Fix doc bug (#78)

* Update index.rst * Update index.rst * Fix doc bug * Update awq.md * Update awq.md --------- Co-authored-by: gushiqiao <[email protected]>
ModelTC · Sep 9, 2024 · ef100d5 · ef100d5
1 parent 10e3120
commit ef100d5
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 3 deletions.
diff --git a/docs/en/source/practice/awq.md b/docs/en/source/practice/awq.md
@@ -41,7 +41,7 @@ task_name=awq_w4a16
 config=${llmc}/configs/quantization/methods/Awq/awq_w_only.yml
 ```
 
-With this improvement, AWQ-LLMC can achieve better accuracy compared to the [original method](), especially showing significant improvement in 2-bit quantization.
+With this improvement, AWQ-LLMC can achieve better accuracy compared to the [original method](https://github.com/mit-han-lab/llm-awq), especially showing significant improvement in 2-bit quantization.
 
 If `clip_sym` is not specified in `config.quant.special`, its value will default to the same as `config.quant.weight.symmetric`. If you want to reproduce academic-level accuracy, you can add `clip_sym` to the config and set it to `True`:
 
@@ -94,4 +94,4 @@ task_name=awq_w8a8
 config=${llmc}/configs/quantization/methods/Awq/awq_w_a.yml
 ```
 
-In weight-activation quantization, AWQ-LLMC can achieve better results than algorithms like SmoothQuant.
+In weight-activation quantization, AWQ-LLMC can achieve better results than algorithms like SmoothQuant.
diff --git a/docs/zh_cn/source/practice/awq.md b/docs/zh_cn/source/practice/awq.md
@@ -42,7 +42,7 @@ task_name=awq_w4a16
 config=${llmc}/configs/quantization/methods/Awq/awq_w_only.yml
 ```
 
-通过这一改进，AWQ-LLMC 可以取得比于 [原始方法]() 更好的精度表现，尤其在2-bit量化，表现出显著的改善。
+通过这一改进，AWQ-LLMC 可以取得比于 [原始方法](https://github.com/mit-han-lab/llm-awq) 更好的精度表现，尤其在2-bit量化，表现出显著的改善。
 
 
 如果在 `config.quant.special` 中未指定 `clip_sym`，那么它的取值将与 `config.quant.weight.symmetric` 保持一致。如果想复现学术精度，可以将 `clip_sym` 写到config里并设置为 `True`：