Upload README.md with huggingface_hub

README.md

@@ -28,13 +28,13 @@ datasets:
 
 ## BioReason-Pro SFT
 
-Supervised fine-tuned (SFT) checkpoint of BioReason-Pro — a multimodal reasoning LLM for protein function prediction. This model integrates ESM3 protein embeddings, a GO graph encoder, and biological context (InterPro domains, STRING interactions) within a Qwen3-4B backbone to generate structured reasoning traces and functional annotations.
+Supervised fine-tuned (SFT) checkpoint of BioReason-Pro, a multimodal reasoning LLM for protein function prediction. This model integrates ESM3 protein embeddings, a GO graph encoder, and biological context (InterPro domains, STRING interactions) within a Qwen3-4B backbone to generate structured reasoning traces and functional annotations.
 
-**Training data:** 117,002 proteins with synthetic reasoning traces generated by GPT-5, available at [wanglab/bioreason-pro-sft-reasoning-data](https://huggingface.co/datasets/wanglab/bioreason-pro-sft-reasoning-data).
+**Training data:** [wanglab/bioreason-pro-sft-reasoning-data](https://huggingface.co/datasets/wanglab/bioreason-pro-sft-reasoning-data)
 
 See also:
-- [BioReason-Pro RL](https://huggingface.co/wanglab/bioreason-pro-rl) — reinforcement learning optimized checkpoint
-- [GO-GPT](https://huggingface.co/wanglab/gogpt) — autoregressive GO term predictor
+- [BioReason-Pro RL](https://huggingface.co/wanglab/bioreason-pro-rl) - reinforcement learning optimized checkpoint
+- [GO-GPT](https://huggingface.co/wanglab/gogpt) - autoregressive GO term predictor
 
 
 ## Citation