| 1203 |
+
"logps/rejected": -530.0,
|
| 1204 |
+
"loss": 0.4087,
|
| 1205 |
+
"nll_loss": 1.119531273841858,
|
| 1206 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1207 |
+
"rewards/chosen": -0.271484375,
|
| 1208 |
+
"rewards/margins": 1.700781226158142,
|
| 1209 |
+
"rewards/rejected": -1.9748046398162842,
|
| 1210 |
+
"step": 740
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 0.5729564553093965,
|
| 1214 |
+
"grad_norm": 149.21207550792738,
|
| 1215 |
+
"learning_rate": 2.3769100169779285e-07,
|
| 1216 |
+
"logits/chosen": 0.2953124940395355,
|
| 1217 |
+
"logits/rejected": 0.20274658501148224,
|
| 1218 |
+
"logps/chosen": -615.4000244140625,
|
| 1219 |
+
"logps/rejected": -604.0,
|
| 1220 |
+
"loss": 0.4149,
|
| 1221 |
+
"nll_loss": 1.070703148841858,
|
| 1222 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1223 |
+
"rewards/chosen": 0.25048828125,
|
| 1224 |
+
"rewards/margins": 2.1714844703674316,
|
| 1225 |
+
"rewards/rejected": -1.924218773841858,
|
| 1226 |
+
"step": 750
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.5805958747135218,
|
| 1230 |
+
"grad_norm": 43.52746784376356,
|
| 1231 |
+
"learning_rate": 2.33446519524618e-07,
|
| 1232 |
+
"logits/chosen": 0.15043945610523224,
|
| 1233 |
+
"logits/rejected": 0.09915771335363388,
|
| 1234 |
+
"logps/chosen": -634.5999755859375,
|
| 1235 |
+
"logps/rejected": -594.0,
|
| 1236 |
+
"loss": 0.3219,
|
| 1237 |
+
"nll_loss": 1.0617187023162842,
|
| 1238 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1239 |
+
"rewards/chosen": -1.091552734375,
|
| 1240 |
+
"rewards/margins": 2.1484375,
|
| 1241 |
+
"rewards/rejected": -3.2367186546325684,
|
| 1242 |
+
"step": 760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.5882352941176471,
|
| 1246 |
+
"grad_norm": 78.35036463903903,
|
| 1247 |
+
"learning_rate": 2.2920203735144312e-07,
|
| 1248 |
+
"logits/chosen": 0.2001953125,
|
| 1249 |
+
"logits/rejected": 0.229248046875,
|
| 1250 |
+
"logps/chosen": -491.0,
|
| 1251 |
+
"logps/rejected": -530.7999877929688,
|
| 1252 |
+
"loss": 0.4278,
|
| 1253 |
+
"nll_loss": 0.9136718511581421,
|
| 1254 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1255 |
+
"rewards/chosen": -1.3425781726837158,
|
| 1256 |
+
"rewards/margins": 1.4572265148162842,
|
| 1257 |
+
"rewards/rejected": -2.7984375953674316,
|
| 1258 |
+
"step": 770
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.5958747135217723,
|
| 1262 |
+
"grad_norm": 63.09040192519054,
|
| 1263 |
+
"learning_rate": 2.2495755517826824e-07,
|
| 1264 |
+
"logits/chosen": 0.22727051377296448,
|
| 1265 |
+
"logits/rejected": 0.2230224609375,
|
| 1266 |
+
"logps/chosen": -537.4000244140625,
|
| 1267 |
+
"logps/rejected": -527.2000122070312,
|
| 1268 |
+
"loss": 0.3625,
|
| 1269 |
+
"nll_loss": 1.072656273841858,
|
| 1270 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1271 |
+
"rewards/chosen": -0.6168457269668579,
|
| 1272 |
+
"rewards/margins": 2.2718749046325684,
|
| 1273 |
+
"rewards/rejected": -2.887500047683716,
|
| 1274 |
+
"step": 780
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.6035141329258976,
|
| 1278 |
+
"grad_norm": 163.65178327754901,
|
| 1279 |
+
"learning_rate": 2.2071307300509337e-07,
|
| 1280 |
+
"logits/chosen": 0.06174926832318306,
|
| 1281 |
+
"logits/rejected": 0.24165038764476776,
|
| 1282 |
+
"logps/chosen": -536.0,
|
| 1283 |
+
"logps/rejected": -516.7999877929688,
|
| 1284 |
+
"loss": 0.4089,
|
| 1285 |
+
"nll_loss": 0.979687511920929,
|
| 1286 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1287 |
+
"rewards/chosen": -0.4857421815395355,
|
| 1288 |
+
"rewards/margins": 1.9304687976837158,
|
| 1289 |
+
"rewards/rejected": -2.4164061546325684,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.6111535523300229,
|
| 1294 |
+
"grad_norm": 88.08094851450458,
|
| 1295 |
+
"learning_rate": 2.164685908319185e-07,
|
| 1296 |
+
"logits/chosen": 0.10258789360523224,
|
| 1297 |
+
"logits/rejected": 0.24887695908546448,
|
| 1298 |
+
"logps/chosen": -622.7999877929688,
|
| 1299 |
+
"logps/rejected": -624.5999755859375,
|
| 1300 |
+
"loss": 0.4898,
|
| 1301 |
+
"nll_loss": 1.0183594226837158,
|
| 1302 |
+
"rewards/accuracies": 0.6875,
|
| 1303 |
+
"rewards/chosen": -1.33203125,
|
| 1304 |
+
"rewards/margins": 1.5109374523162842,
|
| 1305 |
+
"rewards/rejected": -2.839648485183716,
|
| 1306 |
+
"step": 800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.6187929717341482,
|
| 1310 |
+
"grad_norm": 61.96074875142302,
|
| 1311 |
+
"learning_rate": 2.1222410865874364e-07,
|
| 1312 |
+
"logits/chosen": 0.20058593153953552,
|
| 1313 |
+
"logits/rejected": 0.20753173530101776,
|
| 1314 |
+
"logps/chosen": -514.4000244140625,
|
| 1315 |
+
"logps/rejected": -482.79998779296875,
|
| 1316 |
+
"loss": 0.3643,
|
| 1317 |
+
"nll_loss": 0.992968738079071,
|
| 1318 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1319 |
+
"rewards/chosen": -2.125781297683716,
|
| 1320 |
+
"rewards/margins": 1.8703124523162842,
|
| 1321 |
+
"rewards/rejected": -3.9937500953674316,
|
| 1322 |
+
"step": 810
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.6264323911382735,
|
| 1326 |
+
"grad_norm": 78.46100711259405,
|
| 1327 |
+
"learning_rate": 2.0797962648556874e-07,
|
| 1328 |
+
"logits/chosen": 0.11856689304113388,
|
| 1329 |
+
"logits/rejected": 0.21796874701976776,
|
| 1330 |
+
"logps/chosen": -493.79998779296875,
|
| 1331 |
+
"logps/rejected": -470.0,
|
| 1332 |
+
"loss": 0.4282,
|
| 1333 |
+
"nll_loss": 0.8968750238418579,
|
| 1334 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1335 |
+
"rewards/chosen": -1.718164086341858,
|
| 1336 |
+
"rewards/margins": 1.7355468273162842,
|
| 1337 |
+
"rewards/rejected": -3.450976610183716,
|
| 1338 |
+
"step": 820
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 0.6340718105423988,
|
| 1342 |
+
"grad_norm": 77.73118393844226,
|
| 1343 |
+
"learning_rate": 2.037351443123939e-07,
|
| 1344 |
+
"logits/chosen": 0.20204468071460724,
|
| 1345 |
+
"logits/rejected": 0.2997070252895355,
|
| 1346 |
+
"logps/chosen": -508.20001220703125,
|
| 1347 |
+
"logps/rejected": -503.79998779296875,
|
| 1348 |
+
"loss": 0.4684,
|
| 1349 |
+
"nll_loss": 0.9378906488418579,
|
| 1350 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1351 |
+
"rewards/chosen": -2.0494141578674316,
|
| 1352 |
+
"rewards/margins": 1.7734375,
|
| 1353 |
+
"rewards/rejected": -3.8218750953674316,
|
| 1354 |
+
"step": 830
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.6417112299465241,
|
| 1358 |
+
"grad_norm": 51.36288875319757,
|
| 1359 |
+
"learning_rate": 1.99490662139219e-07,
|
| 1360 |
+
"logits/chosen": 0.08984375,
|
| 1361 |
+
"logits/rejected": 0.26904296875,
|
| 1362 |
+
"logps/chosen": -532.5999755859375,
|
| 1363 |
+
"logps/rejected": -528.2000122070312,
|
| 1364 |
+
"loss": 0.39,
|
| 1365 |
+
"nll_loss": 1.1417968273162842,
|
| 1366 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1367 |
+
"rewards/chosen": -1.943359375,
|
| 1368 |
+
"rewards/margins": 2.007031202316284,
|
| 1369 |
+
"rewards/rejected": -3.950000047683716,
|
| 1370 |
+
"step": 840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.6493506493506493,
|
| 1374 |
+
"grad_norm": 62.89771240222646,
|
| 1375 |
+
"learning_rate": 1.9524617996604413e-07,
|
| 1376 |
+
"logits/chosen": 0.18178710341453552,
|
| 1377 |
+
"logits/rejected": 0.3423828184604645,
|
| 1378 |
+
"logps/chosen": -577.2000122070312,
|
| 1379 |
+
"logps/rejected": -608.0,
|
| 1380 |
+
"loss": 0.351,
|
| 1381 |
+
"nll_loss": 1.049218773841858,
|
| 1382 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1383 |
+
"rewards/chosen": -1.882226586341858,
|
| 1384 |
+
"rewards/margins": 1.9597656726837158,
|
| 1385 |
+
"rewards/rejected": -3.8382811546325684,
|
| 1386 |
+
"step": 850
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 0.6569900687547746,
|
| 1390 |
+
"grad_norm": 168.00220492062218,
|
| 1391 |
+
"learning_rate": 1.9100169779286926e-07,
|
| 1392 |
+
"logits/chosen": 0.10942383110523224,
|
| 1393 |
+
"logits/rejected": 0.13002929091453552,
|
| 1394 |
+
"logps/chosen": -529.5999755859375,
|
| 1395 |
+
"logps/rejected": -466.79998779296875,
|
| 1396 |
+
"loss": 0.4862,
|
| 1397 |
+
"nll_loss": 1.0945312976837158,
|
| 1398 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1399 |
+
"rewards/chosen": -1.9791991710662842,
|
| 1400 |
+
"rewards/margins": 1.7999999523162842,
|
| 1401 |
+
"rewards/rejected": -3.78125,
|
| 1402 |
+
"step": 860
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.6646294881588999,
|
| 1406 |
+
"grad_norm": 59.78300945522556,
|
| 1407 |
+
"learning_rate": 1.867572156196944e-07,
|
| 1408 |
+
"logits/chosen": -0.00019531250291038305,
|
| 1409 |
+
"logits/rejected": 0.16667480766773224,
|
| 1410 |
+
"logps/chosen": -494.3999938964844,
|
| 1411 |
+
"logps/rejected": -466.6000061035156,
|
| 1412 |
+
"loss": 0.4736,
|
| 1413 |
+
"nll_loss": 1.0222656726837158,
|
| 1414 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1415 |
+
"rewards/chosen": -1.707128882408142,
|
| 1416 |
+
"rewards/margins": 1.8664062023162842,
|
| 1417 |
+
"rewards/rejected": -3.5796875953674316,
|
| 1418 |
+
"step": 870
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.6722689075630253,
|
| 1422 |
+
"grad_norm": 71.0629204613292,
|
| 1423 |
+
"learning_rate": 1.825127334465195e-07,
|
| 1424 |
+
"logits/chosen": 0.37675780057907104,
|
| 1425 |
+
"logits/rejected": 0.5088866949081421,
|
| 1426 |
+
"logps/chosen": -544.4000244140625,
|
| 1427 |
+
"logps/rejected": -515.2000122070312,
|
| 1428 |
+
"loss": 0.4535,
|
| 1429 |
+
"nll_loss": 0.9624999761581421,
|
| 1430 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1431 |
+
"rewards/chosen": -1.88671875,
|
| 1432 |
+
"rewards/margins": 1.910546898841858,
|
| 1433 |
+
"rewards/rejected": -3.7984375953674316,
|
| 1434 |
+
"step": 880
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.6799083269671505,
|
| 1438 |
+
"grad_norm": 105.14381586804791,
|
| 1439 |
+
"learning_rate": 1.7826825127334463e-07,
|
| 1440 |
+
"logits/chosen": 0.4029541015625,
|
| 1441 |
+
"logits/rejected": 0.09556885063648224,
|
| 1442 |
+
"logps/chosen": -593.5999755859375,
|
| 1443 |
+
"logps/rejected": -555.7999877929688,
|
| 1444 |
+
"loss": 0.4513,
|
| 1445 |
+
"nll_loss": 1.1453125476837158,
|
| 1446 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1447 |
+
"rewards/chosen": -1.828125,
|
| 1448 |
+
"rewards/margins": 2.0414061546325684,
|
| 1449 |
+
"rewards/rejected": -3.8687500953674316,
|
| 1450 |
+
"step": 890
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.6875477463712758,
|
| 1454 |
+
"grad_norm": 92.44231400598672,
|
| 1455 |
+
"learning_rate": 1.7402376910016978e-07,
|
| 1456 |
+
"logits/chosen": 0.2739501893520355,
|
| 1457 |
+
"logits/rejected": 0.19692382216453552,
|
| 1458 |
+
"logps/chosen": -590.7999877929688,
|
| 1459 |
+
"logps/rejected": -493.79998779296875,
|
| 1460 |
+
"loss": 0.3861,
|
| 1461 |
+
"nll_loss": 1.0773437023162842,
|
| 1462 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1463 |
+
"rewards/chosen": -1.638769507408142,
|
| 1464 |
+
"rewards/margins": 1.8623046875,
|
| 1465 |
+
"rewards/rejected": -3.5023436546325684,
|
| 1466 |
+
"step": 900
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.6951871657754011,
|
| 1470 |
+
"grad_norm": 51.021124930196535,
|
| 1471 |
+
"learning_rate": 1.697792869269949e-07,
|
| 1472 |
+
"logits/chosen": 0.13515624403953552,
|
| 1473 |
+
"logits/rejected": 0.11835937201976776,
|
| 1474 |
+
"logps/chosen": -695.5999755859375,
|
| 1475 |
+
"logps/rejected": -717.5999755859375,
|
| 1476 |
+
"loss": 0.4679,
|
| 1477 |
+
"nll_loss": 0.9722656011581421,
|
| 1478 |
+
"rewards/accuracies": 0.6875,
|
| 1479 |
+
"rewards/chosen": -1.2556641101837158,
|
| 1480 |
+
"rewards/margins": 1.419921875,
|
| 1481 |
+
"rewards/rejected": -2.673046827316284,
|
| 1482 |
+
"step": 910
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.7028265851795263,
|
| 1486 |
+
"grad_norm": 106.63758032816845,
|
| 1487 |
+
"learning_rate": 1.6553480475382003e-07,
|
| 1488 |
+
"logits/chosen": 0.11247558891773224,
|
| 1489 |
+
"logits/rejected": 0.16817016899585724,
|
| 1490 |
+
"logps/chosen": -503.20001220703125,
|
| 1491 |
+
"logps/rejected": -501.79998779296875,
|
| 1492 |
+
"loss": 0.516,
|
| 1493 |
+
"nll_loss": 1.0261719226837158,
|
| 1494 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1495 |
+
"rewards/chosen": -1.6069824695587158,
|
| 1496 |
+
"rewards/margins": 1.6521484851837158,
|
| 1497 |
+
"rewards/rejected": -3.2593750953674316,
|
| 1498 |
+
"step": 920
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 0.7104660045836516,
|
| 1502 |
+
"grad_norm": 41.82251046800636,
|
| 1503 |
+
"learning_rate": 1.6129032258064515e-07,
|
| 1504 |
+
"logits/chosen": 0.33271485567092896,
|
| 1505 |
+
"logits/rejected": 0.30549317598342896,
|
| 1506 |
+
"logps/chosen": -570.0,
|
| 1507 |
+
"logps/rejected": -557.2000122070312,
|
| 1508 |
+
"loss": 0.4462,
|
| 1509 |
+
"nll_loss": 0.9378906488418579,
|
| 1510 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1511 |
+
"rewards/chosen": -1.86328125,
|
| 1512 |
+
"rewards/margins": 1.784765601158142,
|
| 1513 |
+
"rewards/rejected": -3.6500000953674316,
|
| 1514 |
+
"step": 930
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.7181054239877769,
|
| 1518 |
+
"grad_norm": 59.265277634488925,
|
| 1519 |
+
"learning_rate": 1.570458404074703e-07,
|
| 1520 |
+
"logits/chosen": 0.16783447563648224,
|
| 1521 |
+
"logits/rejected": 0.21890869736671448,
|
| 1522 |
+
"logps/chosen": -509.6000061035156,
|
| 1523 |
+
"logps/rejected": -499.3999938964844,
|
| 1524 |
+
"loss": 0.4888,
|
| 1525 |
+
"nll_loss": 1.064453125,
|
| 1526 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1527 |
+
"rewards/chosen": -1.707421898841858,
|
| 1528 |
+
"rewards/margins": 1.84375,
|
| 1529 |
+
"rewards/rejected": -3.553906202316284,
|
| 1530 |
+
"step": 940
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.7257448433919023,
|
| 1534 |
+
"grad_norm": 77.92638263470651,
|
| 1535 |
+
"learning_rate": 1.528013582342954e-07,
|
| 1536 |
+
"logits/chosen": 0.24189452826976776,
|
| 1537 |
+
"logits/rejected": 0.16245117783546448,
|
| 1538 |
+
"logps/chosen": -618.0,
|
| 1539 |
+
"logps/rejected": -578.7999877929688,
|
| 1540 |
+
"loss": 0.4762,
|
| 1541 |
+
"nll_loss": 1.064062476158142,
|
| 1542 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1543 |
+
"rewards/chosen": -1.4609375,
|
| 1544 |
+
"rewards/margins": 1.718359351158142,
|
| 1545 |
+
"rewards/rejected": -3.1812500953674316,
|
| 1546 |
+
"step": 950
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 0.7333842627960275,
|
| 1550 |
+
"grad_norm": 175.06103048761182,
|
| 1551 |
+
"learning_rate": 1.4855687606112052e-07,
|
| 1552 |
+
"logits/chosen": 0.02786865271627903,
|
| 1553 |
+
"logits/rejected": 0.24497070908546448,
|
| 1554 |
+
"logps/chosen": -573.2000122070312,
|
| 1555 |
+
"logps/rejected": -503.6000061035156,
|
| 1556 |
+
"loss": 0.4024,
|
| 1557 |
+
"nll_loss": 1.069921851158142,
|
| 1558 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1559 |
+
"rewards/chosen": -2.059985399246216,
|
| 1560 |
+
"rewards/margins": 2.0023436546325684,
|
| 1561 |
+
"rewards/rejected": -4.0625,
|
| 1562 |
+
"step": 960
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.7410236822001528,
|
| 1566 |
+
"grad_norm": 66.36991524452817,
|
| 1567 |
+
"learning_rate": 1.4431239388794567e-07,
|
| 1568 |
+
"logits/chosen": 0.15531310439109802,
|
| 1569 |
+
"logits/rejected": 0.35588377714157104,
|
| 1570 |
+
"logps/chosen": -488.79998779296875,
|
| 1571 |
+
"logps/rejected": -478.6000061035156,
|
| 1572 |
+
"loss": 0.3916,
|
| 1573 |
+
"nll_loss": 1.0046875476837158,
|
| 1574 |
+
"rewards/accuracies": 0.75,
|
| 1575 |
+
"rewards/chosen": -1.4480469226837158,
|
| 1576 |
+
"rewards/margins": 1.8976562023162842,
|
| 1577 |
+
"rewards/rejected": -3.3466796875,
|
| 1578 |
+
"step": 970
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.7486631016042781,
|
| 1582 |
+
"grad_norm": 104.2638833576658,
|
| 1583 |
+
"learning_rate": 1.400679117147708e-07,
|
| 1584 |
+
"logits/chosen": 0.02060546912252903,
|
| 1585 |
+
"logits/rejected": 0.23635253310203552,
|
| 1586 |
+
"logps/chosen": -522.4000244140625,
|
| 1587 |
+
"logps/rejected": -455.3999938964844,
|
| 1588 |
+
"loss": 0.4204,
|
| 1589 |
+
"nll_loss": 1.078515648841858,
|
| 1590 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1591 |
+
"rewards/chosen": -1.4582030773162842,
|
| 1592 |
+
"rewards/margins": 1.611718773841858,
|
| 1593 |
+
"rewards/rejected": -3.071093797683716,
|
| 1594 |
+
"step": 980
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.7563025210084033,
|
| 1598 |
+
"grad_norm": 89.84520085486623,
|
| 1599 |
+
"learning_rate": 1.3582342954159592e-07,
|
| 1600 |
+
"logits/chosen": 0.1976318359375,
|
| 1601 |
+
"logits/rejected": 0.23613281548023224,
|
| 1602 |
+
"logps/chosen": -619.2000122070312,
|
| 1603 |
+
"logps/rejected": -586.2000122070312,
|
| 1604 |
+
"loss": 0.409,
|
| 1605 |
+
"nll_loss": 1.031640648841858,
|
| 1606 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1607 |
+
"rewards/chosen": -1.990625023841858,
|
| 1608 |
+
"rewards/margins": 1.865625023841858,
|
| 1609 |
+
"rewards/rejected": -3.854687452316284,
|
| 1610 |
+
"step": 990
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.7639419404125286,
|
| 1614 |
+
"grad_norm": 153.33070702585832,
|
| 1615 |
+
"learning_rate": 1.3157894736842104e-07,
|
| 1616 |
+
"logits/chosen": 0.16008910536766052,
|
| 1617 |
+
"logits/rejected": 0.16474609076976776,
|
| 1618 |
+
"logps/chosen": -559.4000244140625,
|
| 1619 |
+
"logps/rejected": -610.2000122070312,
|
| 1620 |
+
"loss": 0.4936,
|
| 1621 |
+
"nll_loss": 1.001953125,
|
| 1622 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1623 |
+
"rewards/chosen": -1.138281226158142,
|
| 1624 |
+
"rewards/margins": 1.834375023841858,
|
| 1625 |
+
"rewards/rejected": -2.973437547683716,
|
| 1626 |
+
"step": 1000
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.771581359816654,
|
| 1630 |
+
"grad_norm": 54.641803489764264,
|
| 1631 |
+
"learning_rate": 1.273344651952462e-07,
|
| 1632 |
+
"logits/chosen": 0.2105712890625,
|
| 1633 |
+
"logits/rejected": 0.2726074159145355,
|
| 1634 |
+
"logps/chosen": -532.5999755859375,
|
| 1635 |
+
"logps/rejected": -532.4000244140625,
|
| 1636 |
+
"loss": 0.4568,
|
| 1637 |
+
"nll_loss": 0.927734375,
|
| 1638 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1639 |
+
"rewards/chosen": -1.4357421398162842,
|
| 1640 |
+
"rewards/margins": 1.740625023841858,
|
| 1641 |
+
"rewards/rejected": -3.1781249046325684,
|
| 1642 |
+
"step": 1010
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.7792207792207793,
|
| 1646 |
+
"grad_norm": 76.14721419271173,
|
| 1647 |
+
"learning_rate": 1.2308998302207129e-07,
|
| 1648 |
+
"logits/chosen": -0.04191894456744194,
|
| 1649 |
+
"logits/rejected": 0.11892089992761612,
|
| 1650 |
+
"logps/chosen": -622.2000122070312,
|
| 1651 |
+
"logps/rejected": -604.5999755859375,
|
| 1652 |
+
"loss": 0.5112,
|
| 1653 |
+
"nll_loss": 1.1613280773162842,
|
| 1654 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1655 |
+
"rewards/chosen": -1.048437476158142,
|
| 1656 |
+
"rewards/margins": 1.73828125,
|
| 1657 |
+
"rewards/rejected": -2.787109375,
|
| 1658 |
+
"step": 1020
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.7868601986249045,
|
| 1662 |
+
"grad_norm": 46.81095324265018,
|
| 1663 |
+
"learning_rate": 1.1884550084889642e-07,
|
| 1664 |
+
"logits/chosen": 0.12519530951976776,
|
| 1665 |
+
"logits/rejected": 0.3093627989292145,
|
| 1666 |
+
"logps/chosen": -590.2000122070312,
|
| 1667 |
+
"logps/rejected": -565.5999755859375,
|
| 1668 |
+
"loss": 0.3614,
|
| 1669 |
+
"nll_loss": 1.192968726158142,
|
| 1670 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1671 |
+
"rewards/chosen": -1.818359375,
|
| 1672 |
+
"rewards/margins": 1.951562523841858,
|
| 1673 |
+
"rewards/rejected": -3.768749952316284,
|
| 1674 |
+
"step": 1030
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.7944996180290298,
|
| 1678 |
+
"grad_norm": 188.1614298163614,
|
| 1679 |
+
"learning_rate": 1.1460101867572156e-07,
|
| 1680 |
+
"logits/chosen": 0.06070556491613388,
|
| 1681 |
+
"logits/rejected": 0.09006347507238388,
|
| 1682 |
+
"logps/chosen": -581.7999877929688,
|
| 1683 |
+
"logps/rejected": -629.2000122070312,
|
| 1684 |
+
"loss": 0.3791,
|
| 1685 |
+
"nll_loss": 0.9515625238418579,
|
| 1686 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1687 |
+
"rewards/chosen": -2.4390625953674316,
|
| 1688 |
+
"rewards/margins": 1.7605469226837158,
|
| 1689 |
+
"rewards/rejected": -4.199999809265137,
|
| 1690 |
+
"step": 1040
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.8021390374331551,
|
| 1694 |
+
"grad_norm": 46.116263610995595,
|
| 1695 |
+
"learning_rate": 1.1035653650254668e-07,
|
| 1696 |
+
"logits/chosen": 0.04994506761431694,
|
| 1697 |
+
"logits/rejected": 0.04636230319738388,
|
| 1698 |
+
"logps/chosen": -505.0,
|
| 1699 |
+
"logps/rejected": -514.2000122070312,
|
| 1700 |
+
"loss": 0.4822,
|
| 1701 |
+
"nll_loss": 0.950390636920929,
|
| 1702 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1703 |
+
"rewards/chosen": -1.1416015625,
|
| 1704 |
+
"rewards/margins": 1.7126953601837158,
|
| 1705 |
+
"rewards/rejected": -2.8589844703674316,
|
| 1706 |
+
"step": 1050
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.8097784568372803,
|
| 1710 |
+
"grad_norm": 27.441756440042443,
|
| 1711 |
+
"learning_rate": 1.0611205432937182e-07,
|
| 1712 |
+
"logits/chosen": 0.181640625,
|
| 1713 |
+
"logits/rejected": 0.19467774033546448,
|
| 1714 |
+
"logps/chosen": -621.5999755859375,
|
| 1715 |
+
"logps/rejected": -601.7999877929688,
|
| 1716 |
+
"loss": 0.314,
|
| 1717 |
+
"nll_loss": 0.958984375,
|
| 1718 |
+
"rewards/accuracies": 0.8125,
|
| 1719 |
+
"rewards/chosen": -1.361718773841858,
|
| 1720 |
+
"rewards/margins": 2.022656202316284,
|
| 1721 |
+
"rewards/rejected": -3.38671875,
|
| 1722 |
+
"step": 1060
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.8174178762414056,
|
| 1726 |
+
"grad_norm": 30.958123881153575,
|
| 1727 |
+
"learning_rate": 1.0186757215619694e-07,
|
| 1728 |
+
"logits/chosen": 0.17719726264476776,
|
| 1729 |
+
"logits/rejected": 0.2730468809604645,
|
| 1730 |
+
"logps/chosen": -548.2000122070312,
|
| 1731 |
+
"logps/rejected": -525.2000122070312,
|
| 1732 |
+
"loss": 0.4084,
|
| 1733 |
+
"nll_loss": 0.974609375,
|
| 1734 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1735 |
+
"rewards/chosen": -1.289892554283142,
|
| 1736 |
+
"rewards/margins": 1.989843726158142,
|
| 1737 |
+
"rewards/rejected": -3.28515625,
|
| 1738 |
+
"step": 1070
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.825057295645531,
|
| 1742 |
+
"grad_norm": 46.51626516681703,
|
| 1743 |
+
"learning_rate": 9.762308998302207e-08,
|
| 1744 |
+
"logits/chosen": 0.1400146484375,
|
| 1745 |
+
"logits/rejected": 0.05910644680261612,
|
| 1746 |
+
"logps/chosen": -524.5999755859375,
|
| 1747 |
+
"logps/rejected": -512.5999755859375,
|
| 1748 |
+
"loss": 0.3726,
|
| 1749 |
+
"nll_loss": 1.006250023841858,
|
| 1750 |
+
"rewards/accuracies": 0.75,
|
| 1751 |
+
"rewards/chosen": -1.084082007408142,
|
| 1752 |
+
"rewards/margins": 1.799218773841858,
|
| 1753 |
+
"rewards/rejected": -2.879687547683716,
|
| 1754 |
+
"step": 1080
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.8326967150496563,
|
| 1758 |
+
"grad_norm": 34.1039818844264,
|
| 1759 |
+
"learning_rate": 9.33786078098472e-08,
|
| 1760 |
+
"logits/chosen": 0.08969726413488388,
|
| 1761 |
+
"logits/rejected": 0.2244873046875,
|
| 1762 |
+
"logps/chosen": -499.79998779296875,
|
| 1763 |
+
"logps/rejected": -485.0,
|
| 1764 |
+
"loss": 0.4221,
|
| 1765 |
+
"nll_loss": 0.979296863079071,
|
| 1766 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1767 |
+
"rewards/chosen": -1.2386353015899658,
|
| 1768 |
+
"rewards/margins": 1.769921898841858,
|
| 1769 |
+
"rewards/rejected": -3.0078125,
|
| 1770 |
+
"step": 1090
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.8403361344537815,
|
| 1774 |
+
"grad_norm": 44.51111317109547,
|
| 1775 |
+
"learning_rate": 8.913412563667231e-08,
|
| 1776 |
+
"logits/chosen": 0.05061035230755806,
|
| 1777 |
+
"logits/rejected": -0.0087890625,
|
| 1778 |
+
"logps/chosen": -558.0,
|
| 1779 |
+
"logps/rejected": -569.5999755859375,
|
| 1780 |
+
"loss": 0.4065,
|
| 1781 |
+
"nll_loss": 1.037500023841858,
|
| 1782 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1783 |
+
"rewards/chosen": -1.33984375,
|
| 1784 |
+
"rewards/margins": 1.8230469226837158,
|
| 1785 |
+
"rewards/rejected": -3.1656250953674316,
|
| 1786 |
+
"step": 1100
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.8479755538579068,
|
| 1790 |
+
"grad_norm": 127.58455214847397,
|
| 1791 |
+
"learning_rate": 8.488964346349745e-08,
|
| 1792 |
+
"logits/chosen": 0.427734375,
|
| 1793 |
+
"logits/rejected": 0.2650390565395355,
|
| 1794 |
+
"logps/chosen": -515.7999877929688,
|
| 1795 |
+
"logps/rejected": -480.0,
|
| 1796 |
+
"loss": 0.3811,
|
| 1797 |
+
"nll_loss": 1.0066406726837158,
|
| 1798 |
+
"rewards/accuracies": 0.75,
|
| 1799 |
+
"rewards/chosen": -1.5940430164337158,
|
| 1800 |
+
"rewards/margins": 1.8195312023162842,
|
| 1801 |
+
"rewards/rejected": -3.4124999046325684,
|
| 1802 |
+
"step": 1110
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.8556149732620321,
|
| 1806 |
+
"grad_norm": 84.1325471209257,
|
| 1807 |
+
"learning_rate": 8.064516129032257e-08,
|
| 1808 |
+
"logits/chosen": 0.04128418117761612,
|
| 1809 |
+
"logits/rejected": 0.15388183295726776,
|
| 1810 |
+
"logps/chosen": -484.3999938964844,
|
| 1811 |
+
"logps/rejected": -555.5999755859375,
|
| 1812 |
+
"loss": 0.4469,
|
| 1813 |
+
"nll_loss": 0.9390624761581421,
|
| 1814 |
+
"rewards/accuracies": 0.75,
|
| 1815 |
+
"rewards/chosen": -1.283789038658142,
|
| 1816 |
+
"rewards/margins": 1.693750023841858,
|
| 1817 |
+
"rewards/rejected": -2.9800782203674316,
|
| 1818 |
+
"step": 1120
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.8632543926661573,
|
| 1822 |
+
"grad_norm": 126.51132512166387,
|
| 1823 |
+
"learning_rate": 7.64006791171477e-08,
|
| 1824 |
+
"logits/chosen": 0.2769531309604645,
|
| 1825 |
+
"logits/rejected": 0.24785156548023224,
|
| 1826 |
+
"logps/chosen": -508.20001220703125,
|
| 1827 |
+
"logps/rejected": -510.20001220703125,
|
| 1828 |
+
"loss": 0.4501,
|
| 1829 |
+
"nll_loss": 0.952343761920929,
|
| 1830 |
+
"rewards/accuracies": 0.6875,
|
| 1831 |
+
"rewards/chosen": -1.070703148841858,
|
| 1832 |
+
"rewards/margins": 1.8757812976837158,
|
| 1833 |
+
"rewards/rejected": -2.9507813453674316,
|
| 1834 |
+
"step": 1130
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.8708938120702827,
|
| 1838 |
+
"grad_norm": 68.0062602486858,
|
| 1839 |
+
"learning_rate": 7.215619694397283e-08,
|
| 1840 |
+
"logits/chosen": 0.20705565810203552,
|
| 1841 |
+
"logits/rejected": 0.12968750298023224,
|
| 1842 |
+
"logps/chosen": -541.0,
|
| 1843 |
+
"logps/rejected": -518.2000122070312,
|
| 1844 |
+
"loss": 0.4509,
|
| 1845 |
+
"nll_loss": 0.9488281011581421,
|
| 1846 |
+
"rewards/accuracies": 0.6875,
|
| 1847 |
+
"rewards/chosen": -1.046484351158142,
|
| 1848 |
+
"rewards/margins": 1.5662109851837158,
|
| 1849 |
+
"rewards/rejected": -2.612499952316284,
|
| 1850 |
+
"step": 1140
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.878533231474408,
|
| 1854 |
+
"grad_norm": 56.86465687647158,
|
| 1855 |
+
"learning_rate": 6.791171477079796e-08,
|
| 1856 |
+
"logits/chosen": 0.10419921576976776,
|
| 1857 |
+
"logits/rejected": 0.10698242485523224,
|
| 1858 |
+
"logps/chosen": -586.5999755859375,
|
| 1859 |
+
"logps/rejected": -557.7999877929688,
|
| 1860 |
+
"loss": 0.4839,
|
| 1861 |
+
"nll_loss": 1.083984375,
|
| 1862 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1863 |
+
"rewards/chosen": -0.5614258050918579,
|
| 1864 |
+
"rewards/margins": 1.4835937023162842,
|
| 1865 |
+
"rewards/rejected": -2.046093702316284,
|
| 1866 |
+
"step": 1150
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.8861726508785333,
|
| 1870 |
+
"grad_norm": 249.55831019629667,
|
| 1871 |
+
"learning_rate": 6.36672325976231e-08,
|
| 1872 |
+
"logits/chosen": 0.06125488132238388,
|
| 1873 |
+
"logits/rejected": 0.20297852158546448,
|
| 1874 |
+
"logps/chosen": -528.7999877929688,
|
| 1875 |
+
"logps/rejected": -527.5999755859375,
|
| 1876 |
+
"loss": 0.4153,
|
| 1877 |
+
"nll_loss": 1.016015648841858,
|
| 1878 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1879 |
+
"rewards/chosen": -1.149658203125,
|
| 1880 |
+
"rewards/margins": 1.703125,
|
| 1881 |
+
"rewards/rejected": -2.854296922683716,
|
| 1882 |
+
"step": 1160
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.8938120702826585,
|
| 1886 |
+
"grad_norm": 56.85991905712662,
|
| 1887 |
+
"learning_rate": 5.942275042444821e-08,
|
| 1888 |
+
"logits/chosen": 0.14946289360523224,
|
| 1889 |
+
"logits/rejected": 0.13559570908546448,
|
| 1890 |
+
"logps/chosen": -523.4000244140625,
|
| 1891 |
+
"logps/rejected": -499.3999938964844,
|
| 1892 |
+
"loss": 0.4354,
|
| 1893 |
+
"nll_loss": 0.923046886920929,
|
| 1894 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1895 |
+
"rewards/chosen": -1.0758545398712158,
|
| 1896 |
+
"rewards/margins": 1.5320312976837158,
|
| 1897 |
+
"rewards/rejected": -2.611132860183716,
|
| 1898 |
+
"step": 1170
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.9014514896867838,
|
| 1902 |
+
"grad_norm": 42.322112632764245,
|
| 1903 |
+
"learning_rate": 5.517826825127334e-08,
|
| 1904 |
+
"logits/chosen": 0.15610352158546448,
|
| 1905 |
+
"logits/rejected": 0.15117187798023224,
|
| 1906 |
+
"logps/chosen": -562.5999755859375,
|
| 1907 |
+
"logps/rejected": -543.2000122070312,
|
| 1908 |
+
"loss": 0.4079,
|
| 1909 |
+
"nll_loss": 1.0734374523162842,
|
| 1910 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1911 |
+
"rewards/chosen": -1.0110352039337158,
|
| 1912 |
+
"rewards/margins": 1.935156226158142,
|
| 1913 |
+
"rewards/rejected": -2.9496092796325684,
|
| 1914 |
+
"step": 1180
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.9090909090909091,
|
| 1918 |
+
"grad_norm": 92.98578925608547,
|
| 1919 |
+
"learning_rate": 5.093378607809847e-08,
|
| 1920 |
+
"logits/chosen": 0.04831542819738388,
|
| 1921 |
+
"logits/rejected": 0.24643555283546448,
|
| 1922 |
+
"logps/chosen": -521.4000244140625,
|
| 1923 |
+
"logps/rejected": -481.3999938964844,
|
| 1924 |
+
"loss": 0.4018,
|
| 1925 |
+
"nll_loss": 1.131250023841858,
|
| 1926 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1927 |
+
"rewards/chosen": -0.9009765386581421,
|
| 1928 |
+
"rewards/margins": 2.0601563453674316,
|
| 1929 |
+
"rewards/rejected": -2.9625000953674316,
|
| 1930 |
+
"step": 1190
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.9167303284950343,
|
| 1934 |
+
"grad_norm": 80.1548091064356,
|
| 1935 |
+
"learning_rate": 4.66893039049236e-08,
|
| 1936 |
+
"logits/chosen": 0.18161621689796448,
|
| 1937 |
+
"logits/rejected": 0.4056152403354645,
|
| 1938 |
+
"logps/chosen": -504.0,
|
| 1939 |
+
"logps/rejected": -524.4000244140625,
|
| 1940 |
+
"loss": 0.437,
|
| 1941 |
+
"nll_loss": 0.9574218988418579,
|
| 1942 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1943 |
+
"rewards/chosen": -1.600195288658142,
|
| 1944 |
+
"rewards/margins": 1.701562523841858,
|
| 1945 |
+
"rewards/rejected": -3.2984375953674316,
|
| 1946 |
+
"step": 1200
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.9243697478991597,
|
| 1950 |
+
"grad_norm": 104.97014790303885,
|
| 1951 |
+
"learning_rate": 4.2444821731748725e-08,
|
| 1952 |
+
"logits/chosen": 0.10756836086511612,
|
| 1953 |
+
"logits/rejected": 0.17495116591453552,
|
| 1954 |
+
"logps/chosen": -541.4000244140625,
|
| 1955 |
+
"logps/rejected": -549.5999755859375,
|
| 1956 |
+
"loss": 0.4187,
|
| 1957 |
+
"nll_loss": 1.021093726158142,
|
| 1958 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1959 |
+
"rewards/chosen": -0.9234374761581421,
|
| 1960 |
+
"rewards/margins": 2.3499999046325684,
|
| 1961 |
+
"rewards/rejected": -3.2718749046325684,
|
| 1962 |
+
"step": 1210
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.932009167303285,
|
| 1966 |
+
"grad_norm": 79.03828937393179,
|
| 1967 |
+
"learning_rate": 3.820033955857385e-08,
|
| 1968 |
+
"logits/chosen": 0.04698486253619194,
|
| 1969 |
+
"logits/rejected": 0.06538085639476776,
|
| 1970 |
+
"logps/chosen": -618.4000244140625,
|
| 1971 |
+
"logps/rejected": -604.0,
|
| 1972 |
+
"loss": 0.4579,
|
| 1973 |
+
"nll_loss": 1.101171851158142,
|
| 1974 |
+
"rewards/accuracies": 0.6875,
|
| 1975 |
+
"rewards/chosen": -1.149804711341858,
|
| 1976 |
+
"rewards/margins": 1.558984398841858,
|
| 1977 |
+
"rewards/rejected": -2.710156202316284,
|
| 1978 |
+
"step": 1220
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.9396485867074102,
|
| 1982 |
+
"grad_norm": 49.18189678080444,
|
| 1983 |
+
"learning_rate": 3.395585738539898e-08,
|
| 1984 |
+
"logits/chosen": 0.11579589545726776,
|
| 1985 |
+
"logits/rejected": 0.30366212129592896,
|
| 1986 |
+
"logps/chosen": -533.0,
|
| 1987 |
+
"logps/rejected": -537.0,
|
| 1988 |
+
"loss": 0.2985,
|
| 1989 |
+
"nll_loss": 1.0769531726837158,
|
| 1990 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1991 |
+
"rewards/chosen": -1.183935523033142,
|
| 1992 |
+
"rewards/margins": 2.2054686546325684,
|
| 1993 |
+
"rewards/rejected": -3.38671875,
|
| 1994 |
+
"step": 1230
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.9472880061115355,
|
| 1998 |
+
"grad_norm": 65.08543589597988,
|
| 1999 |
+
"learning_rate": 2.9711375212224106e-08,
|
| 2000 |
+
"logits/chosen": 0.05819091945886612,
|
| 2001 |
+
"logits/rejected": 0.17856445908546448,
|
| 2002 |
+
"logps/chosen": -529.2000122070312,
|
| 2003 |
+
"logps/rejected": -525.4000244140625,
|
| 2004 |
+
"loss": 0.2858,
|
| 2005 |
+
"nll_loss": 1.075781226158142,
|
| 2006 |
+
"rewards/accuracies": 0.824999988079071,
|
| 2007 |
+
"rewards/chosen": -1.2798340320587158,
|
| 2008 |
+
"rewards/margins": 2.4281249046325684,
|
| 2009 |
+
"rewards/rejected": -3.7125000953674316,
|
| 2010 |
+
"step": 1240
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 0.9549274255156608,
|
| 2014 |
+
"grad_norm": 113.15137128253427,
|
| 2015 |
+
"learning_rate": 2.5466893039049236e-08,
|
| 2016 |
+
"logits/chosen": 0.19570311903953552,
|
| 2017 |
+
"logits/rejected": 0.31059569120407104,
|
| 2018 |
+
"logps/chosen": -523.4000244140625,
|
| 2019 |
+
"logps/rejected": -506.6000061035156,
|
| 2020 |
+
"loss": 0.4007,
|
| 2021 |
+
"nll_loss": 0.979296863079071,
|
| 2022 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2023 |
+
"rewards/chosen": -1.1939697265625,
|
| 2024 |
+
"rewards/margins": 1.759765625,
|
| 2025 |
+
"rewards/rejected": -2.9546875953674316,
|
| 2026 |
+
"step": 1250
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.9625668449197861,
|
| 2030 |
+
"grad_norm": 80.05367241552668,
|
| 2031 |
+
"learning_rate": 2.1222410865874363e-08,
|
| 2032 |
+
"logits/chosen": 0.09323730319738388,
|
| 2033 |
+
"logits/rejected": 0.09921874850988388,
|
| 2034 |
+
"logps/chosen": -536.5999755859375,
|
| 2035 |
+
"logps/rejected": -564.4000244140625,
|
| 2036 |
+
"loss": 0.4261,
|
| 2037 |
+
"nll_loss": 1.044921875,
|
| 2038 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2039 |
+
"rewards/chosen": -1.2507812976837158,
|
| 2040 |
+
"rewards/margins": 2.0042967796325684,
|
| 2041 |
+
"rewards/rejected": -3.2562499046325684,
|
| 2042 |
+
"step": 1260
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.9702062643239114,
|
| 2046 |
+
"grad_norm": 57.04509819991225,
|
| 2047 |
+
"learning_rate": 1.697792869269949e-08,
|
| 2048 |
+
"logits/chosen": 0.16401366889476776,
|
| 2049 |
+
"logits/rejected": 0.06118469312787056,
|
| 2050 |
+
"logps/chosen": -505.0,
|
| 2051 |
+
"logps/rejected": -465.6000061035156,
|
| 2052 |
+
"loss": 0.4771,
|
| 2053 |
+
"nll_loss": 0.983203113079071,
|
| 2054 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 2055 |
+
"rewards/chosen": -1.430273413658142,
|
| 2056 |
+
"rewards/margins": 1.6331055164337158,
|
| 2057 |
+
"rewards/rejected": -3.0640625953674316,
|
| 2058 |
+
"step": 1270
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.9778456837280367,
|
| 2062 |
+
"grad_norm": 56.366216029864354,
|
| 2063 |
+
"learning_rate": 1.2733446519524618e-08,
|
| 2064 |
+
"logits/chosen": 0.14274902641773224,
|
| 2065 |
+
"logits/rejected": 0.08376464992761612,
|
| 2066 |
+
"logps/chosen": -548.5999755859375,
|
| 2067 |
+
"logps/rejected": -502.0,
|
| 2068 |
+
"loss": 0.4558,
|
| 2069 |
+
"nll_loss": 0.947265625,
|
| 2070 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 2071 |
+
"rewards/chosen": -1.254296898841858,
|
| 2072 |
+
"rewards/margins": 1.4153320789337158,
|
| 2073 |
+
"rewards/rejected": -2.6664061546325684,
|
| 2074 |
+
"step": 1280
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.985485103132162,
|
| 2078 |
+
"grad_norm": 172.76801282724654,
|
| 2079 |
+
"learning_rate": 8.488964346349745e-09,
|
| 2080 |
+
"logits/chosen": 0.13247069716453552,
|
| 2081 |
+
"logits/rejected": 0.21201172471046448,
|
| 2082 |
+
"logps/chosen": -625.4000244140625,
|
| 2083 |
+
"logps/rejected": -559.2000122070312,
|
| 2084 |
+
"loss": 0.6188,
|
| 2085 |
+
"nll_loss": 1.0027344226837158,
|
| 2086 |
+
"rewards/accuracies": 0.625,
|
| 2087 |
+
"rewards/chosen": -0.865917980670929,
|
| 2088 |
+
"rewards/margins": 1.8015625476837158,
|
| 2089 |
+
"rewards/rejected": -2.668750047683716,
|
| 2090 |
+
"step": 1290
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.9931245225362872,
|
| 2094 |
+
"grad_norm": 33.19453659649471,
|
| 2095 |
+
"learning_rate": 4.244482173174872e-09,
|
| 2096 |
+
"logits/chosen": 0.15334472060203552,
|
| 2097 |
+
"logits/rejected": 0.234130859375,
|
| 2098 |
+
"logps/chosen": -540.4000244140625,
|
| 2099 |
+
"logps/rejected": -483.79998779296875,
|
| 2100 |
+
"loss": 0.475,
|
| 2101 |
+
"nll_loss": 0.89453125,
|
| 2102 |
+
"rewards/accuracies": 0.762499988079071,
|
| 2103 |
+
"rewards/chosen": -0.7779296636581421,
|
| 2104 |
+
"rewards/margins": 1.6785156726837158,
|
| 2105 |
+
"rewards/rejected": -2.4566407203674316,
|
| 2106 |
+
"step": 1300
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 1.0,
|
| 2110 |
+
"eval_logits/chosen": 0.17315204441547394,
|
| 2111 |
+
"eval_logits/rejected": 0.33705490827560425,
|
| 2112 |
+
"eval_logps/chosen": -523.6923217773438,
|
| 2113 |
+
"eval_logps/rejected": -511.69232177734375,
|
| 2114 |
+
"eval_loss": 0.5020715594291687,
|
| 2115 |
+
"eval_nll_loss": 1.0318509340286255,
|
| 2116 |
+
"eval_rewards/accuracies": 0.6538461446762085,
|
| 2117 |
+
"eval_rewards/chosen": -1.2085336446762085,
|
| 2118 |
+
"eval_rewards/margins": 1.761868953704834,
|
| 2119 |
+
"eval_rewards/rejected": -2.966947078704834,
|
| 2120 |
+
"eval_runtime": 16.8968,
|
| 2121 |
+
"eval_samples_per_second": 5.918,
|
| 2122 |
+
"eval_steps_per_second": 0.769,
|
| 2123 |
+
"step": 1309
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.0,
|
| 2127 |
+
"step": 1309,
|
| 2128 |
+
"total_flos": 0.0,
|
| 2129 |
+
"train_loss": 0.4647196615989924,
|
| 2130 |
+
"train_runtime": 4672.0978,
|
| 2131 |
+
"train_samples_per_second": 2.24,
|
| 2132 |
+
"train_steps_per_second": 0.28
|
| 2133 |
+
}
|
| 2134 |
+
],
|
| 2135 |
+
"logging_steps": 10,
|
| 2136 |
+
"max_steps": 1309,
|
| 2137 |
+
"num_input_tokens_seen": 0,
|
| 2138 |
+
"num_train_epochs": 1,
|
| 2139 |
+
"save_steps": 500,
|
| 2140 |
+
"stateful_callbacks": {
|
| 2141 |
+
"TrainerControl": {
|
| 2142 |
+
"args": {
|
| 2143 |
+
"should_epoch_stop": false,
|
| 2144 |
+
"should_evaluate": false,
|
| 2145 |
+
"should_log": false,
|
| 2146 |
+
"should_save": true,
|
| 2147 |
+
"should_training_stop": true
|
| 2148 |
+
},
|
| 2149 |
+
"attributes": {}
|
| 2150 |
+
}
|
| 2151 |
+
},
|
| 2152 |
+
"total_flos": 0.0,
|
| 2153 |
+
"train_batch_size": 2,
|
| 2154 |
+
"trial_name": null,
|
| 2155 |
+
"trial_params": null
|
| 2156 |
+
}
|