Update README.md

README.md

@@ -20,7 +20,8 @@ tags:
 ## Model Overview
 Huihui-MoE-23B-A4B-abliterated is a **Mixture of Experts (MoE)** language model developed by **huihui.ai**, built upon the **[huihui-ai/Huihui-Qwen3-4B-abliterated-v2](https://huggingface.co/huihui-ai/Huihui-Qwen3-4B-abliterated-v2)** base model. It enhances the standard Transformer architecture by replacing MLP layers with MoE layers, each containing 8 experts, to achieve high performance with efficient inference. The model is designed for natural language processing tasks, including text generation, question answering, and conversational applications.
 
-The corresponding version is [huihui-ai/Huihui-MoE-23B-A4B](https://huggingface.co/huihui-ai/Huihui-MoE-23B-A4B)
+The corresponding version is [huihui-ai/Huihui-MoE-23B-A4B](https://huggingface.co/huihui-ai/Huihui-MoE-23B-A4B). 
+Here, simply by changing the instruction version to the ablation version, we can achieve the ablation effect. This is just a try for testing, and we can compare the differences between these two versions.
 
 **Note**:  
 The activated expert can handle numbers from 1 to 8, and can complete normal conversations as well.