Commit
•
da28be2
1
Parent(s):
3641f67
Update README.md
Browse files
README.md
CHANGED
|
@@ -18,7 +18,7 @@ https://github.com/cg123/mergekit/tree/dare-tokenizer
|
|
| 18 |
|
| 19 |
It was quantized with exllamav2 on 200 rows (400K tokens) on a long Vicuna format chat, a single sci fi story and a single fantasy story. This should hopefully yield better chat performance than the default wikitext quantization.
|
| 20 |
|
| 21 |
-
Quantized to 4bpw, enough for **~
|
| 22 |
***
|
| 23 |
|
| 24 |
Merged with the following config, and the tokenizer from Yi Llamafied:
|
|
|
|
| 18 |
|
| 19 |
It was quantized with exllamav2 on 200 rows (400K tokens) on a long Vicuna format chat, a single sci fi story and a single fantasy story. This should hopefully yield better chat performance than the default wikitext quantization.
|
| 20 |
|
| 21 |
+
Quantized to 4bpw, enough for **~45K context on a 24GB GPU.**
|
| 22 |
***
|
| 23 |
|
| 24 |
Merged with the following config, and the tokenizer from Yi Llamafied:
|