File size: 1,545 Bytes
2ae5589 98b631e 2ae5589 eb86576 3991529 eb86576 3991529 7b86662 eb86576 3991529 7b86662 eb86576 3991529 eb86576 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 |
---
tags:
- fp8
- vllm
---
# Meta-Llama-3-8B-Instruct-FP8
## Model Overview
Meta-Llama-3-8B-Instruct quantized to FP8 weights and activations using per-tensor quantization, ready for inference with vLLM >= 0.5.0.
## Usage and Creation
Produced using [AutoFP8 with calibration samples from ultrachat](https://github.com/neuralmagic/AutoFP8/blob/147fa4d9e1a90ef8a93f96fc7d9c33056ddc017a/example_dataset.py).
## Evaluation
### Open LLM Leaderboard evaluation scores
| | Meta-Llama-3-8B-Instruct | Meta-Llama-3-8B-Instruct-FP8<br>(this model) |
| :------------------: | :----------------------: | :------------------------------------------------: |
| arc-c<br>25-shot | 62.54 | 61.77 |
| hellaswag<br>10-shot | 78.83 | 78.56 |
| mmlu<br>5-shot | 66.60 | 66.27 |
| truthfulqa<br>0-shot | 52.44 | 52.35 |
| winogrande<br>5-shot | 75.93 | 76.4 |
| gsm8k<br>5-shot | 75.96 | 73.99 |
| **Average<br>Accuracy** | **68.71** | **68.22** |
| **Recovery** | **100%** | **99.28%** |
|