FP8 LLMs for vLLM
Collection
Accurate FP8 quantized models by Neural Magic, ready for use with vLLM!
•
10 items
•
Updated
•
14
Mixtral-8x7B-Instruct-v0.1-FP8
Model Overview
Mixtral-8x7B-Instruct-v0.1 quantized to FP8 weights and activations, ready for inference with vLLM >= 0.5.0.
Usage and Creation
Produced using AutoFP8 with calibration samples from ultrachat with block_sparse_moe.gate layers kept at original precision.
Evaluation
Open LLM Leaderboard evaluation scores
| Mixtral-8x7B-Instruct-v0.1 | Mixtral-8x7B-Instruct-v0.1-FP8 (this model) |
|
|---|---|---|
| arc-c 25-shot |
71.50 | 71.08 |
| hellaswag 10-shot |
87.53 | 87.38 |
| mmlu 5-shot |
70.33 | 70.00 |
| truthfulqa 0-shot |
64.79 | 64.20 |
| winogrande 5-shot |
82.40 | 82.40 |
| gsm8k 5-shot |
64.36 | 64.06 |
| Average Accuracy |
73.48 | 73.19 |
| Recovery | 100% | 99.61% |
- Downloads last month
- 88