Update README.md (#7)

- Update README.md (6bb726f3f9635ab01ee3fe20d228e8cfa1c8d445)


Co-authored-by: haoyanli <[email protected]>

README.md

@@ -8,7 +8,7 @@ base_model: mistralai/Mixtral-8x7B-Instruct-v0.1
 - ## Introduction
   This model was created by applying [Quark](https://quark.docs.amd.com/latest/index.html) with calibration samples from Pile dataset.
 - ## Quantization Stragegy
-  - ***Quantized Layers***: All linear layers excluding "lm_head", "*gate"
+  - ***Quantized Layers***: All linear layers excluding "lm_head", "*.gate"
   - ***Weight***: FP8 symmetric per-tensor
   - ***Activation***: FP8 symmetric per-tensor
   - ***KV Cache***: FP8 symmetric  per-tensor
@@ -23,16 +23,18 @@ python3 quantize_quark.py \
         --output_dir Mixtral-8x7B-Instruct-v0.1-FP8-KV \                           
         --quant_scheme w_fp8_a_fp8 \
         --kv_cache_dtype fp8 \
-        --num_calib_data 128  \
-        --model_export quark_safetensors
+        --num_calib_data 128 \
+        --model_export quark_safetensors \
+        --no_weight_matrix_merge
 # If model size is too large for single GPU, please use multi GPU instead.
 python3 quantize_quark.py \ 
         --model_dir $MODEL_DIR \
         --output_dir Mixtral-8x7B-Instruct-v0.1-FP8-KV \                           
         --quant_scheme w_fp8_a_fp8 \
         --kv_cache_dtype fp8 \
-        --num_calib_data 128  \
+        --num_calib_data 128 \
         --model_export quark_safetensors \
+        --no_weight_matrix_merge \
         --multi_gpu
 ```
 ## Deployment