Update README.md
Browse files
README.md
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
-
## Mixtral-8x22B-Instruct-v0.1-FP8-
|
| 2 |
|
| 3 |
* Weights and activations are per-tensor quantized to float8_e4m3.
|
| 4 |
-
* Quantization with AutoFP8.
|
| 5 |
* Calibration dataset: Ultrachat (mgoin/ultrachat_2k)
|
| 6 |
* Samples: 2048
|
| 7 |
* Sequence length: 8192
|
|
|
|
| 1 |
+
## Mixtral-8x22B-Instruct-v0.1-FP8-v2
|
| 2 |
|
| 3 |
* Weights and activations are per-tensor quantized to float8_e4m3.
|
| 4 |
+
* Quantization with AutoFP8 with updated activation scaling factor names.
|
| 5 |
* Calibration dataset: Ultrachat (mgoin/ultrachat_2k)
|
| 6 |
* Samples: 2048
|
| 7 |
* Sequence length: 8192
|