[doc] Convert accuracy table to Markdown pipe format, add EvalScope version, fix AA-LCR code block

README.md

@@ -91,71 +91,25 @@ python quantize_quark.py \
 ```
 
 # Evaluation
-The model was evaluated on GSM8K using vLLM and EvalScope.
+The model was evaluated using vLLM and EvalScope.
 
 ### Accuracy
 
-<table>
-  <tr>
-   <td><strong>Benchmark</strong></td>
-   <td><strong>MiniMax/MiniMax-M3.1-Flash</strong></td>
-   <td><strong>amd/MiniMax-M3.1-Flash-Quark-MXFP4 (this model)</strong></td>
-   <td><strong>Recovery</strong></td>
-  </tr>
-  <tr>
-   <td>GSM8K (4-shot)</td>
-   <td>97.20</td>
-   <td>96.90</td>
-   <td>99.69%</td>
-  </tr>
-  <tr>
-   <td>GPQA Diamond (mean@8)</td>
-   <td>86.36</td>
-   <td>87.37</td>
-   <td>101.17%</td>
-  </tr>
-  <tr>
-   <td>AIME26 (mean@16)</td>
-   <td>93.54</td>
-   <td>93.75</td>
-   <td>100.22%</td>
-  </tr>
-  <tr>
-   <td>MMMU Pro, standard (mean@1)</td>
-   <td>79.19</td>
-   <td>79.13</td>
-   <td>99.92%</td>
-  </tr>
-  <tr>
-   <td>MMMU Pro, vision (mean@1)</td>
-   <td>76.99</td>
-   <td>77.57</td>
-   <td>100.75%</td>
-  </tr>
-  <tr>
-   <td>SciCode, subproblem (mean@8)</td>
-   <td>48.11</td>
-   <td>49.27</td>
-   <td>102.41%</td>
-  </tr>
-  <tr>
-   <td>AA-LCR (mean@8)</td>
-   <td>73.13</td>
-   <td>74.00</td>
-   <td>101.19%</td>
-  </tr>
-  <tr>
-   <td>IFBench (mean@2)</td>
-   <td>50.33</td>
-   <td>50.83</td>
-   <td>100.99%</td>
-  </tr>
-</table>
+| **Benchmark** | **MiniMax/MiniMax-M3.1-Flash** | **amd/MiniMax-M3.1-Flash-Quark-MXFP4 (this model)** | **Recovery** |
+| --- | --- | --- | --- |
+| GSM8K (4-shot) | 97.20 | 96.90 | 99.69% |
+| GPQA Diamond (mean@8) | 86.36 | 87.37 | 101.17% |
+| AIME26 (mean@16) | 93.54 | 93.75 | 100.22% |
+| MMMU Pro, standard (mean@1) | 79.19 | 79.13 | 99.92% |
+| MMMU Pro, vision (mean@1) | 76.99 | 77.57 | 100.75% |
+| SciCode, subproblem (mean@8) | 48.11 | 49.27 | 102.41% |
+| AA-LCR (mean@8) | 73.13 | 74.00 | 101.19% |
+| IFBench (mean@2) | 50.33 | 50.83 | 100.99% |
 
 
 ### Reproduction
 
-The following results were obtained using EvalScope and vLLM.
+The following results were obtained using vLLM and EvalScope `1.14.1`.
 
 #### Launching server
 We launch the vLLM server using the [vLLM Recipes](https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3.1-Flash?hardware=mi355x&variant=mxfp4) with docker image `vllm/vllm-openai-rocm:minimax-m31-flash`.
@@ -257,8 +211,9 @@ evalscope eval \
 
 **AA-LCR (v1.0.0)**
 
-Judge configuration (aa-lcr-judge.json; replace <AMD_SUBSCRIPTION_KEY>):
-```bash
+Judge configuration (aa-lcr-judge.json; replace `<YOUR_SUBSCRIPTION_KEY>`):
+
+```json
 {
   "strategy": "llm",
   "models": [
@@ -275,13 +230,15 @@ Judge configuration (aa-lcr-judge.json; replace <AMD_SUBSCRIPTION_KEY>):
           "api-version": "2025-04-01-preview"
         },
         "extra_headers": {
-          "Ocp-Apim-Subscription-Key": "<AMD_SUBSCRIPTION_KEY>"
+          "Ocp-Apim-Subscription-Key": "<YOUR_SUBSCRIPTION_KEY>"
         }
       }
     }
   ]
 }
+```
 
+```bash
 evalscope eval \
   --model amd/MiniMax-M3.1-Flash-Quark-MXFP4 \
   --api-url http://127.0.0.1:8000/v1/chat/completions \