loading model with peft support ==> this files are from checkpoint-200
Browse files- adapter_config.json +31 -0
- adapter_model.safetensors +3 -0
- trainer_state.json +2819 -0
- training_args.bin +3 -0
adapter_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "fblgit/una-cybertron-7b-v2-bf16",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"fan_in_fan_out": false,
|
| 7 |
+
"inference_mode": true,
|
| 8 |
+
"init_lora_weights": true,
|
| 9 |
+
"layers_pattern": null,
|
| 10 |
+
"layers_to_transform": null,
|
| 11 |
+
"loftq_config": {},
|
| 12 |
+
"lora_alpha": 16,
|
| 13 |
+
"lora_dropout": 0.05,
|
| 14 |
+
"megatron_config": null,
|
| 15 |
+
"megatron_core": "megatron.core",
|
| 16 |
+
"modules_to_save": null,
|
| 17 |
+
"peft_type": "LORA",
|
| 18 |
+
"r": 16,
|
| 19 |
+
"rank_pattern": {},
|
| 20 |
+
"revision": null,
|
| 21 |
+
"target_modules": [
|
| 22 |
+
"k_proj",
|
| 23 |
+
"up_proj",
|
| 24 |
+
"q_proj",
|
| 25 |
+
"o_proj",
|
| 26 |
+
"down_proj",
|
| 27 |
+
"gate_proj",
|
| 28 |
+
"v_proj"
|
| 29 |
+
],
|
| 30 |
+
"task_type": "CAUSAL_LM"
|
| 31 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:571440708ecebf18ead92f626a5411e123ec5cbecb85af2cdd7c8305e6e390b7
|
| 3 |
+
size 167832240
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2819 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.24883359253499224,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 5.000000000000001e-07,
|
| 14 |
+
"logits/chosen": -2.0463106632232666,
|
| 15 |
+
"logits/rejected": -1.8486888408660889,
|
| 16 |
+
"logps/chosen": -311.57489013671875,
|
| 17 |
+
"logps/rejected": -415.468505859375,
|
| 18 |
+
"loss": 0.671,
|
| 19 |
+
"rewards/accuracies": 0.75,
|
| 20 |
+
"rewards/chosen": -0.021808436140418053,
|
| 21 |
+
"rewards/margins": 0.04626159369945526,
|
| 22 |
+
"rewards/rejected": -0.06807002425193787,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 28 |
+
"logits/chosen": -2.060105562210083,
|
| 29 |
+
"logits/rejected": -2.106396198272705,
|
| 30 |
+
"logps/chosen": -220.380615234375,
|
| 31 |
+
"logps/rejected": -404.501220703125,
|
| 32 |
+
"loss": 0.6745,
|
| 33 |
+
"rewards/accuracies": 0.6875,
|
| 34 |
+
"rewards/chosen": -0.02021012455224991,
|
| 35 |
+
"rewards/margins": 0.03883519023656845,
|
| 36 |
+
"rewards/rejected": -0.05904531478881836,
|
| 37 |
+
"step": 2
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"learning_rate": 1.5e-06,
|
| 42 |
+
"logits/chosen": -2.239086151123047,
|
| 43 |
+
"logits/rejected": -2.0575525760650635,
|
| 44 |
+
"logps/chosen": -163.8677978515625,
|
| 45 |
+
"logps/rejected": -446.79180908203125,
|
| 46 |
+
"loss": 0.67,
|
| 47 |
+
"rewards/accuracies": 0.6875,
|
| 48 |
+
"rewards/chosen": -0.024322081357240677,
|
| 49 |
+
"rewards/margins": 0.04956565052270889,
|
| 50 |
+
"rewards/rejected": -0.07388772815465927,
|
| 51 |
+
"step": 3
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 56 |
+
"logits/chosen": -2.1058502197265625,
|
| 57 |
+
"logits/rejected": -2.344322681427002,
|
| 58 |
+
"logps/chosen": -318.9349670410156,
|
| 59 |
+
"logps/rejected": -378.9231262207031,
|
| 60 |
+
"loss": 0.6576,
|
| 61 |
+
"rewards/accuracies": 0.9375,
|
| 62 |
+
"rewards/chosen": -0.027724243700504303,
|
| 63 |
+
"rewards/margins": 0.07309143990278244,
|
| 64 |
+
"rewards/rejected": -0.10081568360328674,
|
| 65 |
+
"step": 4
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.01,
|
| 69 |
+
"learning_rate": 2.5e-06,
|
| 70 |
+
"logits/chosen": -2.287339687347412,
|
| 71 |
+
"logits/rejected": -2.3221607208251953,
|
| 72 |
+
"logps/chosen": -221.33163452148438,
|
| 73 |
+
"logps/rejected": -388.6357421875,
|
| 74 |
+
"loss": 0.6498,
|
| 75 |
+
"rewards/accuracies": 0.8125,
|
| 76 |
+
"rewards/chosen": -0.03215093910694122,
|
| 77 |
+
"rewards/margins": 0.09096165001392365,
|
| 78 |
+
"rewards/rejected": -0.12311258912086487,
|
| 79 |
+
"step": 5
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"learning_rate": 3e-06,
|
| 84 |
+
"logits/chosen": -1.98830246925354,
|
| 85 |
+
"logits/rejected": -1.9178718328475952,
|
| 86 |
+
"logps/chosen": -244.0532684326172,
|
| 87 |
+
"logps/rejected": -373.9111328125,
|
| 88 |
+
"loss": 0.6395,
|
| 89 |
+
"rewards/accuracies": 0.8125,
|
| 90 |
+
"rewards/chosen": -0.0789945125579834,
|
| 91 |
+
"rewards/margins": 0.11430712044239044,
|
| 92 |
+
"rewards/rejected": -0.19330161809921265,
|
| 93 |
+
"step": 6
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"learning_rate": 3.5000000000000004e-06,
|
| 98 |
+
"logits/chosen": -2.0385985374450684,
|
| 99 |
+
"logits/rejected": -1.9476547241210938,
|
| 100 |
+
"logps/chosen": -347.14227294921875,
|
| 101 |
+
"logps/rejected": -426.24603271484375,
|
| 102 |
+
"loss": 0.6148,
|
| 103 |
+
"rewards/accuracies": 0.9375,
|
| 104 |
+
"rewards/chosen": -0.0816906988620758,
|
| 105 |
+
"rewards/margins": 0.1658005714416504,
|
| 106 |
+
"rewards/rejected": -0.2474912703037262,
|
| 107 |
+
"step": 7
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"learning_rate": 4.000000000000001e-06,
|
| 112 |
+
"logits/chosen": -1.695872187614441,
|
| 113 |
+
"logits/rejected": -1.7980302572250366,
|
| 114 |
+
"logps/chosen": -287.9168395996094,
|
| 115 |
+
"logps/rejected": -355.14129638671875,
|
| 116 |
+
"loss": 0.5785,
|
| 117 |
+
"rewards/accuracies": 1.0,
|
| 118 |
+
"rewards/chosen": -0.07900090515613556,
|
| 119 |
+
"rewards/margins": 0.25094836950302124,
|
| 120 |
+
"rewards/rejected": -0.329949289560318,
|
| 121 |
+
"step": 8
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"learning_rate": 4.5e-06,
|
| 126 |
+
"logits/chosen": -1.9222339391708374,
|
| 127 |
+
"logits/rejected": -1.8972643613815308,
|
| 128 |
+
"logps/chosen": -211.97979736328125,
|
| 129 |
+
"logps/rejected": -405.848388671875,
|
| 130 |
+
"loss": 0.5229,
|
| 131 |
+
"rewards/accuracies": 1.0,
|
| 132 |
+
"rewards/chosen": -0.09827227890491486,
|
| 133 |
+
"rewards/margins": 0.3891267478466034,
|
| 134 |
+
"rewards/rejected": -0.48739901185035706,
|
| 135 |
+
"step": 9
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"learning_rate": 5e-06,
|
| 140 |
+
"logits/chosen": -1.9703621864318848,
|
| 141 |
+
"logits/rejected": -1.7560880184173584,
|
| 142 |
+
"logps/chosen": -277.2589416503906,
|
| 143 |
+
"logps/rejected": -445.6544494628906,
|
| 144 |
+
"loss": 0.4853,
|
| 145 |
+
"rewards/accuracies": 1.0,
|
| 146 |
+
"rewards/chosen": -0.1426083892583847,
|
| 147 |
+
"rewards/margins": 0.4861423373222351,
|
| 148 |
+
"rewards/rejected": -0.628750741481781,
|
| 149 |
+
"step": 10
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"learning_rate": 5.500000000000001e-06,
|
| 154 |
+
"logits/chosen": -2.0542986392974854,
|
| 155 |
+
"logits/rejected": -2.110658884048462,
|
| 156 |
+
"logps/chosen": -283.9775390625,
|
| 157 |
+
"logps/rejected": -394.361572265625,
|
| 158 |
+
"loss": 0.4476,
|
| 159 |
+
"rewards/accuracies": 1.0,
|
| 160 |
+
"rewards/chosen": -0.14311671257019043,
|
| 161 |
+
"rewards/margins": 0.6046640872955322,
|
| 162 |
+
"rewards/rejected": -0.7477807998657227,
|
| 163 |
+
"step": 11
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"learning_rate": 6e-06,
|
| 168 |
+
"logits/chosen": -2.0650949478149414,
|
| 169 |
+
"logits/rejected": -2.0736098289489746,
|
| 170 |
+
"logps/chosen": -235.07675170898438,
|
| 171 |
+
"logps/rejected": -448.17919921875,
|
| 172 |
+
"loss": 0.3701,
|
| 173 |
+
"rewards/accuracies": 1.0,
|
| 174 |
+
"rewards/chosen": -0.23901399970054626,
|
| 175 |
+
"rewards/margins": 0.9068580269813538,
|
| 176 |
+
"rewards/rejected": -1.1458721160888672,
|
| 177 |
+
"step": 12
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.02,
|
| 181 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 182 |
+
"logits/chosen": -2.011099100112915,
|
| 183 |
+
"logits/rejected": -2.106595754623413,
|
| 184 |
+
"logps/chosen": -289.19970703125,
|
| 185 |
+
"logps/rejected": -417.64208984375,
|
| 186 |
+
"loss": 0.3675,
|
| 187 |
+
"rewards/accuracies": 1.0,
|
| 188 |
+
"rewards/chosen": -0.26357388496398926,
|
| 189 |
+
"rewards/margins": 0.8955046534538269,
|
| 190 |
+
"rewards/rejected": -1.159078598022461,
|
| 191 |
+
"step": 13
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.02,
|
| 195 |
+
"learning_rate": 7.000000000000001e-06,
|
| 196 |
+
"logits/chosen": -2.1066994667053223,
|
| 197 |
+
"logits/rejected": -1.7827305793762207,
|
| 198 |
+
"logps/chosen": -169.27044677734375,
|
| 199 |
+
"logps/rejected": -446.7305908203125,
|
| 200 |
+
"loss": 0.254,
|
| 201 |
+
"rewards/accuracies": 1.0,
|
| 202 |
+
"rewards/chosen": -0.2715146541595459,
|
| 203 |
+
"rewards/margins": 1.3259609937667847,
|
| 204 |
+
"rewards/rejected": -1.5974756479263306,
|
| 205 |
+
"step": 14
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"learning_rate": 7.5e-06,
|
| 210 |
+
"logits/chosen": -1.9394950866699219,
|
| 211 |
+
"logits/rejected": -1.9831421375274658,
|
| 212 |
+
"logps/chosen": -194.98016357421875,
|
| 213 |
+
"logps/rejected": -470.9692687988281,
|
| 214 |
+
"loss": 0.2298,
|
| 215 |
+
"rewards/accuracies": 1.0,
|
| 216 |
+
"rewards/chosen": -0.4095136225223541,
|
| 217 |
+
"rewards/margins": 1.628616452217102,
|
| 218 |
+
"rewards/rejected": -2.038130044937134,
|
| 219 |
+
"step": 15
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"learning_rate": 8.000000000000001e-06,
|
| 224 |
+
"logits/chosen": -1.7606948614120483,
|
| 225 |
+
"logits/rejected": -2.0152878761291504,
|
| 226 |
+
"logps/chosen": -253.0347900390625,
|
| 227 |
+
"logps/rejected": -428.3993835449219,
|
| 228 |
+
"loss": 0.2528,
|
| 229 |
+
"rewards/accuracies": 1.0,
|
| 230 |
+
"rewards/chosen": -0.581798255443573,
|
| 231 |
+
"rewards/margins": 1.3869069814682007,
|
| 232 |
+
"rewards/rejected": -1.968705177307129,
|
| 233 |
+
"step": 16
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"learning_rate": 8.500000000000002e-06,
|
| 238 |
+
"logits/chosen": -2.0609982013702393,
|
| 239 |
+
"logits/rejected": -1.9835376739501953,
|
| 240 |
+
"logps/chosen": -265.3518371582031,
|
| 241 |
+
"logps/rejected": -501.8542785644531,
|
| 242 |
+
"loss": 0.1997,
|
| 243 |
+
"rewards/accuracies": 1.0,
|
| 244 |
+
"rewards/chosen": -0.681609570980072,
|
| 245 |
+
"rewards/margins": 2.0102298259735107,
|
| 246 |
+
"rewards/rejected": -2.6918392181396484,
|
| 247 |
+
"step": 17
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"learning_rate": 9e-06,
|
| 252 |
+
"logits/chosen": -2.117746114730835,
|
| 253 |
+
"logits/rejected": -2.218081474304199,
|
| 254 |
+
"logps/chosen": -202.6907501220703,
|
| 255 |
+
"logps/rejected": -413.13604736328125,
|
| 256 |
+
"loss": 0.1582,
|
| 257 |
+
"rewards/accuracies": 1.0,
|
| 258 |
+
"rewards/chosen": -0.6984068751335144,
|
| 259 |
+
"rewards/margins": 1.8970444202423096,
|
| 260 |
+
"rewards/rejected": -2.5954513549804688,
|
| 261 |
+
"step": 18
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"learning_rate": 9.5e-06,
|
| 266 |
+
"logits/chosen": -2.385519504547119,
|
| 267 |
+
"logits/rejected": -1.8222811222076416,
|
| 268 |
+
"logps/chosen": -175.93809509277344,
|
| 269 |
+
"logps/rejected": -469.17657470703125,
|
| 270 |
+
"loss": 0.0951,
|
| 271 |
+
"rewards/accuracies": 1.0,
|
| 272 |
+
"rewards/chosen": -0.7677481174468994,
|
| 273 |
+
"rewards/margins": 2.7113454341888428,
|
| 274 |
+
"rewards/rejected": -3.479093551635742,
|
| 275 |
+
"step": 19
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"learning_rate": 1e-05,
|
| 280 |
+
"logits/chosen": -2.187441349029541,
|
| 281 |
+
"logits/rejected": -2.1218032836914062,
|
| 282 |
+
"logps/chosen": -301.7166442871094,
|
| 283 |
+
"logps/rejected": -490.854736328125,
|
| 284 |
+
"loss": 0.0944,
|
| 285 |
+
"rewards/accuracies": 1.0,
|
| 286 |
+
"rewards/chosen": -1.2999629974365234,
|
| 287 |
+
"rewards/margins": 3.2447361946105957,
|
| 288 |
+
"rewards/rejected": -4.544699192047119,
|
| 289 |
+
"step": 20
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.03,
|
| 293 |
+
"learning_rate": 1.05e-05,
|
| 294 |
+
"logits/chosen": -1.967098355293274,
|
| 295 |
+
"logits/rejected": -1.9263262748718262,
|
| 296 |
+
"logps/chosen": -176.0479278564453,
|
| 297 |
+
"logps/rejected": -472.8221435546875,
|
| 298 |
+
"loss": 0.0459,
|
| 299 |
+
"rewards/accuracies": 1.0,
|
| 300 |
+
"rewards/chosen": -0.8402897119522095,
|
| 301 |
+
"rewards/margins": 4.1183624267578125,
|
| 302 |
+
"rewards/rejected": -4.958651542663574,
|
| 303 |
+
"step": 21
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.03,
|
| 307 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 308 |
+
"logits/chosen": -2.188459873199463,
|
| 309 |
+
"logits/rejected": -2.156071424484253,
|
| 310 |
+
"logps/chosen": -243.41683959960938,
|
| 311 |
+
"logps/rejected": -465.08685302734375,
|
| 312 |
+
"loss": 0.1389,
|
| 313 |
+
"rewards/accuracies": 0.9375,
|
| 314 |
+
"rewards/chosen": -1.5002142190933228,
|
| 315 |
+
"rewards/margins": 4.108778953552246,
|
| 316 |
+
"rewards/rejected": -5.608993053436279,
|
| 317 |
+
"step": 22
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.03,
|
| 321 |
+
"learning_rate": 1.1500000000000002e-05,
|
| 322 |
+
"logits/chosen": -1.7361280918121338,
|
| 323 |
+
"logits/rejected": -1.8734406232833862,
|
| 324 |
+
"logps/chosen": -274.4007568359375,
|
| 325 |
+
"logps/rejected": -465.7174987792969,
|
| 326 |
+
"loss": 0.074,
|
| 327 |
+
"rewards/accuracies": 1.0,
|
| 328 |
+
"rewards/chosen": -1.6840157508850098,
|
| 329 |
+
"rewards/margins": 4.5255866050720215,
|
| 330 |
+
"rewards/rejected": -6.209602355957031,
|
| 331 |
+
"step": 23
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"learning_rate": 1.2e-05,
|
| 336 |
+
"logits/chosen": -1.8966342210769653,
|
| 337 |
+
"logits/rejected": -1.772972583770752,
|
| 338 |
+
"logps/chosen": -275.3910217285156,
|
| 339 |
+
"logps/rejected": -462.07720947265625,
|
| 340 |
+
"loss": 0.0381,
|
| 341 |
+
"rewards/accuracies": 1.0,
|
| 342 |
+
"rewards/chosen": -2.017077922821045,
|
| 343 |
+
"rewards/margins": 4.740576267242432,
|
| 344 |
+
"rewards/rejected": -6.757654666900635,
|
| 345 |
+
"step": 24
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"learning_rate": 1.25e-05,
|
| 350 |
+
"logits/chosen": -2.2573952674865723,
|
| 351 |
+
"logits/rejected": -1.9883058071136475,
|
| 352 |
+
"logps/chosen": -172.16043090820312,
|
| 353 |
+
"logps/rejected": -391.4390563964844,
|
| 354 |
+
"loss": 0.0666,
|
| 355 |
+
"rewards/accuracies": 1.0,
|
| 356 |
+
"rewards/chosen": -1.9856681823730469,
|
| 357 |
+
"rewards/margins": 4.858271598815918,
|
| 358 |
+
"rewards/rejected": -6.843939781188965,
|
| 359 |
+
"step": 25
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 364 |
+
"logits/chosen": -2.0475640296936035,
|
| 365 |
+
"logits/rejected": -1.8690011501312256,
|
| 366 |
+
"logps/chosen": -184.4323272705078,
|
| 367 |
+
"logps/rejected": -468.86175537109375,
|
| 368 |
+
"loss": 0.0805,
|
| 369 |
+
"rewards/accuracies": 0.9375,
|
| 370 |
+
"rewards/chosen": -2.08463716506958,
|
| 371 |
+
"rewards/margins": 6.207521438598633,
|
| 372 |
+
"rewards/rejected": -8.292158126831055,
|
| 373 |
+
"step": 26
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"learning_rate": 1.3500000000000001e-05,
|
| 378 |
+
"logits/chosen": -1.9634509086608887,
|
| 379 |
+
"logits/rejected": -2.260413646697998,
|
| 380 |
+
"logps/chosen": -398.9476013183594,
|
| 381 |
+
"logps/rejected": -540.3334350585938,
|
| 382 |
+
"loss": 0.029,
|
| 383 |
+
"rewards/accuracies": 1.0,
|
| 384 |
+
"rewards/chosen": -3.3595480918884277,
|
| 385 |
+
"rewards/margins": 6.726611614227295,
|
| 386 |
+
"rewards/rejected": -10.086159706115723,
|
| 387 |
+
"step": 27
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 392 |
+
"logits/chosen": -1.9712620973587036,
|
| 393 |
+
"logits/rejected": -1.827807068824768,
|
| 394 |
+
"logps/chosen": -221.26559448242188,
|
| 395 |
+
"logps/rejected": -483.7889709472656,
|
| 396 |
+
"loss": 0.0063,
|
| 397 |
+
"rewards/accuracies": 1.0,
|
| 398 |
+
"rewards/chosen": -2.3635034561157227,
|
| 399 |
+
"rewards/margins": 7.733745098114014,
|
| 400 |
+
"rewards/rejected": -10.097248077392578,
|
| 401 |
+
"step": 28
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.04,
|
| 405 |
+
"learning_rate": 1.45e-05,
|
| 406 |
+
"logits/chosen": -1.542866587638855,
|
| 407 |
+
"logits/rejected": -1.7318823337554932,
|
| 408 |
+
"logps/chosen": -267.33880615234375,
|
| 409 |
+
"logps/rejected": -558.8424072265625,
|
| 410 |
+
"loss": 0.0746,
|
| 411 |
+
"rewards/accuracies": 0.9375,
|
| 412 |
+
"rewards/chosen": -3.391514301300049,
|
| 413 |
+
"rewards/margins": 9.136863708496094,
|
| 414 |
+
"rewards/rejected": -12.528377532958984,
|
| 415 |
+
"step": 29
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.04,
|
| 419 |
+
"learning_rate": 1.5e-05,
|
| 420 |
+
"logits/chosen": -2.037372589111328,
|
| 421 |
+
"logits/rejected": -1.8312575817108154,
|
| 422 |
+
"logps/chosen": -288.1588134765625,
|
| 423 |
+
"logps/rejected": -670.7269287109375,
|
| 424 |
+
"loss": 0.0007,
|
| 425 |
+
"rewards/accuracies": 1.0,
|
| 426 |
+
"rewards/chosen": -3.795253276824951,
|
| 427 |
+
"rewards/margins": 11.858222961425781,
|
| 428 |
+
"rewards/rejected": -15.653478622436523,
|
| 429 |
+
"step": 30
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.04,
|
| 433 |
+
"learning_rate": 1.55e-05,
|
| 434 |
+
"logits/chosen": -1.5374680757522583,
|
| 435 |
+
"logits/rejected": -1.8392678499221802,
|
| 436 |
+
"logps/chosen": -281.0010070800781,
|
| 437 |
+
"logps/rejected": -553.43017578125,
|
| 438 |
+
"loss": 0.0149,
|
| 439 |
+
"rewards/accuracies": 1.0,
|
| 440 |
+
"rewards/chosen": -3.836461067199707,
|
| 441 |
+
"rewards/margins": 9.380411148071289,
|
| 442 |
+
"rewards/rejected": -13.216873168945312,
|
| 443 |
+
"step": 31
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.04,
|
| 447 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 448 |
+
"logits/chosen": -2.033914566040039,
|
| 449 |
+
"logits/rejected": -1.9804089069366455,
|
| 450 |
+
"logps/chosen": -261.31103515625,
|
| 451 |
+
"logps/rejected": -545.380126953125,
|
| 452 |
+
"loss": 0.0123,
|
| 453 |
+
"rewards/accuracies": 1.0,
|
| 454 |
+
"rewards/chosen": -4.321026802062988,
|
| 455 |
+
"rewards/margins": 9.96163272857666,
|
| 456 |
+
"rewards/rejected": -14.282659530639648,
|
| 457 |
+
"step": 32
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"learning_rate": 1.65e-05,
|
| 462 |
+
"logits/chosen": -1.652747631072998,
|
| 463 |
+
"logits/rejected": -1.6527116298675537,
|
| 464 |
+
"logps/chosen": -250.82638549804688,
|
| 465 |
+
"logps/rejected": -536.0584106445312,
|
| 466 |
+
"loss": 0.0017,
|
| 467 |
+
"rewards/accuracies": 1.0,
|
| 468 |
+
"rewards/chosen": -4.393105506896973,
|
| 469 |
+
"rewards/margins": 11.006621360778809,
|
| 470 |
+
"rewards/rejected": -15.399725914001465,
|
| 471 |
+
"step": 33
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 476 |
+
"logits/chosen": -1.5870184898376465,
|
| 477 |
+
"logits/rejected": -1.9295532703399658,
|
| 478 |
+
"logps/chosen": -276.73248291015625,
|
| 479 |
+
"logps/rejected": -588.6087646484375,
|
| 480 |
+
"loss": 0.2671,
|
| 481 |
+
"rewards/accuracies": 0.875,
|
| 482 |
+
"rewards/chosen": -5.017908573150635,
|
| 483 |
+
"rewards/margins": 12.34096908569336,
|
| 484 |
+
"rewards/rejected": -17.35887908935547,
|
| 485 |
+
"step": 34
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"learning_rate": 1.75e-05,
|
| 490 |
+
"logits/chosen": -1.9227735996246338,
|
| 491 |
+
"logits/rejected": -1.954828143119812,
|
| 492 |
+
"logps/chosen": -266.1369323730469,
|
| 493 |
+
"logps/rejected": -550.742919921875,
|
| 494 |
+
"loss": 0.0035,
|
| 495 |
+
"rewards/accuracies": 1.0,
|
| 496 |
+
"rewards/chosen": -4.8841400146484375,
|
| 497 |
+
"rewards/margins": 13.669473648071289,
|
| 498 |
+
"rewards/rejected": -18.553613662719727,
|
| 499 |
+
"step": 35
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"learning_rate": 1.8e-05,
|
| 504 |
+
"logits/chosen": -1.6637210845947266,
|
| 505 |
+
"logits/rejected": -1.3936012983322144,
|
| 506 |
+
"logps/chosen": -256.7511901855469,
|
| 507 |
+
"logps/rejected": -591.1787109375,
|
| 508 |
+
"loss": 0.0022,
|
| 509 |
+
"rewards/accuracies": 1.0,
|
| 510 |
+
"rewards/chosen": -6.027996063232422,
|
| 511 |
+
"rewards/margins": 13.286176681518555,
|
| 512 |
+
"rewards/rejected": -19.314172744750977,
|
| 513 |
+
"step": 36
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.05,
|
| 517 |
+
"learning_rate": 1.85e-05,
|
| 518 |
+
"logits/chosen": -1.6170068979263306,
|
| 519 |
+
"logits/rejected": -1.6388572454452515,
|
| 520 |
+
"logps/chosen": -257.0255126953125,
|
| 521 |
+
"logps/rejected": -605.7335205078125,
|
| 522 |
+
"loss": 0.0008,
|
| 523 |
+
"rewards/accuracies": 1.0,
|
| 524 |
+
"rewards/chosen": -5.929409027099609,
|
| 525 |
+
"rewards/margins": 14.035564422607422,
|
| 526 |
+
"rewards/rejected": -19.9649715423584,
|
| 527 |
+
"step": 37
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.05,
|
| 531 |
+
"learning_rate": 1.9e-05,
|
| 532 |
+
"logits/chosen": -1.967097520828247,
|
| 533 |
+
"logits/rejected": -1.778485894203186,
|
| 534 |
+
"logps/chosen": -239.90533447265625,
|
| 535 |
+
"logps/rejected": -570.31787109375,
|
| 536 |
+
"loss": 0.0839,
|
| 537 |
+
"rewards/accuracies": 0.9375,
|
| 538 |
+
"rewards/chosen": -6.8724775314331055,
|
| 539 |
+
"rewards/margins": 13.892675399780273,
|
| 540 |
+
"rewards/rejected": -20.765153884887695,
|
| 541 |
+
"step": 38
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.05,
|
| 545 |
+
"learning_rate": 1.9500000000000003e-05,
|
| 546 |
+
"logits/chosen": -1.6693722009658813,
|
| 547 |
+
"logits/rejected": -1.573314905166626,
|
| 548 |
+
"logps/chosen": -349.86480712890625,
|
| 549 |
+
"logps/rejected": -702.0185546875,
|
| 550 |
+
"loss": 0.0008,
|
| 551 |
+
"rewards/accuracies": 1.0,
|
| 552 |
+
"rewards/chosen": -7.373364448547363,
|
| 553 |
+
"rewards/margins": 15.567668914794922,
|
| 554 |
+
"rewards/rejected": -22.941036224365234,
|
| 555 |
+
"step": 39
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.05,
|
| 559 |
+
"learning_rate": 2e-05,
|
| 560 |
+
"logits/chosen": -1.556606650352478,
|
| 561 |
+
"logits/rejected": -1.394392967224121,
|
| 562 |
+
"logps/chosen": -213.81959533691406,
|
| 563 |
+
"logps/rejected": -637.4287109375,
|
| 564 |
+
"loss": 0.041,
|
| 565 |
+
"rewards/accuracies": 1.0,
|
| 566 |
+
"rewards/chosen": -4.418168067932129,
|
| 567 |
+
"rewards/margins": 17.626686096191406,
|
| 568 |
+
"rewards/rejected": -22.04485321044922,
|
| 569 |
+
"step": 40
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.05,
|
| 573 |
+
"learning_rate": 2.05e-05,
|
| 574 |
+
"logits/chosen": -1.6436810493469238,
|
| 575 |
+
"logits/rejected": -1.7172099351882935,
|
| 576 |
+
"logps/chosen": -304.43438720703125,
|
| 577 |
+
"logps/rejected": -704.897216796875,
|
| 578 |
+
"loss": 0.0002,
|
| 579 |
+
"rewards/accuracies": 1.0,
|
| 580 |
+
"rewards/chosen": -6.733709812164307,
|
| 581 |
+
"rewards/margins": 19.767263412475586,
|
| 582 |
+
"rewards/rejected": -26.500972747802734,
|
| 583 |
+
"step": 41
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"learning_rate": 2.1e-05,
|
| 588 |
+
"logits/chosen": -1.5168399810791016,
|
| 589 |
+
"logits/rejected": -1.508206844329834,
|
| 590 |
+
"logps/chosen": -262.28515625,
|
| 591 |
+
"logps/rejected": -556.9771118164062,
|
| 592 |
+
"loss": 0.062,
|
| 593 |
+
"rewards/accuracies": 0.9375,
|
| 594 |
+
"rewards/chosen": -5.817533493041992,
|
| 595 |
+
"rewards/margins": 13.013591766357422,
|
| 596 |
+
"rewards/rejected": -18.831125259399414,
|
| 597 |
+
"step": 42
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"learning_rate": 2.15e-05,
|
| 602 |
+
"logits/chosen": -1.3590738773345947,
|
| 603 |
+
"logits/rejected": -1.1529805660247803,
|
| 604 |
+
"logps/chosen": -205.32371520996094,
|
| 605 |
+
"logps/rejected": -576.4129638671875,
|
| 606 |
+
"loss": 0.002,
|
| 607 |
+
"rewards/accuracies": 1.0,
|
| 608 |
+
"rewards/chosen": -5.326513290405273,
|
| 609 |
+
"rewards/margins": 15.988494873046875,
|
| 610 |
+
"rewards/rejected": -21.315006256103516,
|
| 611 |
+
"step": 43
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 616 |
+
"logits/chosen": -1.5084015130996704,
|
| 617 |
+
"logits/rejected": -1.5834801197052002,
|
| 618 |
+
"logps/chosen": -332.94580078125,
|
| 619 |
+
"logps/rejected": -685.3295288085938,
|
| 620 |
+
"loss": 0.0038,
|
| 621 |
+
"rewards/accuracies": 1.0,
|
| 622 |
+
"rewards/chosen": -7.390817642211914,
|
| 623 |
+
"rewards/margins": 18.214908599853516,
|
| 624 |
+
"rewards/rejected": -25.60572624206543,
|
| 625 |
+
"step": 44
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.06,
|
| 629 |
+
"learning_rate": 2.25e-05,
|
| 630 |
+
"logits/chosen": -1.4891091585159302,
|
| 631 |
+
"logits/rejected": -1.5173286199569702,
|
| 632 |
+
"logps/chosen": -250.24951171875,
|
| 633 |
+
"logps/rejected": -675.5242309570312,
|
| 634 |
+
"loss": 0.0026,
|
| 635 |
+
"rewards/accuracies": 1.0,
|
| 636 |
+
"rewards/chosen": -6.26884126663208,
|
| 637 |
+
"rewards/margins": 19.75680160522461,
|
| 638 |
+
"rewards/rejected": -26.025644302368164,
|
| 639 |
+
"step": 45
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.06,
|
| 643 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 644 |
+
"logits/chosen": -1.433956265449524,
|
| 645 |
+
"logits/rejected": -1.5273690223693848,
|
| 646 |
+
"logps/chosen": -366.8013000488281,
|
| 647 |
+
"logps/rejected": -703.315185546875,
|
| 648 |
+
"loss": 0.0103,
|
| 649 |
+
"rewards/accuracies": 1.0,
|
| 650 |
+
"rewards/chosen": -6.857599258422852,
|
| 651 |
+
"rewards/margins": 19.714298248291016,
|
| 652 |
+
"rewards/rejected": -26.571901321411133,
|
| 653 |
+
"step": 46
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.06,
|
| 657 |
+
"learning_rate": 2.35e-05,
|
| 658 |
+
"logits/chosen": -1.351868987083435,
|
| 659 |
+
"logits/rejected": -1.2347944974899292,
|
| 660 |
+
"logps/chosen": -292.37091064453125,
|
| 661 |
+
"logps/rejected": -687.9886474609375,
|
| 662 |
+
"loss": 0.0,
|
| 663 |
+
"rewards/accuracies": 1.0,
|
| 664 |
+
"rewards/chosen": -7.611140251159668,
|
| 665 |
+
"rewards/margins": 19.30370330810547,
|
| 666 |
+
"rewards/rejected": -26.91484260559082,
|
| 667 |
+
"step": 47
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.06,
|
| 671 |
+
"learning_rate": 2.4e-05,
|
| 672 |
+
"logits/chosen": -1.3243939876556396,
|
| 673 |
+
"logits/rejected": -1.0474777221679688,
|
| 674 |
+
"logps/chosen": -292.1935729980469,
|
| 675 |
+
"logps/rejected": -626.897705078125,
|
| 676 |
+
"loss": 0.0025,
|
| 677 |
+
"rewards/accuracies": 1.0,
|
| 678 |
+
"rewards/chosen": -6.425950527191162,
|
| 679 |
+
"rewards/margins": 17.86639404296875,
|
| 680 |
+
"rewards/rejected": -24.292346954345703,
|
| 681 |
+
"step": 48
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.06,
|
| 685 |
+
"learning_rate": 2.45e-05,
|
| 686 |
+
"logits/chosen": -0.9694186449050903,
|
| 687 |
+
"logits/rejected": -1.2784701585769653,
|
| 688 |
+
"logps/chosen": -260.8387756347656,
|
| 689 |
+
"logps/rejected": -618.62451171875,
|
| 690 |
+
"loss": 0.1257,
|
| 691 |
+
"rewards/accuracies": 0.9375,
|
| 692 |
+
"rewards/chosen": -6.1030497550964355,
|
| 693 |
+
"rewards/margins": 18.54592514038086,
|
| 694 |
+
"rewards/rejected": -24.648975372314453,
|
| 695 |
+
"step": 49
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.06,
|
| 699 |
+
"learning_rate": 2.5e-05,
|
| 700 |
+
"logits/chosen": -0.9199879765510559,
|
| 701 |
+
"logits/rejected": -1.2466750144958496,
|
| 702 |
+
"logps/chosen": -430.0943603515625,
|
| 703 |
+
"logps/rejected": -720.2520141601562,
|
| 704 |
+
"loss": 0.0001,
|
| 705 |
+
"rewards/accuracies": 1.0,
|
| 706 |
+
"rewards/chosen": -8.701850891113281,
|
| 707 |
+
"rewards/margins": 16.992839813232422,
|
| 708 |
+
"rewards/rejected": -25.694692611694336,
|
| 709 |
+
"step": 50
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"learning_rate": 2.5500000000000003e-05,
|
| 714 |
+
"logits/chosen": -0.9574620127677917,
|
| 715 |
+
"logits/rejected": -1.1837575435638428,
|
| 716 |
+
"logps/chosen": -301.4226379394531,
|
| 717 |
+
"logps/rejected": -568.0577392578125,
|
| 718 |
+
"loss": 0.0033,
|
| 719 |
+
"rewards/accuracies": 1.0,
|
| 720 |
+
"rewards/chosen": -6.406881332397461,
|
| 721 |
+
"rewards/margins": 13.906658172607422,
|
| 722 |
+
"rewards/rejected": -20.313539505004883,
|
| 723 |
+
"step": 51
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"learning_rate": 2.6000000000000002e-05,
|
| 728 |
+
"logits/chosen": -1.3775969743728638,
|
| 729 |
+
"logits/rejected": -1.1746721267700195,
|
| 730 |
+
"logps/chosen": -292.12841796875,
|
| 731 |
+
"logps/rejected": -541.6993408203125,
|
| 732 |
+
"loss": 0.0143,
|
| 733 |
+
"rewards/accuracies": 1.0,
|
| 734 |
+
"rewards/chosen": -6.582497596740723,
|
| 735 |
+
"rewards/margins": 13.31058120727539,
|
| 736 |
+
"rewards/rejected": -19.893077850341797,
|
| 737 |
+
"step": 52
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.07,
|
| 741 |
+
"learning_rate": 2.6500000000000004e-05,
|
| 742 |
+
"logits/chosen": -1.5340209007263184,
|
| 743 |
+
"logits/rejected": -1.5800985097885132,
|
| 744 |
+
"logps/chosen": -274.92852783203125,
|
| 745 |
+
"logps/rejected": -618.4048461914062,
|
| 746 |
+
"loss": 0.0005,
|
| 747 |
+
"rewards/accuracies": 1.0,
|
| 748 |
+
"rewards/chosen": -6.103458404541016,
|
| 749 |
+
"rewards/margins": 16.42413330078125,
|
| 750 |
+
"rewards/rejected": -22.527591705322266,
|
| 751 |
+
"step": 53
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.07,
|
| 755 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 756 |
+
"logits/chosen": -1.1566357612609863,
|
| 757 |
+
"logits/rejected": -1.3419249057769775,
|
| 758 |
+
"logps/chosen": -266.595458984375,
|
| 759 |
+
"logps/rejected": -609.8141479492188,
|
| 760 |
+
"loss": 0.0001,
|
| 761 |
+
"rewards/accuracies": 1.0,
|
| 762 |
+
"rewards/chosen": -5.4973931312561035,
|
| 763 |
+
"rewards/margins": 17.105182647705078,
|
| 764 |
+
"rewards/rejected": -22.602575302124023,
|
| 765 |
+
"step": 54
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.07,
|
| 769 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 770 |
+
"logits/chosen": -1.1755287647247314,
|
| 771 |
+
"logits/rejected": -1.436917781829834,
|
| 772 |
+
"logps/chosen": -293.0464172363281,
|
| 773 |
+
"logps/rejected": -614.6860961914062,
|
| 774 |
+
"loss": 0.038,
|
| 775 |
+
"rewards/accuracies": 1.0,
|
| 776 |
+
"rewards/chosen": -6.360505104064941,
|
| 777 |
+
"rewards/margins": 17.14010238647461,
|
| 778 |
+
"rewards/rejected": -23.5006103515625,
|
| 779 |
+
"step": 55
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.07,
|
| 783 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 784 |
+
"logits/chosen": -0.9510018825531006,
|
| 785 |
+
"logits/rejected": -1.4989765882492065,
|
| 786 |
+
"logps/chosen": -246.72573852539062,
|
| 787 |
+
"logps/rejected": -593.3004760742188,
|
| 788 |
+
"loss": 0.0019,
|
| 789 |
+
"rewards/accuracies": 1.0,
|
| 790 |
+
"rewards/chosen": -5.071164131164551,
|
| 791 |
+
"rewards/margins": 16.324851989746094,
|
| 792 |
+
"rewards/rejected": -21.396015167236328,
|
| 793 |
+
"step": 56
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.07,
|
| 797 |
+
"learning_rate": 2.8499999999999998e-05,
|
| 798 |
+
"logits/chosen": -0.8883547782897949,
|
| 799 |
+
"logits/rejected": -0.8475340604782104,
|
| 800 |
+
"logps/chosen": -284.69122314453125,
|
| 801 |
+
"logps/rejected": -582.0223999023438,
|
| 802 |
+
"loss": 0.092,
|
| 803 |
+
"rewards/accuracies": 0.9375,
|
| 804 |
+
"rewards/chosen": -5.555781364440918,
|
| 805 |
+
"rewards/margins": 15.147626876831055,
|
| 806 |
+
"rewards/rejected": -20.703407287597656,
|
| 807 |
+
"step": 57
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.07,
|
| 811 |
+
"learning_rate": 2.9e-05,
|
| 812 |
+
"logits/chosen": -1.420696496963501,
|
| 813 |
+
"logits/rejected": -1.3382985591888428,
|
| 814 |
+
"logps/chosen": -223.16297912597656,
|
| 815 |
+
"logps/rejected": -681.352294921875,
|
| 816 |
+
"loss": 0.0027,
|
| 817 |
+
"rewards/accuracies": 1.0,
|
| 818 |
+
"rewards/chosen": -4.9901018142700195,
|
| 819 |
+
"rewards/margins": 20.429107666015625,
|
| 820 |
+
"rewards/rejected": -25.419208526611328,
|
| 821 |
+
"step": 58
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.07,
|
| 825 |
+
"learning_rate": 2.95e-05,
|
| 826 |
+
"logits/chosen": -1.037627100944519,
|
| 827 |
+
"logits/rejected": -1.745126485824585,
|
| 828 |
+
"logps/chosen": -249.64974975585938,
|
| 829 |
+
"logps/rejected": -605.685546875,
|
| 830 |
+
"loss": 0.097,
|
| 831 |
+
"rewards/accuracies": 0.9375,
|
| 832 |
+
"rewards/chosen": -5.2863945960998535,
|
| 833 |
+
"rewards/margins": 16.568225860595703,
|
| 834 |
+
"rewards/rejected": -21.8546199798584,
|
| 835 |
+
"step": 59
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"learning_rate": 3e-05,
|
| 840 |
+
"logits/chosen": -1.1528346538543701,
|
| 841 |
+
"logits/rejected": -1.5133743286132812,
|
| 842 |
+
"logps/chosen": -319.42352294921875,
|
| 843 |
+
"logps/rejected": -627.826171875,
|
| 844 |
+
"loss": 0.0002,
|
| 845 |
+
"rewards/accuracies": 1.0,
|
| 846 |
+
"rewards/chosen": -5.5534515380859375,
|
| 847 |
+
"rewards/margins": 16.326492309570312,
|
| 848 |
+
"rewards/rejected": -21.87994384765625,
|
| 849 |
+
"step": 60
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.08,
|
| 853 |
+
"learning_rate": 3.05e-05,
|
| 854 |
+
"logits/chosen": -1.4599294662475586,
|
| 855 |
+
"logits/rejected": -1.4529486894607544,
|
| 856 |
+
"logps/chosen": -292.13299560546875,
|
| 857 |
+
"logps/rejected": -663.6395263671875,
|
| 858 |
+
"loss": 0.0036,
|
| 859 |
+
"rewards/accuracies": 1.0,
|
| 860 |
+
"rewards/chosen": -5.249754905700684,
|
| 861 |
+
"rewards/margins": 16.525672912597656,
|
| 862 |
+
"rewards/rejected": -21.775428771972656,
|
| 863 |
+
"step": 61
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.08,
|
| 867 |
+
"learning_rate": 3.1e-05,
|
| 868 |
+
"logits/chosen": -1.3739594221115112,
|
| 869 |
+
"logits/rejected": -1.441535472869873,
|
| 870 |
+
"logps/chosen": -361.3265380859375,
|
| 871 |
+
"logps/rejected": -623.9736938476562,
|
| 872 |
+
"loss": 0.0001,
|
| 873 |
+
"rewards/accuracies": 1.0,
|
| 874 |
+
"rewards/chosen": -4.781492233276367,
|
| 875 |
+
"rewards/margins": 16.213455200195312,
|
| 876 |
+
"rewards/rejected": -20.99494743347168,
|
| 877 |
+
"step": 62
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.08,
|
| 881 |
+
"learning_rate": 3.15e-05,
|
| 882 |
+
"logits/chosen": -1.0977773666381836,
|
| 883 |
+
"logits/rejected": -1.2894166707992554,
|
| 884 |
+
"logps/chosen": -272.4936828613281,
|
| 885 |
+
"logps/rejected": -633.6181640625,
|
| 886 |
+
"loss": 0.0,
|
| 887 |
+
"rewards/accuracies": 1.0,
|
| 888 |
+
"rewards/chosen": -5.058845520019531,
|
| 889 |
+
"rewards/margins": 18.464832305908203,
|
| 890 |
+
"rewards/rejected": -23.523677825927734,
|
| 891 |
+
"step": 63
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.08,
|
| 895 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 896 |
+
"logits/chosen": -1.8300187587738037,
|
| 897 |
+
"logits/rejected": -1.3356730937957764,
|
| 898 |
+
"logps/chosen": -311.8466796875,
|
| 899 |
+
"logps/rejected": -677.257568359375,
|
| 900 |
+
"loss": 0.0002,
|
| 901 |
+
"rewards/accuracies": 1.0,
|
| 902 |
+
"rewards/chosen": -5.877348899841309,
|
| 903 |
+
"rewards/margins": 16.247228622436523,
|
| 904 |
+
"rewards/rejected": -22.124576568603516,
|
| 905 |
+
"step": 64
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.08,
|
| 909 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 910 |
+
"logits/chosen": -1.6267454624176025,
|
| 911 |
+
"logits/rejected": -1.1310009956359863,
|
| 912 |
+
"logps/chosen": -358.5579833984375,
|
| 913 |
+
"logps/rejected": -641.7317504882812,
|
| 914 |
+
"loss": 0.0001,
|
| 915 |
+
"rewards/accuracies": 1.0,
|
| 916 |
+
"rewards/chosen": -5.5671796798706055,
|
| 917 |
+
"rewards/margins": 15.705550193786621,
|
| 918 |
+
"rewards/rejected": -21.27273178100586,
|
| 919 |
+
"step": 65
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.08,
|
| 923 |
+
"learning_rate": 3.3e-05,
|
| 924 |
+
"logits/chosen": -1.8895033597946167,
|
| 925 |
+
"logits/rejected": -1.6860744953155518,
|
| 926 |
+
"logps/chosen": -486.3850402832031,
|
| 927 |
+
"logps/rejected": -664.045166015625,
|
| 928 |
+
"loss": 0.0,
|
| 929 |
+
"rewards/accuracies": 1.0,
|
| 930 |
+
"rewards/chosen": -5.0797834396362305,
|
| 931 |
+
"rewards/margins": 13.979126930236816,
|
| 932 |
+
"rewards/rejected": -19.058910369873047,
|
| 933 |
+
"step": 66
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.08,
|
| 937 |
+
"learning_rate": 3.35e-05,
|
| 938 |
+
"logits/chosen": -1.836645483970642,
|
| 939 |
+
"logits/rejected": -1.4774869680404663,
|
| 940 |
+
"logps/chosen": -261.5281982421875,
|
| 941 |
+
"logps/rejected": -621.8162841796875,
|
| 942 |
+
"loss": 0.0011,
|
| 943 |
+
"rewards/accuracies": 1.0,
|
| 944 |
+
"rewards/chosen": -4.790907382965088,
|
| 945 |
+
"rewards/margins": 14.66439437866211,
|
| 946 |
+
"rewards/rejected": -19.45530128479004,
|
| 947 |
+
"step": 67
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.08,
|
| 951 |
+
"learning_rate": 3.4000000000000007e-05,
|
| 952 |
+
"logits/chosen": -1.4897315502166748,
|
| 953 |
+
"logits/rejected": -1.2426286935806274,
|
| 954 |
+
"logps/chosen": -225.94337463378906,
|
| 955 |
+
"logps/rejected": -547.6304321289062,
|
| 956 |
+
"loss": 0.0002,
|
| 957 |
+
"rewards/accuracies": 1.0,
|
| 958 |
+
"rewards/chosen": -4.037389755249023,
|
| 959 |
+
"rewards/margins": 13.804460525512695,
|
| 960 |
+
"rewards/rejected": -17.84185028076172,
|
| 961 |
+
"step": 68
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.09,
|
| 965 |
+
"learning_rate": 3.45e-05,
|
| 966 |
+
"logits/chosen": -1.3410481214523315,
|
| 967 |
+
"logits/rejected": -1.2696239948272705,
|
| 968 |
+
"logps/chosen": -220.23153686523438,
|
| 969 |
+
"logps/rejected": -545.1270751953125,
|
| 970 |
+
"loss": 0.0035,
|
| 971 |
+
"rewards/accuracies": 1.0,
|
| 972 |
+
"rewards/chosen": -4.0619893074035645,
|
| 973 |
+
"rewards/margins": 14.104720115661621,
|
| 974 |
+
"rewards/rejected": -18.166709899902344,
|
| 975 |
+
"step": 69
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.09,
|
| 979 |
+
"learning_rate": 3.5e-05,
|
| 980 |
+
"logits/chosen": -1.1692075729370117,
|
| 981 |
+
"logits/rejected": -1.4098597764968872,
|
| 982 |
+
"logps/chosen": -291.9318542480469,
|
| 983 |
+
"logps/rejected": -606.8259887695312,
|
| 984 |
+
"loss": 0.037,
|
| 985 |
+
"rewards/accuracies": 1.0,
|
| 986 |
+
"rewards/chosen": -4.749131679534912,
|
| 987 |
+
"rewards/margins": 13.3434476852417,
|
| 988 |
+
"rewards/rejected": -18.09257698059082,
|
| 989 |
+
"step": 70
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.09,
|
| 993 |
+
"learning_rate": 3.55e-05,
|
| 994 |
+
"logits/chosen": -1.291015386581421,
|
| 995 |
+
"logits/rejected": -1.0744752883911133,
|
| 996 |
+
"logps/chosen": -235.30746459960938,
|
| 997 |
+
"logps/rejected": -571.0240478515625,
|
| 998 |
+
"loss": 0.0002,
|
| 999 |
+
"rewards/accuracies": 1.0,
|
| 1000 |
+
"rewards/chosen": -3.615610122680664,
|
| 1001 |
+
"rewards/margins": 14.394880294799805,
|
| 1002 |
+
"rewards/rejected": -18.01049041748047,
|
| 1003 |
+
"step": 71
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.09,
|
| 1007 |
+
"learning_rate": 3.6e-05,
|
| 1008 |
+
"logits/chosen": -1.3118032217025757,
|
| 1009 |
+
"logits/rejected": -1.3245186805725098,
|
| 1010 |
+
"logps/chosen": -278.8104248046875,
|
| 1011 |
+
"logps/rejected": -621.2532958984375,
|
| 1012 |
+
"loss": 0.0,
|
| 1013 |
+
"rewards/accuracies": 1.0,
|
| 1014 |
+
"rewards/chosen": -4.277045249938965,
|
| 1015 |
+
"rewards/margins": 17.554424285888672,
|
| 1016 |
+
"rewards/rejected": -21.831466674804688,
|
| 1017 |
+
"step": 72
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.09,
|
| 1021 |
+
"learning_rate": 3.65e-05,
|
| 1022 |
+
"logits/chosen": -1.5896751880645752,
|
| 1023 |
+
"logits/rejected": -1.3609411716461182,
|
| 1024 |
+
"logps/chosen": -243.6151123046875,
|
| 1025 |
+
"logps/rejected": -572.5302124023438,
|
| 1026 |
+
"loss": 0.0006,
|
| 1027 |
+
"rewards/accuracies": 1.0,
|
| 1028 |
+
"rewards/chosen": -4.514758110046387,
|
| 1029 |
+
"rewards/margins": 12.668429374694824,
|
| 1030 |
+
"rewards/rejected": -17.183185577392578,
|
| 1031 |
+
"step": 73
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.09,
|
| 1035 |
+
"learning_rate": 3.7e-05,
|
| 1036 |
+
"logits/chosen": -1.431384801864624,
|
| 1037 |
+
"logits/rejected": -1.5531063079833984,
|
| 1038 |
+
"logps/chosen": -306.5356750488281,
|
| 1039 |
+
"logps/rejected": -600.71484375,
|
| 1040 |
+
"loss": 0.0001,
|
| 1041 |
+
"rewards/accuracies": 1.0,
|
| 1042 |
+
"rewards/chosen": -4.694637298583984,
|
| 1043 |
+
"rewards/margins": 18.718229293823242,
|
| 1044 |
+
"rewards/rejected": -23.412864685058594,
|
| 1045 |
+
"step": 74
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.09,
|
| 1049 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 1050 |
+
"logits/chosen": -1.557645559310913,
|
| 1051 |
+
"logits/rejected": -1.5208816528320312,
|
| 1052 |
+
"logps/chosen": -258.01324462890625,
|
| 1053 |
+
"logps/rejected": -619.49462890625,
|
| 1054 |
+
"loss": 0.0005,
|
| 1055 |
+
"rewards/accuracies": 1.0,
|
| 1056 |
+
"rewards/chosen": -5.0902862548828125,
|
| 1057 |
+
"rewards/margins": 15.939676284790039,
|
| 1058 |
+
"rewards/rejected": -21.029964447021484,
|
| 1059 |
+
"step": 75
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.09,
|
| 1063 |
+
"learning_rate": 3.8e-05,
|
| 1064 |
+
"logits/chosen": -1.0553147792816162,
|
| 1065 |
+
"logits/rejected": -1.5524070262908936,
|
| 1066 |
+
"logps/chosen": -309.97412109375,
|
| 1067 |
+
"logps/rejected": -524.318603515625,
|
| 1068 |
+
"loss": 0.0002,
|
| 1069 |
+
"rewards/accuracies": 1.0,
|
| 1070 |
+
"rewards/chosen": -4.439201354980469,
|
| 1071 |
+
"rewards/margins": 13.217429161071777,
|
| 1072 |
+
"rewards/rejected": -17.656631469726562,
|
| 1073 |
+
"step": 76
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.1,
|
| 1077 |
+
"learning_rate": 3.85e-05,
|
| 1078 |
+
"logits/chosen": -1.3808860778808594,
|
| 1079 |
+
"logits/rejected": -1.490052342414856,
|
| 1080 |
+
"logps/chosen": -249.54974365234375,
|
| 1081 |
+
"logps/rejected": -647.080078125,
|
| 1082 |
+
"loss": 0.0,
|
| 1083 |
+
"rewards/accuracies": 1.0,
|
| 1084 |
+
"rewards/chosen": -5.671748161315918,
|
| 1085 |
+
"rewards/margins": 17.54851531982422,
|
| 1086 |
+
"rewards/rejected": -23.220260620117188,
|
| 1087 |
+
"step": 77
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.1,
|
| 1091 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 1092 |
+
"logits/chosen": -1.6017792224884033,
|
| 1093 |
+
"logits/rejected": -1.437800407409668,
|
| 1094 |
+
"logps/chosen": -270.47894287109375,
|
| 1095 |
+
"logps/rejected": -523.1994018554688,
|
| 1096 |
+
"loss": 0.0107,
|
| 1097 |
+
"rewards/accuracies": 1.0,
|
| 1098 |
+
"rewards/chosen": -5.677766799926758,
|
| 1099 |
+
"rewards/margins": 13.664579391479492,
|
| 1100 |
+
"rewards/rejected": -19.34234619140625,
|
| 1101 |
+
"step": 78
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.1,
|
| 1105 |
+
"learning_rate": 3.9500000000000005e-05,
|
| 1106 |
+
"logits/chosen": -1.7644150257110596,
|
| 1107 |
+
"logits/rejected": -1.1649854183197021,
|
| 1108 |
+
"logps/chosen": -352.513671875,
|
| 1109 |
+
"logps/rejected": -660.660888671875,
|
| 1110 |
+
"loss": 0.0002,
|
| 1111 |
+
"rewards/accuracies": 1.0,
|
| 1112 |
+
"rewards/chosen": -6.161036968231201,
|
| 1113 |
+
"rewards/margins": 15.742412567138672,
|
| 1114 |
+
"rewards/rejected": -21.90345001220703,
|
| 1115 |
+
"step": 79
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.1,
|
| 1119 |
+
"learning_rate": 4e-05,
|
| 1120 |
+
"logits/chosen": -1.7110097408294678,
|
| 1121 |
+
"logits/rejected": -1.4986464977264404,
|
| 1122 |
+
"logps/chosen": -285.781005859375,
|
| 1123 |
+
"logps/rejected": -601.8524780273438,
|
| 1124 |
+
"loss": 0.0002,
|
| 1125 |
+
"rewards/accuracies": 1.0,
|
| 1126 |
+
"rewards/chosen": -6.4479594230651855,
|
| 1127 |
+
"rewards/margins": 17.196250915527344,
|
| 1128 |
+
"rewards/rejected": -23.644210815429688,
|
| 1129 |
+
"step": 80
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.1,
|
| 1133 |
+
"learning_rate": 4.05e-05,
|
| 1134 |
+
"logits/chosen": -1.5364179611206055,
|
| 1135 |
+
"logits/rejected": -1.5603926181793213,
|
| 1136 |
+
"logps/chosen": -205.1665802001953,
|
| 1137 |
+
"logps/rejected": -514.8365478515625,
|
| 1138 |
+
"loss": 0.0004,
|
| 1139 |
+
"rewards/accuracies": 1.0,
|
| 1140 |
+
"rewards/chosen": -5.383946418762207,
|
| 1141 |
+
"rewards/margins": 13.479593276977539,
|
| 1142 |
+
"rewards/rejected": -18.863540649414062,
|
| 1143 |
+
"step": 81
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.1,
|
| 1147 |
+
"learning_rate": 4.1e-05,
|
| 1148 |
+
"logits/chosen": -1.2903649806976318,
|
| 1149 |
+
"logits/rejected": -1.077599287033081,
|
| 1150 |
+
"logps/chosen": -240.99749755859375,
|
| 1151 |
+
"logps/rejected": -672.7728271484375,
|
| 1152 |
+
"loss": 0.0001,
|
| 1153 |
+
"rewards/accuracies": 1.0,
|
| 1154 |
+
"rewards/chosen": -5.660610675811768,
|
| 1155 |
+
"rewards/margins": 18.02043914794922,
|
| 1156 |
+
"rewards/rejected": -23.681049346923828,
|
| 1157 |
+
"step": 82
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.1,
|
| 1161 |
+
"learning_rate": 4.15e-05,
|
| 1162 |
+
"logits/chosen": -1.2030022144317627,
|
| 1163 |
+
"logits/rejected": -1.4182593822479248,
|
| 1164 |
+
"logps/chosen": -302.1430358886719,
|
| 1165 |
+
"logps/rejected": -677.5120849609375,
|
| 1166 |
+
"loss": 0.0056,
|
| 1167 |
+
"rewards/accuracies": 1.0,
|
| 1168 |
+
"rewards/chosen": -6.275785446166992,
|
| 1169 |
+
"rewards/margins": 16.768428802490234,
|
| 1170 |
+
"rewards/rejected": -23.044212341308594,
|
| 1171 |
+
"step": 83
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.1,
|
| 1175 |
+
"learning_rate": 4.2e-05,
|
| 1176 |
+
"logits/chosen": -1.459336519241333,
|
| 1177 |
+
"logits/rejected": -1.0312726497650146,
|
| 1178 |
+
"logps/chosen": -303.3758544921875,
|
| 1179 |
+
"logps/rejected": -617.7279052734375,
|
| 1180 |
+
"loss": 0.0002,
|
| 1181 |
+
"rewards/accuracies": 1.0,
|
| 1182 |
+
"rewards/chosen": -7.219513416290283,
|
| 1183 |
+
"rewards/margins": 14.163326263427734,
|
| 1184 |
+
"rewards/rejected": -21.38283920288086,
|
| 1185 |
+
"step": 84
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.11,
|
| 1189 |
+
"learning_rate": 4.25e-05,
|
| 1190 |
+
"logits/chosen": -1.6425917148590088,
|
| 1191 |
+
"logits/rejected": -1.1666898727416992,
|
| 1192 |
+
"logps/chosen": -349.3736267089844,
|
| 1193 |
+
"logps/rejected": -678.3955078125,
|
| 1194 |
+
"loss": 0.0,
|
| 1195 |
+
"rewards/accuracies": 1.0,
|
| 1196 |
+
"rewards/chosen": -7.792992115020752,
|
| 1197 |
+
"rewards/margins": 19.550434112548828,
|
| 1198 |
+
"rewards/rejected": -27.343425750732422,
|
| 1199 |
+
"step": 85
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.11,
|
| 1203 |
+
"learning_rate": 4.3e-05,
|
| 1204 |
+
"logits/chosen": -1.159060001373291,
|
| 1205 |
+
"logits/rejected": -1.2317029237747192,
|
| 1206 |
+
"logps/chosen": -377.60040283203125,
|
| 1207 |
+
"logps/rejected": -658.4701538085938,
|
| 1208 |
+
"loss": 0.0001,
|
| 1209 |
+
"rewards/accuracies": 1.0,
|
| 1210 |
+
"rewards/chosen": -8.494105339050293,
|
| 1211 |
+
"rewards/margins": 16.188596725463867,
|
| 1212 |
+
"rewards/rejected": -24.682701110839844,
|
| 1213 |
+
"step": 86
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.11,
|
| 1217 |
+
"learning_rate": 4.35e-05,
|
| 1218 |
+
"logits/chosen": -1.5545859336853027,
|
| 1219 |
+
"logits/rejected": -1.2424968481063843,
|
| 1220 |
+
"logps/chosen": -332.1812744140625,
|
| 1221 |
+
"logps/rejected": -686.550537109375,
|
| 1222 |
+
"loss": 0.0002,
|
| 1223 |
+
"rewards/accuracies": 1.0,
|
| 1224 |
+
"rewards/chosen": -7.362907409667969,
|
| 1225 |
+
"rewards/margins": 18.219987869262695,
|
| 1226 |
+
"rewards/rejected": -25.58289337158203,
|
| 1227 |
+
"step": 87
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.11,
|
| 1231 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 1232 |
+
"logits/chosen": -1.329568862915039,
|
| 1233 |
+
"logits/rejected": -1.5836312770843506,
|
| 1234 |
+
"logps/chosen": -344.522705078125,
|
| 1235 |
+
"logps/rejected": -663.6416015625,
|
| 1236 |
+
"loss": 0.0001,
|
| 1237 |
+
"rewards/accuracies": 1.0,
|
| 1238 |
+
"rewards/chosen": -8.556987762451172,
|
| 1239 |
+
"rewards/margins": 16.531614303588867,
|
| 1240 |
+
"rewards/rejected": -25.08860206604004,
|
| 1241 |
+
"step": 88
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.11,
|
| 1245 |
+
"learning_rate": 4.4500000000000004e-05,
|
| 1246 |
+
"logits/chosen": -0.9687823057174683,
|
| 1247 |
+
"logits/rejected": -1.0615956783294678,
|
| 1248 |
+
"logps/chosen": -248.99139404296875,
|
| 1249 |
+
"logps/rejected": -666.9677734375,
|
| 1250 |
+
"loss": 0.0001,
|
| 1251 |
+
"rewards/accuracies": 1.0,
|
| 1252 |
+
"rewards/chosen": -7.284422874450684,
|
| 1253 |
+
"rewards/margins": 18.494182586669922,
|
| 1254 |
+
"rewards/rejected": -25.778606414794922,
|
| 1255 |
+
"step": 89
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.11,
|
| 1259 |
+
"learning_rate": 4.5e-05,
|
| 1260 |
+
"logits/chosen": -1.5608181953430176,
|
| 1261 |
+
"logits/rejected": -1.603400707244873,
|
| 1262 |
+
"logps/chosen": -406.6678161621094,
|
| 1263 |
+
"logps/rejected": -738.91015625,
|
| 1264 |
+
"loss": 0.0006,
|
| 1265 |
+
"rewards/accuracies": 1.0,
|
| 1266 |
+
"rewards/chosen": -8.280235290527344,
|
| 1267 |
+
"rewards/margins": 20.654300689697266,
|
| 1268 |
+
"rewards/rejected": -28.93453598022461,
|
| 1269 |
+
"step": 90
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.11,
|
| 1273 |
+
"learning_rate": 4.55e-05,
|
| 1274 |
+
"logits/chosen": -1.1856168508529663,
|
| 1275 |
+
"logits/rejected": -1.2493767738342285,
|
| 1276 |
+
"logps/chosen": -248.82559204101562,
|
| 1277 |
+
"logps/rejected": -637.666259765625,
|
| 1278 |
+
"loss": 0.0007,
|
| 1279 |
+
"rewards/accuracies": 1.0,
|
| 1280 |
+
"rewards/chosen": -6.917981147766113,
|
| 1281 |
+
"rewards/margins": 18.320941925048828,
|
| 1282 |
+
"rewards/rejected": -25.238922119140625,
|
| 1283 |
+
"step": 91
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.11,
|
| 1287 |
+
"learning_rate": 4.600000000000001e-05,
|
| 1288 |
+
"logits/chosen": -1.4266817569732666,
|
| 1289 |
+
"logits/rejected": -1.3622314929962158,
|
| 1290 |
+
"logps/chosen": -236.7372283935547,
|
| 1291 |
+
"logps/rejected": -677.9640502929688,
|
| 1292 |
+
"loss": 0.0001,
|
| 1293 |
+
"rewards/accuracies": 1.0,
|
| 1294 |
+
"rewards/chosen": -7.995168685913086,
|
| 1295 |
+
"rewards/margins": 21.01068687438965,
|
| 1296 |
+
"rewards/rejected": -29.005855560302734,
|
| 1297 |
+
"step": 92
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.12,
|
| 1301 |
+
"learning_rate": 4.6500000000000005e-05,
|
| 1302 |
+
"logits/chosen": -1.1082837581634521,
|
| 1303 |
+
"logits/rejected": -1.383658528327942,
|
| 1304 |
+
"logps/chosen": -314.92681884765625,
|
| 1305 |
+
"logps/rejected": -658.524169921875,
|
| 1306 |
+
"loss": 0.0,
|
| 1307 |
+
"rewards/accuracies": 1.0,
|
| 1308 |
+
"rewards/chosen": -9.042872428894043,
|
| 1309 |
+
"rewards/margins": 17.163429260253906,
|
| 1310 |
+
"rewards/rejected": -26.206302642822266,
|
| 1311 |
+
"step": 93
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.12,
|
| 1315 |
+
"learning_rate": 4.7e-05,
|
| 1316 |
+
"logits/chosen": -1.5014522075653076,
|
| 1317 |
+
"logits/rejected": -1.7101225852966309,
|
| 1318 |
+
"logps/chosen": -260.1839904785156,
|
| 1319 |
+
"logps/rejected": -683.700927734375,
|
| 1320 |
+
"loss": 0.0,
|
| 1321 |
+
"rewards/accuracies": 1.0,
|
| 1322 |
+
"rewards/chosen": -9.463061332702637,
|
| 1323 |
+
"rewards/margins": 17.54586410522461,
|
| 1324 |
+
"rewards/rejected": -27.008926391601562,
|
| 1325 |
+
"step": 94
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.12,
|
| 1329 |
+
"learning_rate": 4.75e-05,
|
| 1330 |
+
"logits/chosen": -1.5970165729522705,
|
| 1331 |
+
"logits/rejected": -1.3586769104003906,
|
| 1332 |
+
"logps/chosen": -318.0454406738281,
|
| 1333 |
+
"logps/rejected": -684.9884033203125,
|
| 1334 |
+
"loss": 0.0,
|
| 1335 |
+
"rewards/accuracies": 1.0,
|
| 1336 |
+
"rewards/chosen": -9.960655212402344,
|
| 1337 |
+
"rewards/margins": 20.025028228759766,
|
| 1338 |
+
"rewards/rejected": -29.985679626464844,
|
| 1339 |
+
"step": 95
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.12,
|
| 1343 |
+
"learning_rate": 4.8e-05,
|
| 1344 |
+
"logits/chosen": -1.1033860445022583,
|
| 1345 |
+
"logits/rejected": -0.8102812767028809,
|
| 1346 |
+
"logps/chosen": -301.2135925292969,
|
| 1347 |
+
"logps/rejected": -720.0145263671875,
|
| 1348 |
+
"loss": 0.0,
|
| 1349 |
+
"rewards/accuracies": 1.0,
|
| 1350 |
+
"rewards/chosen": -8.136268615722656,
|
| 1351 |
+
"rewards/margins": 20.93104362487793,
|
| 1352 |
+
"rewards/rejected": -29.067312240600586,
|
| 1353 |
+
"step": 96
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.12,
|
| 1357 |
+
"learning_rate": 4.85e-05,
|
| 1358 |
+
"logits/chosen": -1.1413370370864868,
|
| 1359 |
+
"logits/rejected": -0.9585047960281372,
|
| 1360 |
+
"logps/chosen": -329.30133056640625,
|
| 1361 |
+
"logps/rejected": -630.4498291015625,
|
| 1362 |
+
"loss": 0.0,
|
| 1363 |
+
"rewards/accuracies": 1.0,
|
| 1364 |
+
"rewards/chosen": -9.226066589355469,
|
| 1365 |
+
"rewards/margins": 16.47555923461914,
|
| 1366 |
+
"rewards/rejected": -25.70162582397461,
|
| 1367 |
+
"step": 97
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.12,
|
| 1371 |
+
"learning_rate": 4.9e-05,
|
| 1372 |
+
"logits/chosen": -1.5005364418029785,
|
| 1373 |
+
"logits/rejected": -1.1674431562423706,
|
| 1374 |
+
"logps/chosen": -264.9545593261719,
|
| 1375 |
+
"logps/rejected": -639.041259765625,
|
| 1376 |
+
"loss": 0.0,
|
| 1377 |
+
"rewards/accuracies": 1.0,
|
| 1378 |
+
"rewards/chosen": -8.097036361694336,
|
| 1379 |
+
"rewards/margins": 17.606611251831055,
|
| 1380 |
+
"rewards/rejected": -25.70364761352539,
|
| 1381 |
+
"step": 98
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.12,
|
| 1385 |
+
"learning_rate": 4.9500000000000004e-05,
|
| 1386 |
+
"logits/chosen": -1.181979775428772,
|
| 1387 |
+
"logits/rejected": -1.1194593906402588,
|
| 1388 |
+
"logps/chosen": -295.6553955078125,
|
| 1389 |
+
"logps/rejected": -625.8030395507812,
|
| 1390 |
+
"loss": 0.0,
|
| 1391 |
+
"rewards/accuracies": 1.0,
|
| 1392 |
+
"rewards/chosen": -7.671294212341309,
|
| 1393 |
+
"rewards/margins": 18.466283798217773,
|
| 1394 |
+
"rewards/rejected": -26.1375789642334,
|
| 1395 |
+
"step": 99
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.12,
|
| 1399 |
+
"learning_rate": 5e-05,
|
| 1400 |
+
"logits/chosen": -1.6407731771469116,
|
| 1401 |
+
"logits/rejected": -1.7556546926498413,
|
| 1402 |
+
"logps/chosen": -251.4920654296875,
|
| 1403 |
+
"logps/rejected": -707.2771606445312,
|
| 1404 |
+
"loss": 0.0,
|
| 1405 |
+
"rewards/accuracies": 1.0,
|
| 1406 |
+
"rewards/chosen": -9.03681468963623,
|
| 1407 |
+
"rewards/margins": 19.950708389282227,
|
| 1408 |
+
"rewards/rejected": -28.98752212524414,
|
| 1409 |
+
"step": 100
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.13,
|
| 1413 |
+
"learning_rate": 4.998766400914329e-05,
|
| 1414 |
+
"logits/chosen": -1.49093496799469,
|
| 1415 |
+
"logits/rejected": -1.454042673110962,
|
| 1416 |
+
"logps/chosen": -401.38555908203125,
|
| 1417 |
+
"logps/rejected": -656.6805419921875,
|
| 1418 |
+
"loss": 0.0004,
|
| 1419 |
+
"rewards/accuracies": 1.0,
|
| 1420 |
+
"rewards/chosen": -9.029119491577148,
|
| 1421 |
+
"rewards/margins": 17.11648941040039,
|
| 1422 |
+
"rewards/rejected": -26.14560890197754,
|
| 1423 |
+
"step": 101
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.13,
|
| 1427 |
+
"learning_rate": 4.995066821070679e-05,
|
| 1428 |
+
"logits/chosen": -1.1679431200027466,
|
| 1429 |
+
"logits/rejected": -1.1486090421676636,
|
| 1430 |
+
"logps/chosen": -378.21893310546875,
|
| 1431 |
+
"logps/rejected": -720.4594116210938,
|
| 1432 |
+
"loss": 0.0,
|
| 1433 |
+
"rewards/accuracies": 1.0,
|
| 1434 |
+
"rewards/chosen": -9.716445922851562,
|
| 1435 |
+
"rewards/margins": 19.0845947265625,
|
| 1436 |
+
"rewards/rejected": -28.801040649414062,
|
| 1437 |
+
"step": 102
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.13,
|
| 1441 |
+
"learning_rate": 4.9889049115077005e-05,
|
| 1442 |
+
"logits/chosen": -1.0046123266220093,
|
| 1443 |
+
"logits/rejected": -1.2339215278625488,
|
| 1444 |
+
"logps/chosen": -258.95635986328125,
|
| 1445 |
+
"logps/rejected": -639.7680053710938,
|
| 1446 |
+
"loss": 0.0002,
|
| 1447 |
+
"rewards/accuracies": 1.0,
|
| 1448 |
+
"rewards/chosen": -8.76823902130127,
|
| 1449 |
+
"rewards/margins": 18.9598388671875,
|
| 1450 |
+
"rewards/rejected": -27.728076934814453,
|
| 1451 |
+
"step": 103
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.13,
|
| 1455 |
+
"learning_rate": 4.980286753286195e-05,
|
| 1456 |
+
"logits/chosen": -0.9866207242012024,
|
| 1457 |
+
"logits/rejected": -1.2342731952667236,
|
| 1458 |
+
"logps/chosen": -353.4874267578125,
|
| 1459 |
+
"logps/rejected": -652.6710205078125,
|
| 1460 |
+
"loss": 0.0,
|
| 1461 |
+
"rewards/accuracies": 1.0,
|
| 1462 |
+
"rewards/chosen": -8.72028923034668,
|
| 1463 |
+
"rewards/margins": 18.17083740234375,
|
| 1464 |
+
"rewards/rejected": -26.891128540039062,
|
| 1465 |
+
"step": 104
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.13,
|
| 1469 |
+
"learning_rate": 4.9692208514878444e-05,
|
| 1470 |
+
"logits/chosen": -1.644493818283081,
|
| 1471 |
+
"logits/rejected": -1.3671150207519531,
|
| 1472 |
+
"logps/chosen": -407.6358337402344,
|
| 1473 |
+
"logps/rejected": -680.9503784179688,
|
| 1474 |
+
"loss": 0.0001,
|
| 1475 |
+
"rewards/accuracies": 1.0,
|
| 1476 |
+
"rewards/chosen": -9.58061408996582,
|
| 1477 |
+
"rewards/margins": 16.537212371826172,
|
| 1478 |
+
"rewards/rejected": -26.117826461791992,
|
| 1479 |
+
"step": 105
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.13,
|
| 1483 |
+
"learning_rate": 4.9557181268217227e-05,
|
| 1484 |
+
"logits/chosen": -1.3883347511291504,
|
| 1485 |
+
"logits/rejected": -1.1226028203964233,
|
| 1486 |
+
"logps/chosen": -298.3403625488281,
|
| 1487 |
+
"logps/rejected": -661.0068969726562,
|
| 1488 |
+
"loss": 0.0,
|
| 1489 |
+
"rewards/accuracies": 1.0,
|
| 1490 |
+
"rewards/chosen": -8.479090690612793,
|
| 1491 |
+
"rewards/margins": 19.53042984008789,
|
| 1492 |
+
"rewards/rejected": -28.009523391723633,
|
| 1493 |
+
"step": 106
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.13,
|
| 1497 |
+
"learning_rate": 4.939791904846869e-05,
|
| 1498 |
+
"logits/chosen": -1.2119803428649902,
|
| 1499 |
+
"logits/rejected": -1.5936832427978516,
|
| 1500 |
+
"logps/chosen": -269.82684326171875,
|
| 1501 |
+
"logps/rejected": -645.09619140625,
|
| 1502 |
+
"loss": 0.0,
|
| 1503 |
+
"rewards/accuracies": 1.0,
|
| 1504 |
+
"rewards/chosen": -8.107880592346191,
|
| 1505 |
+
"rewards/margins": 20.64764976501465,
|
| 1506 |
+
"rewards/rejected": -28.755531311035156,
|
| 1507 |
+
"step": 107
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.13,
|
| 1511 |
+
"learning_rate": 4.9214579028215776e-05,
|
| 1512 |
+
"logits/chosen": -1.2132232189178467,
|
| 1513 |
+
"logits/rejected": -1.3179054260253906,
|
| 1514 |
+
"logps/chosen": -438.98321533203125,
|
| 1515 |
+
"logps/rejected": -713.7950439453125,
|
| 1516 |
+
"loss": 0.0,
|
| 1517 |
+
"rewards/accuracies": 1.0,
|
| 1518 |
+
"rewards/chosen": -10.158727645874023,
|
| 1519 |
+
"rewards/margins": 17.90839958190918,
|
| 1520 |
+
"rewards/rejected": -28.06713104248047,
|
| 1521 |
+
"step": 108
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.14,
|
| 1525 |
+
"learning_rate": 4.900734214192358e-05,
|
| 1526 |
+
"logits/chosen": -1.0279672145843506,
|
| 1527 |
+
"logits/rejected": -0.8536261320114136,
|
| 1528 |
+
"logps/chosen": -382.6248779296875,
|
| 1529 |
+
"logps/rejected": -738.4147338867188,
|
| 1530 |
+
"loss": 0.0034,
|
| 1531 |
+
"rewards/accuracies": 1.0,
|
| 1532 |
+
"rewards/chosen": -8.301639556884766,
|
| 1533 |
+
"rewards/margins": 20.556997299194336,
|
| 1534 |
+
"rewards/rejected": -28.858638763427734,
|
| 1535 |
+
"step": 109
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.14,
|
| 1539 |
+
"learning_rate": 4.877641290737884e-05,
|
| 1540 |
+
"logits/chosen": -1.1681768894195557,
|
| 1541 |
+
"logits/rejected": -0.6321386694908142,
|
| 1542 |
+
"logps/chosen": -295.9759521484375,
|
| 1543 |
+
"logps/rejected": -673.0234375,
|
| 1544 |
+
"loss": 0.0168,
|
| 1545 |
+
"rewards/accuracies": 1.0,
|
| 1546 |
+
"rewards/chosen": -10.42588996887207,
|
| 1547 |
+
"rewards/margins": 17.69922637939453,
|
| 1548 |
+
"rewards/rejected": -28.12511444091797,
|
| 1549 |
+
"step": 110
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.14,
|
| 1553 |
+
"learning_rate": 4.852201922385564e-05,
|
| 1554 |
+
"logits/chosen": -0.6276636123657227,
|
| 1555 |
+
"logits/rejected": -0.6425186991691589,
|
| 1556 |
+
"logps/chosen": -268.58050537109375,
|
| 1557 |
+
"logps/rejected": -601.134033203125,
|
| 1558 |
+
"loss": 0.0005,
|
| 1559 |
+
"rewards/accuracies": 1.0,
|
| 1560 |
+
"rewards/chosen": -6.623042583465576,
|
| 1561 |
+
"rewards/margins": 20.134323120117188,
|
| 1562 |
+
"rewards/rejected": -26.75736427307129,
|
| 1563 |
+
"step": 111
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.14,
|
| 1567 |
+
"learning_rate": 4.8244412147206284e-05,
|
| 1568 |
+
"logits/chosen": -1.298043966293335,
|
| 1569 |
+
"logits/rejected": -1.8934366703033447,
|
| 1570 |
+
"logps/chosen": -363.50616455078125,
|
| 1571 |
+
"logps/rejected": -793.5181274414062,
|
| 1572 |
+
"loss": 0.0,
|
| 1573 |
+
"rewards/accuracies": 1.0,
|
| 1574 |
+
"rewards/chosen": -9.013703346252441,
|
| 1575 |
+
"rewards/margins": 21.266841888427734,
|
| 1576 |
+
"rewards/rejected": -30.28054428100586,
|
| 1577 |
+
"step": 112
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.14,
|
| 1581 |
+
"learning_rate": 4.794386564209953e-05,
|
| 1582 |
+
"logits/chosen": -1.3218209743499756,
|
| 1583 |
+
"logits/rejected": -0.8168056011199951,
|
| 1584 |
+
"logps/chosen": -306.5829772949219,
|
| 1585 |
+
"logps/rejected": -696.270263671875,
|
| 1586 |
+
"loss": 0.0,
|
| 1587 |
+
"rewards/accuracies": 1.0,
|
| 1588 |
+
"rewards/chosen": -7.502519607543945,
|
| 1589 |
+
"rewards/margins": 23.955406188964844,
|
| 1590 |
+
"rewards/rejected": -31.457927703857422,
|
| 1591 |
+
"step": 113
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.14,
|
| 1595 |
+
"learning_rate": 4.762067631165049e-05,
|
| 1596 |
+
"logits/chosen": -1.1450629234313965,
|
| 1597 |
+
"logits/rejected": -1.393950343132019,
|
| 1598 |
+
"logps/chosen": -376.8450622558594,
|
| 1599 |
+
"logps/rejected": -673.3822631835938,
|
| 1600 |
+
"loss": 0.0,
|
| 1601 |
+
"rewards/accuracies": 1.0,
|
| 1602 |
+
"rewards/chosen": -9.097296714782715,
|
| 1603 |
+
"rewards/margins": 18.731908798217773,
|
| 1604 |
+
"rewards/rejected": -27.829204559326172,
|
| 1605 |
+
"step": 114
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.14,
|
| 1609 |
+
"learning_rate": 4.72751631047092e-05,
|
| 1610 |
+
"logits/chosen": -1.3087085485458374,
|
| 1611 |
+
"logits/rejected": -1.6303530931472778,
|
| 1612 |
+
"logps/chosen": -272.7848205566406,
|
| 1613 |
+
"logps/rejected": -697.6502685546875,
|
| 1614 |
+
"loss": 0.0,
|
| 1615 |
+
"rewards/accuracies": 1.0,
|
| 1616 |
+
"rewards/chosen": -7.2123308181762695,
|
| 1617 |
+
"rewards/margins": 22.104957580566406,
|
| 1618 |
+
"rewards/rejected": -29.31728744506836,
|
| 1619 |
+
"step": 115
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.14,
|
| 1623 |
+
"learning_rate": 4.690766700109659e-05,
|
| 1624 |
+
"logits/chosen": -1.3388525247573853,
|
| 1625 |
+
"logits/rejected": -1.523690938949585,
|
| 1626 |
+
"logps/chosen": -281.05694580078125,
|
| 1627 |
+
"logps/rejected": -593.7879028320312,
|
| 1628 |
+
"loss": 0.0035,
|
| 1629 |
+
"rewards/accuracies": 1.0,
|
| 1630 |
+
"rewards/chosen": -5.953818321228027,
|
| 1631 |
+
"rewards/margins": 17.667112350463867,
|
| 1632 |
+
"rewards/rejected": -23.620929718017578,
|
| 1633 |
+
"step": 116
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.15,
|
| 1637 |
+
"learning_rate": 4.65185506750986e-05,
|
| 1638 |
+
"logits/chosen": -1.3922251462936401,
|
| 1639 |
+
"logits/rejected": -1.2891855239868164,
|
| 1640 |
+
"logps/chosen": -322.259765625,
|
| 1641 |
+
"logps/rejected": -654.0626220703125,
|
| 1642 |
+
"loss": 0.0001,
|
| 1643 |
+
"rewards/accuracies": 1.0,
|
| 1644 |
+
"rewards/chosen": -6.73513126373291,
|
| 1645 |
+
"rewards/margins": 19.120628356933594,
|
| 1646 |
+
"rewards/rejected": -25.855758666992188,
|
| 1647 |
+
"step": 117
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.15,
|
| 1651 |
+
"learning_rate": 4.610819813755038e-05,
|
| 1652 |
+
"logits/chosen": -1.0762147903442383,
|
| 1653 |
+
"logits/rejected": -1.2459900379180908,
|
| 1654 |
+
"logps/chosen": -339.0086364746094,
|
| 1655 |
+
"logps/rejected": -729.5966796875,
|
| 1656 |
+
"loss": 0.0,
|
| 1657 |
+
"rewards/accuracies": 1.0,
|
| 1658 |
+
"rewards/chosen": -7.075998306274414,
|
| 1659 |
+
"rewards/margins": 22.65298843383789,
|
| 1660 |
+
"rewards/rejected": -29.728988647460938,
|
| 1661 |
+
"step": 118
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.15,
|
| 1665 |
+
"learning_rate": 4.567701435686404e-05,
|
| 1666 |
+
"logits/chosen": -1.4852521419525146,
|
| 1667 |
+
"logits/rejected": -1.3211164474487305,
|
| 1668 |
+
"logps/chosen": -221.89089965820312,
|
| 1669 |
+
"logps/rejected": -554.428466796875,
|
| 1670 |
+
"loss": 0.0,
|
| 1671 |
+
"rewards/accuracies": 1.0,
|
| 1672 |
+
"rewards/chosen": -6.827755928039551,
|
| 1673 |
+
"rewards/margins": 15.966878890991211,
|
| 1674 |
+
"rewards/rejected": -22.794635772705078,
|
| 1675 |
+
"step": 119
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.15,
|
| 1679 |
+
"learning_rate": 4.522542485937369e-05,
|
| 1680 |
+
"logits/chosen": -1.4120067358016968,
|
| 1681 |
+
"logits/rejected": -1.1943210363388062,
|
| 1682 |
+
"logps/chosen": -251.79034423828125,
|
| 1683 |
+
"logps/rejected": -700.6851196289062,
|
| 1684 |
+
"loss": 0.0,
|
| 1685 |
+
"rewards/accuracies": 1.0,
|
| 1686 |
+
"rewards/chosen": -6.087055206298828,
|
| 1687 |
+
"rewards/margins": 20.131147384643555,
|
| 1688 |
+
"rewards/rejected": -26.218202590942383,
|
| 1689 |
+
"step": 120
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.15,
|
| 1693 |
+
"learning_rate": 4.4753875309392266e-05,
|
| 1694 |
+
"logits/chosen": -1.564077377319336,
|
| 1695 |
+
"logits/rejected": -1.4048410654067993,
|
| 1696 |
+
"logps/chosen": -260.0417175292969,
|
| 1697 |
+
"logps/rejected": -702.1083984375,
|
| 1698 |
+
"loss": 0.0,
|
| 1699 |
+
"rewards/accuracies": 1.0,
|
| 1700 |
+
"rewards/chosen": -7.242954730987549,
|
| 1701 |
+
"rewards/margins": 23.306888580322266,
|
| 1702 |
+
"rewards/rejected": -30.549842834472656,
|
| 1703 |
+
"step": 121
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.15,
|
| 1707 |
+
"learning_rate": 4.426283106939474e-05,
|
| 1708 |
+
"logits/chosen": -1.6456904411315918,
|
| 1709 |
+
"logits/rejected": -1.3609850406646729,
|
| 1710 |
+
"logps/chosen": -267.5194396972656,
|
| 1711 |
+
"logps/rejected": -620.3526611328125,
|
| 1712 |
+
"loss": 0.0,
|
| 1713 |
+
"rewards/accuracies": 1.0,
|
| 1714 |
+
"rewards/chosen": -6.6236772537231445,
|
| 1715 |
+
"rewards/margins": 17.591167449951172,
|
| 1716 |
+
"rewards/rejected": -24.21484375,
|
| 1717 |
+
"step": 122
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.15,
|
| 1721 |
+
"learning_rate": 4.375277674076149e-05,
|
| 1722 |
+
"logits/chosen": -0.8970831632614136,
|
| 1723 |
+
"logits/rejected": -1.2291415929794312,
|
| 1724 |
+
"logps/chosen": -263.4637451171875,
|
| 1725 |
+
"logps/rejected": -561.6110229492188,
|
| 1726 |
+
"loss": 0.0,
|
| 1727 |
+
"rewards/accuracies": 1.0,
|
| 1728 |
+
"rewards/chosen": -5.951724052429199,
|
| 1729 |
+
"rewards/margins": 17.361831665039062,
|
| 1730 |
+
"rewards/rejected": -23.313554763793945,
|
| 1731 |
+
"step": 123
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.15,
|
| 1735 |
+
"learning_rate": 4.3224215685535294e-05,
|
| 1736 |
+
"logits/chosen": -1.5338877439498901,
|
| 1737 |
+
"logits/rejected": -1.3900439739227295,
|
| 1738 |
+
"logps/chosen": -266.0079345703125,
|
| 1739 |
+
"logps/rejected": -646.4955444335938,
|
| 1740 |
+
"loss": 0.0002,
|
| 1741 |
+
"rewards/accuracies": 1.0,
|
| 1742 |
+
"rewards/chosen": -6.709009170532227,
|
| 1743 |
+
"rewards/margins": 20.428083419799805,
|
| 1744 |
+
"rewards/rejected": -27.1370906829834,
|
| 1745 |
+
"step": 124
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.16,
|
| 1749 |
+
"learning_rate": 4.267766952966369e-05,
|
| 1750 |
+
"logits/chosen": -1.0960302352905273,
|
| 1751 |
+
"logits/rejected": -1.468554139137268,
|
| 1752 |
+
"logps/chosen": -285.5207214355469,
|
| 1753 |
+
"logps/rejected": -672.1173095703125,
|
| 1754 |
+
"loss": 0.0,
|
| 1755 |
+
"rewards/accuracies": 1.0,
|
| 1756 |
+
"rewards/chosen": -4.691971302032471,
|
| 1757 |
+
"rewards/margins": 24.13559913635254,
|
| 1758 |
+
"rewards/rejected": -28.82756996154785,
|
| 1759 |
+
"step": 125
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.16,
|
| 1763 |
+
"learning_rate": 4.211367764821722e-05,
|
| 1764 |
+
"logits/chosen": -2.010265588760376,
|
| 1765 |
+
"logits/rejected": -1.6019049882888794,
|
| 1766 |
+
"logps/chosen": -354.89190673828125,
|
| 1767 |
+
"logps/rejected": -666.0311279296875,
|
| 1768 |
+
"loss": 0.0,
|
| 1769 |
+
"rewards/accuracies": 1.0,
|
| 1770 |
+
"rewards/chosen": -6.3796706199646,
|
| 1771 |
+
"rewards/margins": 19.254438400268555,
|
| 1772 |
+
"rewards/rejected": -25.634109497070312,
|
| 1773 |
+
"step": 126
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.16,
|
| 1777 |
+
"learning_rate": 4.1532796633091296e-05,
|
| 1778 |
+
"logits/chosen": -1.8576483726501465,
|
| 1779 |
+
"logits/rejected": -1.5691324472427368,
|
| 1780 |
+
"logps/chosen": -286.91375732421875,
|
| 1781 |
+
"logps/rejected": -716.0498046875,
|
| 1782 |
+
"loss": 0.0,
|
| 1783 |
+
"rewards/accuracies": 1.0,
|
| 1784 |
+
"rewards/chosen": -6.725906848907471,
|
| 1785 |
+
"rewards/margins": 24.17666244506836,
|
| 1786 |
+
"rewards/rejected": -30.902565002441406,
|
| 1787 |
+
"step": 127
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.16,
|
| 1791 |
+
"learning_rate": 4.093559974371725e-05,
|
| 1792 |
+
"logits/chosen": -1.9219647645950317,
|
| 1793 |
+
"logits/rejected": -1.7591352462768555,
|
| 1794 |
+
"logps/chosen": -364.5614318847656,
|
| 1795 |
+
"logps/rejected": -736.0914306640625,
|
| 1796 |
+
"loss": 0.0015,
|
| 1797 |
+
"rewards/accuracies": 1.0,
|
| 1798 |
+
"rewards/chosen": -7.333095550537109,
|
| 1799 |
+
"rewards/margins": 23.11652946472168,
|
| 1800 |
+
"rewards/rejected": -30.44962501525879,
|
| 1801 |
+
"step": 128
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.16,
|
| 1805 |
+
"learning_rate": 4.0322676341324415e-05,
|
| 1806 |
+
"logits/chosen": -1.2601250410079956,
|
| 1807 |
+
"logits/rejected": -1.2032322883605957,
|
| 1808 |
+
"logps/chosen": -316.52935791015625,
|
| 1809 |
+
"logps/rejected": -660.7392578125,
|
| 1810 |
+
"loss": 0.0001,
|
| 1811 |
+
"rewards/accuracies": 1.0,
|
| 1812 |
+
"rewards/chosen": -6.681999206542969,
|
| 1813 |
+
"rewards/margins": 18.961524963378906,
|
| 1814 |
+
"rewards/rejected": -25.643526077270508,
|
| 1815 |
+
"step": 129
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.16,
|
| 1819 |
+
"learning_rate": 3.969463130731183e-05,
|
| 1820 |
+
"logits/chosen": -1.1819369792938232,
|
| 1821 |
+
"logits/rejected": -1.330411672592163,
|
| 1822 |
+
"logps/chosen": -334.45440673828125,
|
| 1823 |
+
"logps/rejected": -668.8825073242188,
|
| 1824 |
+
"loss": 0.0001,
|
| 1825 |
+
"rewards/accuracies": 1.0,
|
| 1826 |
+
"rewards/chosen": -6.4721527099609375,
|
| 1827 |
+
"rewards/margins": 20.78248405456543,
|
| 1828 |
+
"rewards/rejected": -27.254636764526367,
|
| 1829 |
+
"step": 130
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.16,
|
| 1833 |
+
"learning_rate": 3.905208444630327e-05,
|
| 1834 |
+
"logits/chosen": -1.359390377998352,
|
| 1835 |
+
"logits/rejected": -1.4633076190948486,
|
| 1836 |
+
"logps/chosen": -449.515869140625,
|
| 1837 |
+
"logps/rejected": -722.9254760742188,
|
| 1838 |
+
"loss": 0.0009,
|
| 1839 |
+
"rewards/accuracies": 1.0,
|
| 1840 |
+
"rewards/chosen": -8.208673477172852,
|
| 1841 |
+
"rewards/margins": 17.24679946899414,
|
| 1842 |
+
"rewards/rejected": -25.45547103881836,
|
| 1843 |
+
"step": 131
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.16,
|
| 1847 |
+
"learning_rate": 3.8395669874474915e-05,
|
| 1848 |
+
"logits/chosen": -1.5751495361328125,
|
| 1849 |
+
"logits/rejected": -1.4079082012176514,
|
| 1850 |
+
"logps/chosen": -316.51824951171875,
|
| 1851 |
+
"logps/rejected": -690.4049072265625,
|
| 1852 |
+
"loss": 0.0,
|
| 1853 |
+
"rewards/accuracies": 1.0,
|
| 1854 |
+
"rewards/chosen": -6.716471195220947,
|
| 1855 |
+
"rewards/margins": 21.65338134765625,
|
| 1856 |
+
"rewards/rejected": -28.369853973388672,
|
| 1857 |
+
"step": 132
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.17,
|
| 1861 |
+
"learning_rate": 3.7726035393759285e-05,
|
| 1862 |
+
"logits/chosen": -1.5979351997375488,
|
| 1863 |
+
"logits/rejected": -1.5114738941192627,
|
| 1864 |
+
"logps/chosen": -395.20916748046875,
|
| 1865 |
+
"logps/rejected": -718.419677734375,
|
| 1866 |
+
"loss": 0.0001,
|
| 1867 |
+
"rewards/accuracies": 1.0,
|
| 1868 |
+
"rewards/chosen": -7.750358581542969,
|
| 1869 |
+
"rewards/margins": 18.212371826171875,
|
| 1870 |
+
"rewards/rejected": -25.96272850036621,
|
| 1871 |
+
"step": 133
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.17,
|
| 1875 |
+
"learning_rate": 3.704384185254288e-05,
|
| 1876 |
+
"logits/chosen": -1.5029571056365967,
|
| 1877 |
+
"logits/rejected": -1.4255585670471191,
|
| 1878 |
+
"logps/chosen": -254.38461303710938,
|
| 1879 |
+
"logps/rejected": -803.1172485351562,
|
| 1880 |
+
"loss": 0.0,
|
| 1881 |
+
"rewards/accuracies": 1.0,
|
| 1882 |
+
"rewards/chosen": -6.148038387298584,
|
| 1883 |
+
"rewards/margins": 28.40212059020996,
|
| 1884 |
+
"rewards/rejected": -34.5501594543457,
|
| 1885 |
+
"step": 134
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.17,
|
| 1889 |
+
"learning_rate": 3.634976249348867e-05,
|
| 1890 |
+
"logits/chosen": -1.0641860961914062,
|
| 1891 |
+
"logits/rejected": -1.3426963090896606,
|
| 1892 |
+
"logps/chosen": -322.1287536621094,
|
| 1893 |
+
"logps/rejected": -627.5282592773438,
|
| 1894 |
+
"loss": 0.0,
|
| 1895 |
+
"rewards/accuracies": 1.0,
|
| 1896 |
+
"rewards/chosen": -7.379766464233398,
|
| 1897 |
+
"rewards/margins": 17.175968170166016,
|
| 1898 |
+
"rewards/rejected": -24.555734634399414,
|
| 1899 |
+
"step": 135
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.17,
|
| 1903 |
+
"learning_rate": 3.564448228912682e-05,
|
| 1904 |
+
"logits/chosen": -1.4602770805358887,
|
| 1905 |
+
"logits/rejected": -1.4505391120910645,
|
| 1906 |
+
"logps/chosen": -343.05450439453125,
|
| 1907 |
+
"logps/rejected": -651.352783203125,
|
| 1908 |
+
"loss": 0.0052,
|
| 1909 |
+
"rewards/accuracies": 1.0,
|
| 1910 |
+
"rewards/chosen": -8.195021629333496,
|
| 1911 |
+
"rewards/margins": 18.67833709716797,
|
| 1912 |
+
"rewards/rejected": -26.87335777282715,
|
| 1913 |
+
"step": 136
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.17,
|
| 1917 |
+
"learning_rate": 3.4928697265869515e-05,
|
| 1918 |
+
"logits/chosen": -1.2342597246170044,
|
| 1919 |
+
"logits/rejected": -0.98235023021698,
|
| 1920 |
+
"logps/chosen": -267.2305603027344,
|
| 1921 |
+
"logps/rejected": -662.8800048828125,
|
| 1922 |
+
"loss": 0.0,
|
| 1923 |
+
"rewards/accuracies": 1.0,
|
| 1924 |
+
"rewards/chosen": -5.990522384643555,
|
| 1925 |
+
"rewards/margins": 20.408323287963867,
|
| 1926 |
+
"rewards/rejected": -26.39884376525879,
|
| 1927 |
+
"step": 137
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.17,
|
| 1931 |
+
"learning_rate": 3.4203113817116957e-05,
|
| 1932 |
+
"logits/chosen": -1.5317420959472656,
|
| 1933 |
+
"logits/rejected": -1.4648504257202148,
|
| 1934 |
+
"logps/chosen": -291.79547119140625,
|
| 1935 |
+
"logps/rejected": -663.3065795898438,
|
| 1936 |
+
"loss": 0.0,
|
| 1937 |
+
"rewards/accuracies": 1.0,
|
| 1938 |
+
"rewards/chosen": -7.487759590148926,
|
| 1939 |
+
"rewards/margins": 20.33439064025879,
|
| 1940 |
+
"rewards/rejected": -27.8221492767334,
|
| 1941 |
+
"step": 138
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.17,
|
| 1945 |
+
"learning_rate": 3.346844800613229e-05,
|
| 1946 |
+
"logits/chosen": -1.0398391485214233,
|
| 1947 |
+
"logits/rejected": -1.127720832824707,
|
| 1948 |
+
"logps/chosen": -310.21142578125,
|
| 1949 |
+
"logps/rejected": -683.2669677734375,
|
| 1950 |
+
"loss": 0.0,
|
| 1951 |
+
"rewards/accuracies": 1.0,
|
| 1952 |
+
"rewards/chosen": -6.429044723510742,
|
| 1953 |
+
"rewards/margins": 19.585290908813477,
|
| 1954 |
+
"rewards/rejected": -26.01433753967285,
|
| 1955 |
+
"step": 139
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.17,
|
| 1959 |
+
"learning_rate": 3.272542485937369e-05,
|
| 1960 |
+
"logits/chosen": -1.1515860557556152,
|
| 1961 |
+
"logits/rejected": -1.2663438320159912,
|
| 1962 |
+
"logps/chosen": -311.65545654296875,
|
| 1963 |
+
"logps/rejected": -805.0521850585938,
|
| 1964 |
+
"loss": 0.0,
|
| 1965 |
+
"rewards/accuracies": 1.0,
|
| 1966 |
+
"rewards/chosen": -8.75864315032959,
|
| 1967 |
+
"rewards/margins": 26.559768676757812,
|
| 1968 |
+
"rewards/rejected": -35.31841278076172,
|
| 1969 |
+
"step": 140
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.18,
|
| 1973 |
+
"learning_rate": 3.1974777650980735e-05,
|
| 1974 |
+
"logits/chosen": -1.6724903583526611,
|
| 1975 |
+
"logits/rejected": -1.2102152109146118,
|
| 1976 |
+
"logps/chosen": -216.79110717773438,
|
| 1977 |
+
"logps/rejected": -664.1553344726562,
|
| 1978 |
+
"loss": 0.0,
|
| 1979 |
+
"rewards/accuracies": 1.0,
|
| 1980 |
+
"rewards/chosen": -6.002614498138428,
|
| 1981 |
+
"rewards/margins": 23.148998260498047,
|
| 1982 |
+
"rewards/rejected": -29.151615142822266,
|
| 1983 |
+
"step": 141
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.18,
|
| 1987 |
+
"learning_rate": 3.121724717912138e-05,
|
| 1988 |
+
"logits/chosen": -1.2503281831741333,
|
| 1989 |
+
"logits/rejected": -1.4109008312225342,
|
| 1990 |
+
"logps/chosen": -257.79205322265625,
|
| 1991 |
+
"logps/rejected": -626.7386474609375,
|
| 1992 |
+
"loss": 0.0,
|
| 1993 |
+
"rewards/accuracies": 1.0,
|
| 1994 |
+
"rewards/chosen": -8.028193473815918,
|
| 1995 |
+
"rewards/margins": 20.606430053710938,
|
| 1996 |
+
"rewards/rejected": -28.63462257385254,
|
| 1997 |
+
"step": 142
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.18,
|
| 2001 |
+
"learning_rate": 3.045358103491357e-05,
|
| 2002 |
+
"logits/chosen": -1.6434905529022217,
|
| 2003 |
+
"logits/rejected": -1.5915906429290771,
|
| 2004 |
+
"logps/chosen": -333.8968811035156,
|
| 2005 |
+
"logps/rejected": -733.0169677734375,
|
| 2006 |
+
"loss": 0.0,
|
| 2007 |
+
"rewards/accuracies": 1.0,
|
| 2008 |
+
"rewards/chosen": -8.511116027832031,
|
| 2009 |
+
"rewards/margins": 27.476917266845703,
|
| 2010 |
+
"rewards/rejected": -35.988033294677734,
|
| 2011 |
+
"step": 143
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.18,
|
| 2015 |
+
"learning_rate": 2.9684532864643122e-05,
|
| 2016 |
+
"logits/chosen": -1.4675549268722534,
|
| 2017 |
+
"logits/rejected": -1.9043865203857422,
|
| 2018 |
+
"logps/chosen": -309.0293273925781,
|
| 2019 |
+
"logps/rejected": -591.2290649414062,
|
| 2020 |
+
"loss": 0.0,
|
| 2021 |
+
"rewards/accuracies": 1.0,
|
| 2022 |
+
"rewards/chosen": -8.682902336120605,
|
| 2023 |
+
"rewards/margins": 18.137462615966797,
|
| 2024 |
+
"rewards/rejected": -26.820363998413086,
|
| 2025 |
+
"step": 144
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.18,
|
| 2029 |
+
"learning_rate": 2.8910861626005776e-05,
|
| 2030 |
+
"logits/chosen": -1.7422269582748413,
|
| 2031 |
+
"logits/rejected": -1.2487080097198486,
|
| 2032 |
+
"logps/chosen": -260.97528076171875,
|
| 2033 |
+
"logps/rejected": -634.4310302734375,
|
| 2034 |
+
"loss": 0.0,
|
| 2035 |
+
"rewards/accuracies": 1.0,
|
| 2036 |
+
"rewards/chosen": -7.767115592956543,
|
| 2037 |
+
"rewards/margins": 18.91771697998047,
|
| 2038 |
+
"rewards/rejected": -26.684831619262695,
|
| 2039 |
+
"step": 145
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.18,
|
| 2043 |
+
"learning_rate": 2.8133330839107608e-05,
|
| 2044 |
+
"logits/chosen": -1.277583122253418,
|
| 2045 |
+
"logits/rejected": -1.6845176219940186,
|
| 2046 |
+
"logps/chosen": -352.37738037109375,
|
| 2047 |
+
"logps/rejected": -807.046875,
|
| 2048 |
+
"loss": 0.0,
|
| 2049 |
+
"rewards/accuracies": 1.0,
|
| 2050 |
+
"rewards/chosen": -10.428441047668457,
|
| 2051 |
+
"rewards/margins": 24.504329681396484,
|
| 2052 |
+
"rewards/rejected": -34.932769775390625,
|
| 2053 |
+
"step": 146
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 0.18,
|
| 2057 |
+
"learning_rate": 2.7352707832962865e-05,
|
| 2058 |
+
"logits/chosen": -1.4795787334442139,
|
| 2059 |
+
"logits/rejected": -1.813377857208252,
|
| 2060 |
+
"logps/chosen": -331.2845153808594,
|
| 2061 |
+
"logps/rejected": -801.8402099609375,
|
| 2062 |
+
"loss": 0.0001,
|
| 2063 |
+
"rewards/accuracies": 1.0,
|
| 2064 |
+
"rewards/chosen": -10.046164512634277,
|
| 2065 |
+
"rewards/margins": 27.718435287475586,
|
| 2066 |
+
"rewards/rejected": -37.76459884643555,
|
| 2067 |
+
"step": 147
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.18,
|
| 2071 |
+
"learning_rate": 2.656976298823284e-05,
|
| 2072 |
+
"logits/chosen": -1.283496379852295,
|
| 2073 |
+
"logits/rejected": -1.0843333005905151,
|
| 2074 |
+
"logps/chosen": -287.734619140625,
|
| 2075 |
+
"logps/rejected": -661.9713745117188,
|
| 2076 |
+
"loss": 0.0,
|
| 2077 |
+
"rewards/accuracies": 1.0,
|
| 2078 |
+
"rewards/chosen": -8.094979286193848,
|
| 2079 |
+
"rewards/margins": 21.19150161743164,
|
| 2080 |
+
"rewards/rejected": -29.286481857299805,
|
| 2081 |
+
"step": 148
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.19,
|
| 2085 |
+
"learning_rate": 2.578526897695321e-05,
|
| 2086 |
+
"logits/chosen": -1.5580577850341797,
|
| 2087 |
+
"logits/rejected": -1.6201229095458984,
|
| 2088 |
+
"logps/chosen": -301.7952880859375,
|
| 2089 |
+
"logps/rejected": -713.572998046875,
|
| 2090 |
+
"loss": 0.0,
|
| 2091 |
+
"rewards/accuracies": 1.0,
|
| 2092 |
+
"rewards/chosen": -7.939479827880859,
|
| 2093 |
+
"rewards/margins": 23.91385269165039,
|
| 2094 |
+
"rewards/rejected": -31.85333251953125,
|
| 2095 |
+
"step": 149
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 0.19,
|
| 2099 |
+
"learning_rate": 2.5e-05,
|
| 2100 |
+
"logits/chosen": -1.5227057933807373,
|
| 2101 |
+
"logits/rejected": -1.5960218906402588,
|
| 2102 |
+
"logps/chosen": -350.0317077636719,
|
| 2103 |
+
"logps/rejected": -813.5361938476562,
|
| 2104 |
+
"loss": 0.0,
|
| 2105 |
+
"rewards/accuracies": 1.0,
|
| 2106 |
+
"rewards/chosen": -9.528999328613281,
|
| 2107 |
+
"rewards/margins": 25.71453094482422,
|
| 2108 |
+
"rewards/rejected": -35.2435302734375,
|
| 2109 |
+
"step": 150
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.19,
|
| 2113 |
+
"learning_rate": 2.4214731023046793e-05,
|
| 2114 |
+
"logits/chosen": -1.5164449214935303,
|
| 2115 |
+
"logits/rejected": -1.3943390846252441,
|
| 2116 |
+
"logps/chosen": -262.6053466796875,
|
| 2117 |
+
"logps/rejected": -667.7622680664062,
|
| 2118 |
+
"loss": 0.0,
|
| 2119 |
+
"rewards/accuracies": 1.0,
|
| 2120 |
+
"rewards/chosen": -6.690636157989502,
|
| 2121 |
+
"rewards/margins": 21.432931900024414,
|
| 2122 |
+
"rewards/rejected": -28.123567581176758,
|
| 2123 |
+
"step": 151
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 0.19,
|
| 2127 |
+
"learning_rate": 2.3430237011767167e-05,
|
| 2128 |
+
"logits/chosen": -1.4828455448150635,
|
| 2129 |
+
"logits/rejected": -1.2360996007919312,
|
| 2130 |
+
"logps/chosen": -435.7225036621094,
|
| 2131 |
+
"logps/rejected": -812.917724609375,
|
| 2132 |
+
"loss": 0.0,
|
| 2133 |
+
"rewards/accuracies": 1.0,
|
| 2134 |
+
"rewards/chosen": -10.94736385345459,
|
| 2135 |
+
"rewards/margins": 23.513948440551758,
|
| 2136 |
+
"rewards/rejected": -34.46131134033203,
|
| 2137 |
+
"step": 152
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 0.19,
|
| 2141 |
+
"learning_rate": 2.2647292167037144e-05,
|
| 2142 |
+
"logits/chosen": -1.7452964782714844,
|
| 2143 |
+
"logits/rejected": -1.3689957857131958,
|
| 2144 |
+
"logps/chosen": -338.7836608886719,
|
| 2145 |
+
"logps/rejected": -629.8165283203125,
|
| 2146 |
+
"loss": 0.0089,
|
| 2147 |
+
"rewards/accuracies": 1.0,
|
| 2148 |
+
"rewards/chosen": -9.47596549987793,
|
| 2149 |
+
"rewards/margins": 18.36334228515625,
|
| 2150 |
+
"rewards/rejected": -27.839309692382812,
|
| 2151 |
+
"step": 153
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.19,
|
| 2155 |
+
"learning_rate": 2.186666916089239e-05,
|
| 2156 |
+
"logits/chosen": -1.1591405868530273,
|
| 2157 |
+
"logits/rejected": -1.1171772480010986,
|
| 2158 |
+
"logps/chosen": -251.10696411132812,
|
| 2159 |
+
"logps/rejected": -677.5640258789062,
|
| 2160 |
+
"loss": 0.0,
|
| 2161 |
+
"rewards/accuracies": 1.0,
|
| 2162 |
+
"rewards/chosen": -7.3279805183410645,
|
| 2163 |
+
"rewards/margins": 23.134410858154297,
|
| 2164 |
+
"rewards/rejected": -30.462392807006836,
|
| 2165 |
+
"step": 154
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 0.19,
|
| 2169 |
+
"learning_rate": 2.1089138373994223e-05,
|
| 2170 |
+
"logits/chosen": -1.800222635269165,
|
| 2171 |
+
"logits/rejected": -1.1949923038482666,
|
| 2172 |
+
"logps/chosen": -363.82037353515625,
|
| 2173 |
+
"logps/rejected": -834.6056518554688,
|
| 2174 |
+
"loss": 0.0001,
|
| 2175 |
+
"rewards/accuracies": 1.0,
|
| 2176 |
+
"rewards/chosen": -10.19363784790039,
|
| 2177 |
+
"rewards/margins": 27.392547607421875,
|
| 2178 |
+
"rewards/rejected": -37.586185455322266,
|
| 2179 |
+
"step": 155
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 0.19,
|
| 2183 |
+
"learning_rate": 2.031546713535688e-05,
|
| 2184 |
+
"logits/chosen": -1.80062735080719,
|
| 2185 |
+
"logits/rejected": -1.8262879848480225,
|
| 2186 |
+
"logps/chosen": -405.14239501953125,
|
| 2187 |
+
"logps/rejected": -789.0555419921875,
|
| 2188 |
+
"loss": 0.0,
|
| 2189 |
+
"rewards/accuracies": 1.0,
|
| 2190 |
+
"rewards/chosen": -10.376029968261719,
|
| 2191 |
+
"rewards/margins": 25.058115005493164,
|
| 2192 |
+
"rewards/rejected": -35.43414306640625,
|
| 2193 |
+
"step": 156
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.2,
|
| 2197 |
+
"learning_rate": 1.9546418965086442e-05,
|
| 2198 |
+
"logits/chosen": -1.6067843437194824,
|
| 2199 |
+
"logits/rejected": -1.5892972946166992,
|
| 2200 |
+
"logps/chosen": -300.63055419921875,
|
| 2201 |
+
"logps/rejected": -763.7952880859375,
|
| 2202 |
+
"loss": 0.0,
|
| 2203 |
+
"rewards/accuracies": 1.0,
|
| 2204 |
+
"rewards/chosen": -8.260177612304688,
|
| 2205 |
+
"rewards/margins": 28.313352584838867,
|
| 2206 |
+
"rewards/rejected": -36.57353210449219,
|
| 2207 |
+
"step": 157
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.2,
|
| 2211 |
+
"learning_rate": 1.8782752820878634e-05,
|
| 2212 |
+
"logits/chosen": -1.0016999244689941,
|
| 2213 |
+
"logits/rejected": -1.4391802549362183,
|
| 2214 |
+
"logps/chosen": -352.371826171875,
|
| 2215 |
+
"logps/rejected": -763.2830810546875,
|
| 2216 |
+
"loss": 0.0,
|
| 2217 |
+
"rewards/accuracies": 1.0,
|
| 2218 |
+
"rewards/chosen": -9.006278991699219,
|
| 2219 |
+
"rewards/margins": 22.58731460571289,
|
| 2220 |
+
"rewards/rejected": -31.593591690063477,
|
| 2221 |
+
"step": 158
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.2,
|
| 2225 |
+
"learning_rate": 1.802522234901927e-05,
|
| 2226 |
+
"logits/chosen": -1.1005672216415405,
|
| 2227 |
+
"logits/rejected": -1.2123079299926758,
|
| 2228 |
+
"logps/chosen": -329.65667724609375,
|
| 2229 |
+
"logps/rejected": -706.2532348632812,
|
| 2230 |
+
"loss": 0.0,
|
| 2231 |
+
"rewards/accuracies": 1.0,
|
| 2232 |
+
"rewards/chosen": -7.16185998916626,
|
| 2233 |
+
"rewards/margins": 22.066129684448242,
|
| 2234 |
+
"rewards/rejected": -29.22798728942871,
|
| 2235 |
+
"step": 159
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.2,
|
| 2239 |
+
"learning_rate": 1.7274575140626318e-05,
|
| 2240 |
+
"logits/chosen": -1.442047119140625,
|
| 2241 |
+
"logits/rejected": -1.6045359373092651,
|
| 2242 |
+
"logps/chosen": -415.3704833984375,
|
| 2243 |
+
"logps/rejected": -721.92724609375,
|
| 2244 |
+
"loss": 0.0,
|
| 2245 |
+
"rewards/accuracies": 1.0,
|
| 2246 |
+
"rewards/chosen": -6.923215389251709,
|
| 2247 |
+
"rewards/margins": 19.695026397705078,
|
| 2248 |
+
"rewards/rejected": -26.618240356445312,
|
| 2249 |
+
"step": 160
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.2,
|
| 2253 |
+
"learning_rate": 1.6531551993867717e-05,
|
| 2254 |
+
"logits/chosen": -1.3954782485961914,
|
| 2255 |
+
"logits/rejected": -1.5011147260665894,
|
| 2256 |
+
"logps/chosen": -298.94036865234375,
|
| 2257 |
+
"logps/rejected": -701.531982421875,
|
| 2258 |
+
"loss": 0.0002,
|
| 2259 |
+
"rewards/accuracies": 1.0,
|
| 2260 |
+
"rewards/chosen": -7.553921699523926,
|
| 2261 |
+
"rewards/margins": 21.85742950439453,
|
| 2262 |
+
"rewards/rejected": -29.41135025024414,
|
| 2263 |
+
"step": 161
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.2,
|
| 2267 |
+
"learning_rate": 1.5796886182883053e-05,
|
| 2268 |
+
"logits/chosen": -1.3542156219482422,
|
| 2269 |
+
"logits/rejected": -2.0877578258514404,
|
| 2270 |
+
"logps/chosen": -350.46441650390625,
|
| 2271 |
+
"logps/rejected": -648.8909912109375,
|
| 2272 |
+
"loss": 0.0,
|
| 2273 |
+
"rewards/accuracies": 1.0,
|
| 2274 |
+
"rewards/chosen": -9.726862907409668,
|
| 2275 |
+
"rewards/margins": 18.973695755004883,
|
| 2276 |
+
"rewards/rejected": -28.700557708740234,
|
| 2277 |
+
"step": 162
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.2,
|
| 2281 |
+
"learning_rate": 1.5071302734130489e-05,
|
| 2282 |
+
"logits/chosen": -1.6200387477874756,
|
| 2283 |
+
"logits/rejected": -1.3909564018249512,
|
| 2284 |
+
"logps/chosen": -253.26058959960938,
|
| 2285 |
+
"logps/rejected": -643.187255859375,
|
| 2286 |
+
"loss": 0.0002,
|
| 2287 |
+
"rewards/accuracies": 1.0,
|
| 2288 |
+
"rewards/chosen": -7.470417022705078,
|
| 2289 |
+
"rewards/margins": 19.356704711914062,
|
| 2290 |
+
"rewards/rejected": -26.82712173461914,
|
| 2291 |
+
"step": 163
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 0.2,
|
| 2295 |
+
"learning_rate": 1.4355517710873184e-05,
|
| 2296 |
+
"logits/chosen": -1.630284070968628,
|
| 2297 |
+
"logits/rejected": -1.851732611656189,
|
| 2298 |
+
"logps/chosen": -357.91229248046875,
|
| 2299 |
+
"logps/rejected": -663.6168212890625,
|
| 2300 |
+
"loss": 0.0001,
|
| 2301 |
+
"rewards/accuracies": 1.0,
|
| 2302 |
+
"rewards/chosen": -9.298179626464844,
|
| 2303 |
+
"rewards/margins": 19.182233810424805,
|
| 2304 |
+
"rewards/rejected": -28.480411529541016,
|
| 2305 |
+
"step": 164
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.21,
|
| 2309 |
+
"learning_rate": 1.3650237506511331e-05,
|
| 2310 |
+
"logits/chosen": -1.681291103363037,
|
| 2311 |
+
"logits/rejected": -1.6187834739685059,
|
| 2312 |
+
"logps/chosen": -228.45509338378906,
|
| 2313 |
+
"logps/rejected": -587.3668823242188,
|
| 2314 |
+
"loss": 0.0,
|
| 2315 |
+
"rewards/accuracies": 1.0,
|
| 2316 |
+
"rewards/chosen": -7.012322425842285,
|
| 2317 |
+
"rewards/margins": 18.644020080566406,
|
| 2318 |
+
"rewards/rejected": -25.656343460083008,
|
| 2319 |
+
"step": 165
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.21,
|
| 2323 |
+
"learning_rate": 1.2956158147457115e-05,
|
| 2324 |
+
"logits/chosen": -1.7261227369308472,
|
| 2325 |
+
"logits/rejected": -1.3109016418457031,
|
| 2326 |
+
"logps/chosen": -337.2093505859375,
|
| 2327 |
+
"logps/rejected": -821.0774536132812,
|
| 2328 |
+
"loss": 0.0,
|
| 2329 |
+
"rewards/accuracies": 1.0,
|
| 2330 |
+
"rewards/chosen": -9.458036422729492,
|
| 2331 |
+
"rewards/margins": 26.85298728942871,
|
| 2332 |
+
"rewards/rejected": -36.3110237121582,
|
| 2333 |
+
"step": 166
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 0.21,
|
| 2337 |
+
"learning_rate": 1.2273964606240718e-05,
|
| 2338 |
+
"logits/chosen": -1.6195597648620605,
|
| 2339 |
+
"logits/rejected": -1.2597343921661377,
|
| 2340 |
+
"logps/chosen": -377.4043884277344,
|
| 2341 |
+
"logps/rejected": -714.3167724609375,
|
| 2342 |
+
"loss": 0.0,
|
| 2343 |
+
"rewards/accuracies": 1.0,
|
| 2344 |
+
"rewards/chosen": -7.308032035827637,
|
| 2345 |
+
"rewards/margins": 21.946781158447266,
|
| 2346 |
+
"rewards/rejected": -29.25481414794922,
|
| 2347 |
+
"step": 167
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.21,
|
| 2351 |
+
"learning_rate": 1.1604330125525079e-05,
|
| 2352 |
+
"logits/chosen": -1.6287131309509277,
|
| 2353 |
+
"logits/rejected": -1.2217836380004883,
|
| 2354 |
+
"logps/chosen": -302.2106628417969,
|
| 2355 |
+
"logps/rejected": -777.7568969726562,
|
| 2356 |
+
"loss": 0.0,
|
| 2357 |
+
"rewards/accuracies": 1.0,
|
| 2358 |
+
"rewards/chosen": -7.82155704498291,
|
| 2359 |
+
"rewards/margins": 26.31945037841797,
|
| 2360 |
+
"rewards/rejected": -34.14101028442383,
|
| 2361 |
+
"step": 168
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.21,
|
| 2365 |
+
"learning_rate": 1.0947915553696742e-05,
|
| 2366 |
+
"logits/chosen": -1.947331428527832,
|
| 2367 |
+
"logits/rejected": -1.336834192276001,
|
| 2368 |
+
"logps/chosen": -336.48870849609375,
|
| 2369 |
+
"logps/rejected": -743.7005004882812,
|
| 2370 |
+
"loss": 0.0,
|
| 2371 |
+
"rewards/accuracies": 1.0,
|
| 2372 |
+
"rewards/chosen": -8.903692245483398,
|
| 2373 |
+
"rewards/margins": 23.143495559692383,
|
| 2374 |
+
"rewards/rejected": -32.04718780517578,
|
| 2375 |
+
"step": 169
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 0.21,
|
| 2379 |
+
"learning_rate": 1.0305368692688174e-05,
|
| 2380 |
+
"logits/chosen": -1.1448383331298828,
|
| 2381 |
+
"logits/rejected": -1.5337557792663574,
|
| 2382 |
+
"logps/chosen": -329.9434814453125,
|
| 2383 |
+
"logps/rejected": -650.752197265625,
|
| 2384 |
+
"loss": 0.0001,
|
| 2385 |
+
"rewards/accuracies": 1.0,
|
| 2386 |
+
"rewards/chosen": -8.674652099609375,
|
| 2387 |
+
"rewards/margins": 20.816728591918945,
|
| 2388 |
+
"rewards/rejected": -29.49138069152832,
|
| 2389 |
+
"step": 170
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 0.21,
|
| 2393 |
+
"learning_rate": 9.677323658675594e-06,
|
| 2394 |
+
"logits/chosen": -1.6498054265975952,
|
| 2395 |
+
"logits/rejected": -1.8353989124298096,
|
| 2396 |
+
"logps/chosen": -380.84136962890625,
|
| 2397 |
+
"logps/rejected": -634.1046142578125,
|
| 2398 |
+
"loss": 0.0001,
|
| 2399 |
+
"rewards/accuracies": 1.0,
|
| 2400 |
+
"rewards/chosen": -6.794053077697754,
|
| 2401 |
+
"rewards/margins": 17.891712188720703,
|
| 2402 |
+
"rewards/rejected": -24.68576431274414,
|
| 2403 |
+
"step": 171
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.21,
|
| 2407 |
+
"learning_rate": 9.064400256282757e-06,
|
| 2408 |
+
"logits/chosen": -1.6227014064788818,
|
| 2409 |
+
"logits/rejected": -1.2250831127166748,
|
| 2410 |
+
"logps/chosen": -291.1907043457031,
|
| 2411 |
+
"logps/rejected": -713.7255859375,
|
| 2412 |
+
"loss": 0.0,
|
| 2413 |
+
"rewards/accuracies": 1.0,
|
| 2414 |
+
"rewards/chosen": -9.592133522033691,
|
| 2415 |
+
"rewards/margins": 23.09697723388672,
|
| 2416 |
+
"rewards/rejected": -32.689109802246094,
|
| 2417 |
+
"step": 172
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.22,
|
| 2421 |
+
"learning_rate": 8.467203366908707e-06,
|
| 2422 |
+
"logits/chosen": -1.8008649349212646,
|
| 2423 |
+
"logits/rejected": -1.8741885423660278,
|
| 2424 |
+
"logps/chosen": -314.5689697265625,
|
| 2425 |
+
"logps/rejected": -654.071533203125,
|
| 2426 |
+
"loss": 0.0004,
|
| 2427 |
+
"rewards/accuracies": 1.0,
|
| 2428 |
+
"rewards/chosen": -7.73466682434082,
|
| 2429 |
+
"rewards/margins": 20.07561683654785,
|
| 2430 |
+
"rewards/rejected": -27.810285568237305,
|
| 2431 |
+
"step": 173
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.22,
|
| 2435 |
+
"learning_rate": 7.886322351782783e-06,
|
| 2436 |
+
"logits/chosen": -1.5283660888671875,
|
| 2437 |
+
"logits/rejected": -1.358813762664795,
|
| 2438 |
+
"logps/chosen": -236.36203002929688,
|
| 2439 |
+
"logps/rejected": -611.1893920898438,
|
| 2440 |
+
"loss": 0.0001,
|
| 2441 |
+
"rewards/accuracies": 1.0,
|
| 2442 |
+
"rewards/chosen": -7.122480869293213,
|
| 2443 |
+
"rewards/margins": 21.218101501464844,
|
| 2444 |
+
"rewards/rejected": -28.3405818939209,
|
| 2445 |
+
"step": 174
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.22,
|
| 2449 |
+
"learning_rate": 7.3223304703363135e-06,
|
| 2450 |
+
"logits/chosen": -1.43361234664917,
|
| 2451 |
+
"logits/rejected": -1.4089072942733765,
|
| 2452 |
+
"logps/chosen": -317.1131286621094,
|
| 2453 |
+
"logps/rejected": -796.5771484375,
|
| 2454 |
+
"loss": 0.0,
|
| 2455 |
+
"rewards/accuracies": 1.0,
|
| 2456 |
+
"rewards/chosen": -8.25259780883789,
|
| 2457 |
+
"rewards/margins": 25.549585342407227,
|
| 2458 |
+
"rewards/rejected": -33.802181243896484,
|
| 2459 |
+
"step": 175
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 0.22,
|
| 2463 |
+
"learning_rate": 6.775784314464717e-06,
|
| 2464 |
+
"logits/chosen": -1.5645939111709595,
|
| 2465 |
+
"logits/rejected": -1.5477921962738037,
|
| 2466 |
+
"logps/chosen": -406.3544921875,
|
| 2467 |
+
"logps/rejected": -669.9266357421875,
|
| 2468 |
+
"loss": 0.0,
|
| 2469 |
+
"rewards/accuracies": 1.0,
|
| 2470 |
+
"rewards/chosen": -7.228372573852539,
|
| 2471 |
+
"rewards/margins": 19.12314796447754,
|
| 2472 |
+
"rewards/rejected": -26.351518630981445,
|
| 2473 |
+
"step": 176
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.22,
|
| 2477 |
+
"learning_rate": 6.247223259238511e-06,
|
| 2478 |
+
"logits/chosen": -1.9002436399459839,
|
| 2479 |
+
"logits/rejected": -1.8102644681930542,
|
| 2480 |
+
"logps/chosen": -274.9958190917969,
|
| 2481 |
+
"logps/rejected": -686.9537963867188,
|
| 2482 |
+
"loss": 0.0,
|
| 2483 |
+
"rewards/accuracies": 1.0,
|
| 2484 |
+
"rewards/chosen": -8.03482723236084,
|
| 2485 |
+
"rewards/margins": 25.2858943939209,
|
| 2486 |
+
"rewards/rejected": -33.320716857910156,
|
| 2487 |
+
"step": 177
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.22,
|
| 2491 |
+
"learning_rate": 5.737168930605272e-06,
|
| 2492 |
+
"logits/chosen": -1.7001311779022217,
|
| 2493 |
+
"logits/rejected": -1.597381353378296,
|
| 2494 |
+
"logps/chosen": -380.9153137207031,
|
| 2495 |
+
"logps/rejected": -645.8336181640625,
|
| 2496 |
+
"loss": 0.0,
|
| 2497 |
+
"rewards/accuracies": 1.0,
|
| 2498 |
+
"rewards/chosen": -9.033845901489258,
|
| 2499 |
+
"rewards/margins": 17.564502716064453,
|
| 2500 |
+
"rewards/rejected": -26.598346710205078,
|
| 2501 |
+
"step": 178
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.22,
|
| 2505 |
+
"learning_rate": 5.24612469060774e-06,
|
| 2506 |
+
"logits/chosen": -1.4437607526779175,
|
| 2507 |
+
"logits/rejected": -1.4907622337341309,
|
| 2508 |
+
"logps/chosen": -279.7784729003906,
|
| 2509 |
+
"logps/rejected": -584.3414916992188,
|
| 2510 |
+
"loss": 0.0,
|
| 2511 |
+
"rewards/accuracies": 1.0,
|
| 2512 |
+
"rewards/chosen": -8.12205696105957,
|
| 2513 |
+
"rewards/margins": 17.455833435058594,
|
| 2514 |
+
"rewards/rejected": -25.577890396118164,
|
| 2515 |
+
"step": 179
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.22,
|
| 2519 |
+
"learning_rate": 4.7745751406263165e-06,
|
| 2520 |
+
"logits/chosen": -1.514746904373169,
|
| 2521 |
+
"logits/rejected": -1.4715900421142578,
|
| 2522 |
+
"logps/chosen": -302.8192138671875,
|
| 2523 |
+
"logps/rejected": -739.8018188476562,
|
| 2524 |
+
"loss": 0.0,
|
| 2525 |
+
"rewards/accuracies": 1.0,
|
| 2526 |
+
"rewards/chosen": -7.81146240234375,
|
| 2527 |
+
"rewards/margins": 23.765052795410156,
|
| 2528 |
+
"rewards/rejected": -31.576513290405273,
|
| 2529 |
+
"step": 180
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.23,
|
| 2533 |
+
"learning_rate": 4.322985643135952e-06,
|
| 2534 |
+
"logits/chosen": -1.4820040464401245,
|
| 2535 |
+
"logits/rejected": -1.354321002960205,
|
| 2536 |
+
"logps/chosen": -244.31655883789062,
|
| 2537 |
+
"logps/rejected": -604.16015625,
|
| 2538 |
+
"loss": 0.0,
|
| 2539 |
+
"rewards/accuracies": 1.0,
|
| 2540 |
+
"rewards/chosen": -6.212098121643066,
|
| 2541 |
+
"rewards/margins": 21.11400032043457,
|
| 2542 |
+
"rewards/rejected": -27.32609748840332,
|
| 2543 |
+
"step": 181
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.23,
|
| 2547 |
+
"learning_rate": 3.891801862449629e-06,
|
| 2548 |
+
"logits/chosen": -1.1967613697052002,
|
| 2549 |
+
"logits/rejected": -1.6249971389770508,
|
| 2550 |
+
"logps/chosen": -297.40936279296875,
|
| 2551 |
+
"logps/rejected": -727.44091796875,
|
| 2552 |
+
"loss": 0.0007,
|
| 2553 |
+
"rewards/accuracies": 1.0,
|
| 2554 |
+
"rewards/chosen": -8.703887939453125,
|
| 2555 |
+
"rewards/margins": 24.147249221801758,
|
| 2556 |
+
"rewards/rejected": -32.851139068603516,
|
| 2557 |
+
"step": 182
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.23,
|
| 2561 |
+
"learning_rate": 3.4814493249014116e-06,
|
| 2562 |
+
"logits/chosen": -1.896475911140442,
|
| 2563 |
+
"logits/rejected": -1.8446130752563477,
|
| 2564 |
+
"logps/chosen": -347.37335205078125,
|
| 2565 |
+
"logps/rejected": -795.5216064453125,
|
| 2566 |
+
"loss": 0.0003,
|
| 2567 |
+
"rewards/accuracies": 1.0,
|
| 2568 |
+
"rewards/chosen": -7.598814964294434,
|
| 2569 |
+
"rewards/margins": 22.283912658691406,
|
| 2570 |
+
"rewards/rejected": -29.882726669311523,
|
| 2571 |
+
"step": 183
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.23,
|
| 2575 |
+
"learning_rate": 3.092332998903416e-06,
|
| 2576 |
+
"logits/chosen": -1.5609359741210938,
|
| 2577 |
+
"logits/rejected": -1.3855328559875488,
|
| 2578 |
+
"logps/chosen": -378.8002014160156,
|
| 2579 |
+
"logps/rejected": -675.4521484375,
|
| 2580 |
+
"loss": 0.0003,
|
| 2581 |
+
"rewards/accuracies": 1.0,
|
| 2582 |
+
"rewards/chosen": -9.198368072509766,
|
| 2583 |
+
"rewards/margins": 19.104259490966797,
|
| 2584 |
+
"rewards/rejected": -28.302629470825195,
|
| 2585 |
+
"step": 184
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 0.23,
|
| 2589 |
+
"learning_rate": 2.7248368952908053e-06,
|
| 2590 |
+
"logits/chosen": -1.6999045610427856,
|
| 2591 |
+
"logits/rejected": -1.468465805053711,
|
| 2592 |
+
"logps/chosen": -367.71343994140625,
|
| 2593 |
+
"logps/rejected": -818.8544311523438,
|
| 2594 |
+
"loss": 0.001,
|
| 2595 |
+
"rewards/accuracies": 1.0,
|
| 2596 |
+
"rewards/chosen": -7.415655136108398,
|
| 2597 |
+
"rewards/margins": 24.159696578979492,
|
| 2598 |
+
"rewards/rejected": -31.575353622436523,
|
| 2599 |
+
"step": 185
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 0.23,
|
| 2603 |
+
"learning_rate": 2.379323688349516e-06,
|
| 2604 |
+
"logits/chosen": -1.8641347885131836,
|
| 2605 |
+
"logits/rejected": -1.7553801536560059,
|
| 2606 |
+
"logps/chosen": -328.0618896484375,
|
| 2607 |
+
"logps/rejected": -807.257080078125,
|
| 2608 |
+
"loss": 0.0,
|
| 2609 |
+
"rewards/accuracies": 1.0,
|
| 2610 |
+
"rewards/chosen": -8.81100082397461,
|
| 2611 |
+
"rewards/margins": 22.967178344726562,
|
| 2612 |
+
"rewards/rejected": -31.778179168701172,
|
| 2613 |
+
"step": 186
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 0.23,
|
| 2617 |
+
"learning_rate": 2.0561343579004715e-06,
|
| 2618 |
+
"logits/chosen": -1.3991014957427979,
|
| 2619 |
+
"logits/rejected": -1.234046220779419,
|
| 2620 |
+
"logps/chosen": -348.1552734375,
|
| 2621 |
+
"logps/rejected": -707.0328979492188,
|
| 2622 |
+
"loss": 0.0,
|
| 2623 |
+
"rewards/accuracies": 1.0,
|
| 2624 |
+
"rewards/chosen": -8.648429870605469,
|
| 2625 |
+
"rewards/margins": 20.976667404174805,
|
| 2626 |
+
"rewards/rejected": -29.625097274780273,
|
| 2627 |
+
"step": 187
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.23,
|
| 2631 |
+
"learning_rate": 1.7555878527937164e-06,
|
| 2632 |
+
"logits/chosen": -1.6236505508422852,
|
| 2633 |
+
"logits/rejected": -1.8800938129425049,
|
| 2634 |
+
"logps/chosen": -319.60369873046875,
|
| 2635 |
+
"logps/rejected": -695.4336547851562,
|
| 2636 |
+
"loss": 0.0,
|
| 2637 |
+
"rewards/accuracies": 1.0,
|
| 2638 |
+
"rewards/chosen": -7.873907089233398,
|
| 2639 |
+
"rewards/margins": 24.007831573486328,
|
| 2640 |
+
"rewards/rejected": -31.88174057006836,
|
| 2641 |
+
"step": 188
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.24,
|
| 2645 |
+
"learning_rate": 1.4779807761443636e-06,
|
| 2646 |
+
"logits/chosen": -1.3024158477783203,
|
| 2647 |
+
"logits/rejected": -1.4170106649398804,
|
| 2648 |
+
"logps/chosen": -295.1216735839844,
|
| 2649 |
+
"logps/rejected": -666.1689453125,
|
| 2650 |
+
"loss": 0.0,
|
| 2651 |
+
"rewards/accuracies": 1.0,
|
| 2652 |
+
"rewards/chosen": -7.164243221282959,
|
| 2653 |
+
"rewards/margins": 19.821399688720703,
|
| 2654 |
+
"rewards/rejected": -26.98564338684082,
|
| 2655 |
+
"step": 189
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 0.24,
|
| 2659 |
+
"learning_rate": 1.2235870926211619e-06,
|
| 2660 |
+
"logits/chosen": -1.5904102325439453,
|
| 2661 |
+
"logits/rejected": -1.3836643695831299,
|
| 2662 |
+
"logps/chosen": -425.983642578125,
|
| 2663 |
+
"logps/rejected": -806.5205688476562,
|
| 2664 |
+
"loss": 0.0,
|
| 2665 |
+
"rewards/accuracies": 1.0,
|
| 2666 |
+
"rewards/chosen": -8.154449462890625,
|
| 2667 |
+
"rewards/margins": 24.296112060546875,
|
| 2668 |
+
"rewards/rejected": -32.4505615234375,
|
| 2669 |
+
"step": 190
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 0.24,
|
| 2673 |
+
"learning_rate": 9.926578580764234e-07,
|
| 2674 |
+
"logits/chosen": -1.5797605514526367,
|
| 2675 |
+
"logits/rejected": -1.5394536256790161,
|
| 2676 |
+
"logps/chosen": -266.3468017578125,
|
| 2677 |
+
"logps/rejected": -691.3595581054688,
|
| 2678 |
+
"loss": 0.0,
|
| 2679 |
+
"rewards/accuracies": 1.0,
|
| 2680 |
+
"rewards/chosen": -7.95579719543457,
|
| 2681 |
+
"rewards/margins": 20.96570587158203,
|
| 2682 |
+
"rewards/rejected": -28.9215030670166,
|
| 2683 |
+
"step": 191
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.24,
|
| 2687 |
+
"learning_rate": 7.854209717842231e-07,
|
| 2688 |
+
"logits/chosen": -1.2808557748794556,
|
| 2689 |
+
"logits/rejected": -1.168199896812439,
|
| 2690 |
+
"logps/chosen": -260.4226379394531,
|
| 2691 |
+
"logps/rejected": -559.0167846679688,
|
| 2692 |
+
"loss": 0.0,
|
| 2693 |
+
"rewards/accuracies": 1.0,
|
| 2694 |
+
"rewards/chosen": -6.135287284851074,
|
| 2695 |
+
"rewards/margins": 17.63001251220703,
|
| 2696 |
+
"rewards/rejected": -23.76529884338379,
|
| 2697 |
+
"step": 192
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.24,
|
| 2701 |
+
"learning_rate": 6.020809515313142e-07,
|
| 2702 |
+
"logits/chosen": -1.7369805574417114,
|
| 2703 |
+
"logits/rejected": -1.5230166912078857,
|
| 2704 |
+
"logps/chosen": -487.14990234375,
|
| 2705 |
+
"logps/rejected": -802.9796142578125,
|
| 2706 |
+
"loss": 0.0,
|
| 2707 |
+
"rewards/accuracies": 1.0,
|
| 2708 |
+
"rewards/chosen": -8.786242485046387,
|
| 2709 |
+
"rewards/margins": 19.918052673339844,
|
| 2710 |
+
"rewards/rejected": -28.70429229736328,
|
| 2711 |
+
"step": 193
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.24,
|
| 2715 |
+
"learning_rate": 4.4281873178278475e-07,
|
| 2716 |
+
"logits/chosen": -1.9259474277496338,
|
| 2717 |
+
"logits/rejected": -1.6450233459472656,
|
| 2718 |
+
"logps/chosen": -345.2897033691406,
|
| 2719 |
+
"logps/rejected": -656.6492919921875,
|
| 2720 |
+
"loss": 0.0,
|
| 2721 |
+
"rewards/accuracies": 1.0,
|
| 2722 |
+
"rewards/chosen": -8.629081726074219,
|
| 2723 |
+
"rewards/margins": 17.748437881469727,
|
| 2724 |
+
"rewards/rejected": -26.377517700195312,
|
| 2725 |
+
"step": 194
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 0.24,
|
| 2729 |
+
"learning_rate": 3.077914851215585e-07,
|
| 2730 |
+
"logits/chosen": -1.5199689865112305,
|
| 2731 |
+
"logits/rejected": -1.5939421653747559,
|
| 2732 |
+
"logps/chosen": -313.1384582519531,
|
| 2733 |
+
"logps/rejected": -633.854248046875,
|
| 2734 |
+
"loss": 0.0,
|
| 2735 |
+
"rewards/accuracies": 1.0,
|
| 2736 |
+
"rewards/chosen": -7.756176471710205,
|
| 2737 |
+
"rewards/margins": 19.526309967041016,
|
| 2738 |
+
"rewards/rejected": -27.282485961914062,
|
| 2739 |
+
"step": 195
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.24,
|
| 2743 |
+
"learning_rate": 1.9713246713805588e-07,
|
| 2744 |
+
"logits/chosen": -1.5103416442871094,
|
| 2745 |
+
"logits/rejected": -1.331744909286499,
|
| 2746 |
+
"logps/chosen": -355.4280700683594,
|
| 2747 |
+
"logps/rejected": -676.4717407226562,
|
| 2748 |
+
"loss": 0.0,
|
| 2749 |
+
"rewards/accuracies": 1.0,
|
| 2750 |
+
"rewards/chosen": -7.107924461364746,
|
| 2751 |
+
"rewards/margins": 20.338895797729492,
|
| 2752 |
+
"rewards/rejected": -27.446819305419922,
|
| 2753 |
+
"step": 196
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.25,
|
| 2757 |
+
"learning_rate": 1.109508849230001e-07,
|
| 2758 |
+
"logits/chosen": -1.657146692276001,
|
| 2759 |
+
"logits/rejected": -1.5730295181274414,
|
| 2760 |
+
"logps/chosen": -284.8887634277344,
|
| 2761 |
+
"logps/rejected": -743.6079711914062,
|
| 2762 |
+
"loss": 0.0001,
|
| 2763 |
+
"rewards/accuracies": 1.0,
|
| 2764 |
+
"rewards/chosen": -7.696743011474609,
|
| 2765 |
+
"rewards/margins": 24.68915367126465,
|
| 2766 |
+
"rewards/rejected": -32.38589859008789,
|
| 2767 |
+
"step": 197
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 0.25,
|
| 2771 |
+
"learning_rate": 4.9331789293211026e-08,
|
| 2772 |
+
"logits/chosen": -1.585029125213623,
|
| 2773 |
+
"logits/rejected": -1.5069087743759155,
|
| 2774 |
+
"logps/chosen": -341.88458251953125,
|
| 2775 |
+
"logps/rejected": -764.3339233398438,
|
| 2776 |
+
"loss": 0.0,
|
| 2777 |
+
"rewards/accuracies": 1.0,
|
| 2778 |
+
"rewards/chosen": -7.408685684204102,
|
| 2779 |
+
"rewards/margins": 23.2781982421875,
|
| 2780 |
+
"rewards/rejected": -30.686885833740234,
|
| 2781 |
+
"step": 198
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 0.25,
|
| 2785 |
+
"learning_rate": 1.233599085671e-08,
|
| 2786 |
+
"logits/chosen": -1.403904914855957,
|
| 2787 |
+
"logits/rejected": -1.301183819770813,
|
| 2788 |
+
"logps/chosen": -278.1524658203125,
|
| 2789 |
+
"logps/rejected": -659.3203125,
|
| 2790 |
+
"loss": 0.0,
|
| 2791 |
+
"rewards/accuracies": 1.0,
|
| 2792 |
+
"rewards/chosen": -6.859063625335693,
|
| 2793 |
+
"rewards/margins": 19.299898147583008,
|
| 2794 |
+
"rewards/rejected": -26.15896224975586,
|
| 2795 |
+
"step": 199
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 0.25,
|
| 2799 |
+
"learning_rate": 0.0,
|
| 2800 |
+
"logits/chosen": -1.8759424686431885,
|
| 2801 |
+
"logits/rejected": -1.6900622844696045,
|
| 2802 |
+
"logps/chosen": -227.09156799316406,
|
| 2803 |
+
"logps/rejected": -601.8648681640625,
|
| 2804 |
+
"loss": 0.0,
|
| 2805 |
+
"rewards/accuracies": 1.0,
|
| 2806 |
+
"rewards/chosen": -7.408708572387695,
|
| 2807 |
+
"rewards/margins": 20.892513275146484,
|
| 2808 |
+
"rewards/rejected": -28.301219940185547,
|
| 2809 |
+
"step": 200
|
| 2810 |
+
}
|
| 2811 |
+
],
|
| 2812 |
+
"logging_steps": 1,
|
| 2813 |
+
"max_steps": 200,
|
| 2814 |
+
"num_train_epochs": 1,
|
| 2815 |
+
"save_steps": 40,
|
| 2816 |
+
"total_flos": 0.0,
|
| 2817 |
+
"trial_name": null,
|
| 2818 |
+
"trial_params": null
|
| 2819 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b52ed983d3e7f22985f201323e2d10cf6955e6d68f6bfb834c007eae1cae02b3
|
| 3 |
+
size 4600
|