Update README.md

README.md

@@ -13,14 +13,18 @@ pipeline_tag: text-generation
 
 English | [中文](README_ZH.md)
 
-This model is a fine-tuned version of **Qwen3-4B-Instruct-2507**, specialized for **emotion recognition and emotionally-aware text generation**.  
+This model is a fine-tuned version of **Qwen3-4B-Instruct-2507**, specialized for **emotion recognition and emotionally-aware text generation**.
 
-We provide two weight formats:  
+We provide this model in this format:
+- [safetensor(fp16)](https://huggingface.co/OmniDimen/OmniDimen-v1.1-4B-Emotion) (~13GB on disk)
+
+The **safetensors weights** can be used directly with PyTorch-based setups and other frameworks that accept safetensors.
+
+We also provide two easy-to-use weight formats:
 - **fp16** (~8GB on disk)  
 - [q4_K_M quantized](https://huggingface.co/OmniDimen/OmniDimen-v1.1-4B-Emotion-GGUF-q4_K_M) (~2.5GB on disk)  
 
 Both formats are released in **GGUF**, easily usable with [LM Studio](https://lmstudio.ai/), [Ollama](https://ollama.com/), and [PocketPal](https://github.com/a-ghorbani/pocketpal-ai).  
-Later, we will also upload [safetensors weights](https://huggingface.co/OmniDimen/OmniDimen-4B-Emotion) for research and framework integration.  
 
 ---
 
@@ -91,8 +95,8 @@ Take all the time you need, sweetheart. Just remember that no matter what happen
 ## 📝 Changelog  
 
 - **v1.1 (2025-09-29)**
-  - First upload of **Safetensors** .
   - Fix some bugs that output abnormal characters.
+  - First upload of **safetensor weights**.
 
 - **v1.0 (2025-09-19)**  
   - First upload of **GGUF weights** (fp16 and q4_K_M).