new version
Browse files- config.json +1 -1
- trainer_state.json +0 -9
config.json
CHANGED
|
@@ -36,7 +36,7 @@
|
|
| 36 |
"torch_dtype": "bfloat16",
|
| 37 |
"transformers_version": "4.42.3",
|
| 38 |
"tune_mm_mlp_adapter": false,
|
| 39 |
-
"use_cache":
|
| 40 |
"use_mm_proj": true,
|
| 41 |
"use_qk_norm": false,
|
| 42 |
"vocab_size": 256000
|
|
|
|
| 36 |
"torch_dtype": "bfloat16",
|
| 37 |
"transformers_version": "4.42.3",
|
| 38 |
"tune_mm_mlp_adapter": false,
|
| 39 |
+
"use_cache": false,
|
| 40 |
"use_mm_proj": true,
|
| 41 |
"use_qk_norm": false,
|
| 42 |
"vocab_size": 256000
|
trainer_state.json
CHANGED
|
@@ -114030,15 +114030,6 @@
|
|
| 114030 |
"learning_rate": 0.0,
|
| 114031 |
"loss": 0.6631,
|
| 114032 |
"step": 16289
|
| 114033 |
-
},
|
| 114034 |
-
{
|
| 114035 |
-
"epoch": 0.9999693053807668,
|
| 114036 |
-
"step": 16289,
|
| 114037 |
-
"total_flos": 2.9484990053482496e+16,
|
| 114038 |
-
"train_loss": 0.8763824042031848,
|
| 114039 |
-
"train_runtime": 165677.34,
|
| 114040 |
-
"train_samples_per_second": 12.585,
|
| 114041 |
-
"train_steps_per_second": 0.098
|
| 114042 |
}
|
| 114043 |
],
|
| 114044 |
"logging_steps": 1.0,
|
|
|
|
| 114030 |
"learning_rate": 0.0,
|
| 114031 |
"loss": 0.6631,
|
| 114032 |
"step": 16289
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114033 |
}
|
| 114034 |
],
|
| 114035 |
"logging_steps": 1.0,
|