Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -24,7 +24,8 @@ python3 quantize_quark.py \
|
|
| 24 |
--kv_cache_dtype fp8 \
|
| 25 |
--num_calib_data 128 \
|
| 26 |
--model_export quark_safetensors \
|
| 27 |
-
--no_weight_matrix_merge
|
|
|
|
| 28 |
# If model size is too large for single GPU, please use multi GPU instead.
|
| 29 |
python3 quantize_quark.py \
|
| 30 |
--model_dir $MODEL_DIR \
|
|
@@ -34,7 +35,8 @@ python3 quantize_quark.py \
|
|
| 34 |
--num_calib_data 128 \
|
| 35 |
--model_export quark_safetensors \
|
| 36 |
--no_weight_matrix_merge \
|
| 37 |
-
--multi_gpu
|
|
|
|
| 38 |
```
|
| 39 |
## Deployment
|
| 40 |
Quark has its own export format and allows FP8 quantized models to be efficiently deployed using the vLLM backend(vLLM-compatible).
|
|
|
|
| 24 |
--kv_cache_dtype fp8 \
|
| 25 |
--num_calib_data 128 \
|
| 26 |
--model_export quark_safetensors \
|
| 27 |
+
--no_weight_matrix_merge \
|
| 28 |
+
--custom_mode fp8
|
| 29 |
# If model size is too large for single GPU, please use multi GPU instead.
|
| 30 |
python3 quantize_quark.py \
|
| 31 |
--model_dir $MODEL_DIR \
|
|
|
|
| 35 |
--num_calib_data 128 \
|
| 36 |
--model_export quark_safetensors \
|
| 37 |
--no_weight_matrix_merge \
|
| 38 |
+
--multi_gpu \
|
| 39 |
+
--custom_mode fp8
|
| 40 |
```
|
| 41 |
## Deployment
|
| 42 |
Quark has its own export format and allows FP8 quantized models to be efficiently deployed using the vLLM backend(vLLM-compatible).
|