Update README.md
Browse files
README.md
CHANGED
|
@@ -6,7 +6,7 @@ base_model:
|
|
| 6 |
## Model Details
|
| 7 |
|
| 8 |
This model card is for mxfp4 quantization of [meta-llama/Llama-4-Scout-17B-16E-Instruct](https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct) based on [intel/auto-round](https://github.com/intel/auto-round).
|
| 9 |
-
The quantized model is not able to be published due to
|
| 10 |
|
| 11 |
## How to Use
|
| 12 |
|
|
@@ -14,14 +14,16 @@ The step-by-step README of quantization and evaluation can be found in [Intel Ne
|
|
| 14 |
|
| 15 |
## Evaluate Results
|
| 16 |
|
| 17 |
-
|
|
| 18 |
-
|
| 19 |
-
|
|
| 20 |
-
|
|
| 21 |
-
|
|
| 22 |
-
| gsm8k | vllm | 0.9090 | 0.9121 |
|
| 23 |
-
|
|
| 24 |
-
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
## Ethical Considerations and Limitations
|
| 27 |
|
|
|
|
| 6 |
## Model Details
|
| 7 |
|
| 8 |
This model card is for mxfp4 quantization of [meta-llama/Llama-4-Scout-17B-16E-Instruct](https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct) based on [intel/auto-round](https://github.com/intel/auto-round).
|
| 9 |
+
The quantized model is not able to be published due to license limitation. Please follow the INC example README to generate and evaluate the low precision model.
|
| 10 |
|
| 11 |
## How to Use
|
| 12 |
|
|
|
|
| 14 |
|
| 15 |
## Evaluate Results
|
| 16 |
|
| 17 |
+
| Task | backend | BF16 | MXFP4 |
|
| 18 |
+
|:-----------------:|:-------:|:--------:|:------:|
|
| 19 |
+
| hellaswag | vllm | 0.6389 | 0.6349 |
|
| 20 |
+
| piqa | vllm | 0.8156 | 0.8107 |
|
| 21 |
+
| mmlu | vllm | 0.7997 | 0.7921 |
|
| 22 |
+
| gsm8k(strict) | vllm | 0.9090 | 0.9121 |
|
| 23 |
+
| chartqa(relaxed) | vllm | 0.8900 | 0.8884 |
|
| 24 |
+
| mmmu_val | vllm | 0.5989 | 0.5844 |
|
| 25 |
+
| average | vllm | 0.7754 | 0.7704 |
|
| 26 |
+
|
| 27 |
|
| 28 |
## Ethical Considerations and Limitations
|
| 29 |
|