Update README.md
Browse files
README.md
CHANGED
|
@@ -12,7 +12,7 @@ tags:
|
|
| 12 |
- merge
|
| 13 |
license: llama3.3
|
| 14 |
---
|
| 15 |
-
**Upon further testing I found some logic issues
|
| 16 |
|
| 17 |
//User y-ryan discovered an issue where the model had invalid tensor.Shape for weights ([1, 8192]), raising following errors when loading with transformers, and fixed it here: [tmfi-us/Progenitor-V5-Final-LLaMa-70B](https://huggingface.co/tmfi-us/Progenitor-V5-Final-LLaMa-70B) I have no clue what is the reason, but despite that I was still able to use and even quant this model?! Testing the fixed version and this gave me different outputs too, with this version's stuff being good?. If anyone understands this I would love to hear about it.//
|
| 18 |
|
|
|
|
| 12 |
- merge
|
| 13 |
license: llama3.3
|
| 14 |
---
|
| 15 |
+
**Upon further testing I found some logic issues!! WEIGHTS ARE BROKEN**
|
| 16 |
|
| 17 |
//User y-ryan discovered an issue where the model had invalid tensor.Shape for weights ([1, 8192]), raising following errors when loading with transformers, and fixed it here: [tmfi-us/Progenitor-V5-Final-LLaMa-70B](https://huggingface.co/tmfi-us/Progenitor-V5-Final-LLaMa-70B) I have no clue what is the reason, but despite that I was still able to use and even quant this model?! Testing the fixed version and this gave me different outputs too, with this version's stuff being good?. If anyone understands this I would love to hear about it.//
|
| 18 |
|