Update README.md
Browse files
README.md
CHANGED
|
@@ -13,12 +13,14 @@ language:
|
|
| 13 |
## Model details
|
| 14 |
|
| 15 |
First attempt to quantize a 20B model so it can run on 16GB VRAM with the highest quality possible.
|
| 16 |
-
Quantized at 3.18bpw with hb 6
|
| 17 |
|
| 18 |
Perplexity:
|
| 19 |
|
| 20 |
Base = 6.4744
|
| 21 |
|
|
|
|
|
|
|
| 22 |
3.18 h6 = 6.5705
|
| 23 |
|
| 24 |
Dataset = [wikitext](https://huggingface.co/datasets/wikitext/resolve/refs%2Fconvert%2Fparquet/wikitext-2-v1/test/0000.parquet)
|
|
|
|
| 13 |
## Model details
|
| 14 |
|
| 15 |
First attempt to quantize a 20B model so it can run on 16GB VRAM with the highest quality possible.
|
| 16 |
+
Quantized at 3.18bpw with hb 6. 8.13bpw also available for those who want it (exl2 is very fast with flash-attention and the quality is (almost) the same with fp16.)
|
| 17 |
|
| 18 |
Perplexity:
|
| 19 |
|
| 20 |
Base = 6.4744
|
| 21 |
|
| 22 |
+
8bpw h8 = 6.4471
|
| 23 |
+
|
| 24 |
3.18 h6 = 6.5705
|
| 25 |
|
| 26 |
Dataset = [wikitext](https://huggingface.co/datasets/wikitext/resolve/refs%2Fconvert%2Fparquet/wikitext-2-v1/test/0000.parquet)
|