Update README.md

README.md

@@ -16,8 +16,9 @@ base_model:
 - **Operating System(s):** Linux
 - **Inference Engine:** [vLLM](https://docs.vllm.ai/en/latest/)
 - **Model Optimizer:** [AMD-Quark](https://quark.docs.amd.com/latest/index.html) (V0.11.1)
-  - **Weight quantization:** MOE-only, OCP MXFP4, Static 
-  - **Activation quantization:** MOE-only, OCP MXFP4, Dynamic
+  - **Quantized layers:** Experts, Shared_experts
+  - **Weight quantization:** OCP MXFP4, Static 
+  - **Activation quantization:** OCP MXFP4, Dynamic
 - **Calibration Dataset:** [Pile](https://huggingface.co/datasets/mit-han-lab/pile-val-backup)
 
 This model was built with Kimi-K2-Instruct model by applying [AMD-Quark](https://quark.docs.amd.com/latest/index.html) for MXFP4 quantization.
@@ -30,7 +31,7 @@ The model was quantized from [unsloth/Kimi-K2-Instruct-0905-BF16](https://huggin
 **Quantization scripts:**
 ```
 cd Quark/examples/torch/language_modeling/llm_ptq/
-exclude_layers="*self_attn* *mlp.gate *lm_head *mlp.gate_proj *mlp.up_proj *mlp.down_proj *shared_experts*"
+exclude_layers="*self_attn* *mlp.gate *lm_head *mlp.gate_proj *mlp.up_proj *mlp.down_proj"
 
 python quantize_quark.py \
     --model_dir unsloth/Kimi-K2-Instruct-0905-BF16 \
@@ -62,13 +63,13 @@ The model was evaluated on GSM8K benchmarks.
    </td>
   </tr>
   <tr>
-   <td>GSM8K (strict-match)
+   <td>GSM8K (flexible-extract)
    </td>
-   <td>95.53
+   <td>95.45
    </td>
-   <td>94.47
+   <td>93.78
    </td>
-   <td>98.89%
+   <td>98.25%
    </td>
   </tr>
 </table>