Commit
•
1139e72
1
Parent(s):
282cbe3
Update README.md
Browse files
README.md
CHANGED
|
@@ -13,7 +13,7 @@ Official [AQLM](https://arxiv.org/abs/2401.06118) quantization of [CohereForAI/c
|
|
| 13 |
For this quantization, we used 1 codebook of 16 bits.
|
| 14 |
|
| 15 |
Results:
|
| 16 |
-
| Model | Quantization | MMLU | GSM8k | Model size, Gb |
|
| 17 |
|------|------|-------|------|------|
|
| 18 |
|CohereForAI/c4ai-command-r-v01| None |0.6755 | 0.6065 | 70.0 |
|
| 19 |
| | 1x16 | 0.5719 | 0.3760 | 12.7 |
|
|
|
|
| 13 |
For this quantization, we used 1 codebook of 16 bits.
|
| 14 |
|
| 15 |
Results:
|
| 16 |
+
| Model | Quantization | MMLU (5-shot) | GSM8k (8-shot) | Model size, Gb |
|
| 17 |
|------|------|-------|------|------|
|
| 18 |
|CohereForAI/c4ai-command-r-v01| None |0.6755 | 0.6065 | 70.0 |
|
| 19 |
| | 1x16 | 0.5719 | 0.3760 | 12.7 |
|