File size: 248 Bytes
fee9f43 |
1 2 3 4 5 6 7 8 9 10 11 |
## Mixtral-8x22B-Instruct-v0.1-FP8-v1
* Weights and activations are per-tensor quantized to float8_e4m3.
* Quantization with AutoFP8.
* Calibration dataset: Ultrachat (mgoin/ultrachat_2k)
* Samples: 2048
* Sequence length: 8192
## Evaluation
TBA |