update README

README.md

@@ -1,5 +1,7 @@
 ---
 license: apache-2.0
+base_model:
+- Qwen/Qwen2.5-7B-Instruct
 ---
 
 # AutoL2S-7B
@@ -8,16 +10,25 @@ This is the official model repository for **AutoL2S-7B**, a model fine-tuned for
 
 ## 💡 Overview
 
-AutoL2S enables automatically switching between short and long reasoning paths based on input complexity. This repository contains:
+AutoL2S enables automatically switching between short and long reasoning paths based on input complexity. 
+Auto Long-Short Reasoning (AutoL2S), a dynamic and model-agnostic framework that enables LLMs to dynamically compress their generated reasoning
+path based on the complexity of the reasoning question. AutoL2S enables a learned paradigm, in which LLMs themselves can decide when longer reasoning is necessary and when shorter reasoning
+suffices, by training on data annotated with our proposed method, which includes both long and short CoT paths and a special \<EASY\> token (\<specialLong\> in the implementation). We then use <EASY> token to indicate when the model can
+skip generating lengthy CoT reasoning. This proposed annotation strategy can enhance the LLMs’ ability to generate shorter CoT reasoning paths with improved quality after training. 
+
+This repository contains:
 
 - Model weights
 - Configuration files
 - necessary scripts in the `examples/` directory
 
----
+<p align="left">
+  <img src="https://cdn-uploads.huggingface.co/production/uploads/66f9bb2dd5575ad6914756ce/dVpIjeIaU8Hv1M5z5VWYS.png" width="35%" style="display:inline-block; margin-right: 10px;" />
+  <img src="https://cdn-uploads.huggingface.co/production/uploads/66f9bb2dd5575ad6914756ce/qxHTE-ZGTpxVjmkIX6Fk-.png" width="35%" style="display:inline-block;" />
+</p>
 
+---
 ## 🧩 Dependencies
-
 We recommend using the model with [vLLM](https://github.com/vllm-project/vllm).  
 The code has been tested with:
 
@@ -26,7 +37,6 @@ vLLM == 0.6.2
 ```
 
 ---
-
 ## 🚀 How to Use
 
 Run the inference example:
@@ -72,5 +82,4 @@ If you use this model in your work, please consider citing:
   journal={arXiv preprint},
   year={2025}
 }
-```
-
+```