Commit
•
5ebccd1
1
Parent(s):
d5d729d
Update training statistics
Browse files
README.md
CHANGED
|
@@ -2319,14 +2319,15 @@ See this repository for JSON files: https://github.com/bigscience-workshop/evalu
|
|
| 2319 |
|
| 2320 |
**Train-time Evaluation:**
|
| 2321 |
|
| 2322 |
-
|
| 2323 |
|
| 2324 |
-
- Training Loss:
|
| 2325 |
|
| 2326 |
-
- Validation Loss: 2.
|
| 2327 |
|
| 2328 |
-
- Perplexity:
|
| 2329 |
|
|
|
|
| 2330 |
|
| 2331 |
</details>
|
| 2332 |
|
|
|
|
| 2319 |
|
| 2320 |
**Train-time Evaluation:**
|
| 2321 |
|
| 2322 |
+
Final checkpoint after 95K steps:
|
| 2323 |
|
| 2324 |
+
- Training Loss: 1.939
|
| 2325 |
|
| 2326 |
+
- Validation Loss: 2.061
|
| 2327 |
|
| 2328 |
+
- Perplexity: 7.045
|
| 2329 |
|
| 2330 |
+
For more see: https://huggingface.co/bigscience/tr11-176B-ml-logs
|
| 2331 |
|
| 2332 |
</details>
|
| 2333 |
|