Add model card for NCoTS

README.md

@@ -1,3 +1,37 @@
----
-license: mit
----
+---
+license: mit
+pipeline_tag: text-generation
+---
+
+# Neural Chain-of-Thought Search (NCoTS)
+
+This repository contains artifacts for **Neural Chain-of-Thought Search (NCoTS)**, a framework that reformulates reasoning as a dynamic search for the optimal thinking strategy to enhance Large Language Models.
+
+- **Paper:** [Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Models](https://huggingface.co/papers/2601.11340)
+- **Repository:** [MilkThink-Lab/Neural-CoT-Search](https://github.com/MilkThink-Lab/Neural-CoT-Search)
+
+## Introduction
+NCoTS addresses the limitations of sequential Chain-of-Thought (CoT) generation by treating reasoning as a search process. By characterizing the solution space, the framework identifies "sparse superior reasoning paths" that are more accurate and concise than standard outputs. NCoTS navigates these paths using a dual-factor heuristic that balances correctness and computational cost, achieving a Pareto improvement across diverse reasoning benchmarks.
+
+## 🚀 Quick Start
+
+### ⚙️ Environment Setup
+
+```bash
+pip install -r requirements.txt
+```
+
+### 📊 Evaluation
+
+To evaluate the model using the provided codebase:
+
+```bash
+python evaluation.py \
+    --checkpoint <path_to_head_checkpoint> \
+    --dataset <dataset_name> \
+    --max_new_tokens 4096 \
+    --output_dir ./evaluation_results \
+    --verbose
+```
+
+For more details on training and project structure, please refer to the [GitHub repository](https://github.com/MilkThink-Lab/Neural-CoT-Search).