Update README.md
Browse files
README.md
CHANGED
|
@@ -2,6 +2,7 @@
|
|
| 2 |
tags:
|
| 3 |
- fp8
|
| 4 |
- vllm
|
|
|
|
| 5 |
---
|
| 6 |
|
| 7 |
# Mixtral-8x22B-Instruct-v0.1-FP8
|
|
@@ -19,12 +20,12 @@ tags:
|
|
| 19 |
- **Version:** 1.0
|
| 20 |
- **Model Developers:** Neural Magic
|
| 21 |
|
| 22 |
-
Quantized version of [Mixtral-8x22B-Instruct-v0.1](mistralai/Mixtral-8x22B-Instruct-v0.1).
|
| 23 |
It achieves an average score of 78.47 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 79.15.
|
| 24 |
|
| 25 |
### Model Optimizations
|
| 26 |
|
| 27 |
-
This model was obtained by quantizing the weights and activations of [Mixtral-8x22B-Instruct-v0.1](mistralai/Mixtral-8x22B-Instruct-v0.1) to FP8 data type, ready for inference with vLLM >= 0.5.0.
|
| 28 |
This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%.
|
| 29 |
|
| 30 |
Only the weights and activations of the linear operators within transformers blocks are quantized. Symmetric per-tensor quantization is applied, in which a linear scaling per output dimension maps the FP8 representations of the quantized weights and activations.
|
|
|
|
| 2 |
tags:
|
| 3 |
- fp8
|
| 4 |
- vllm
|
| 5 |
+
license: apache-2.0
|
| 6 |
---
|
| 7 |
|
| 8 |
# Mixtral-8x22B-Instruct-v0.1-FP8
|
|
|
|
| 20 |
- **Version:** 1.0
|
| 21 |
- **Model Developers:** Neural Magic
|
| 22 |
|
| 23 |
+
Quantized version of [Mixtral-8x22B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1).
|
| 24 |
It achieves an average score of 78.47 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 79.15.
|
| 25 |
|
| 26 |
### Model Optimizations
|
| 27 |
|
| 28 |
+
This model was obtained by quantizing the weights and activations of [Mixtral-8x22B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1) to FP8 data type, ready for inference with vLLM >= 0.5.0.
|
| 29 |
This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%.
|
| 30 |
|
| 31 |
Only the weights and activations of the linear operators within transformers blocks are quantized. Symmetric per-tensor quantization is applied, in which a linear scaling per output dimension maps the FP8 representations of the quantized weights and activations.
|