Improve dataset card: Add paper/project/code links, task categories, sample usage, and correct license

README.md

@@ -1,3 +1,155 @@
----
-license: apache-2.0
----
+---
+license: mit
+task_categories:
+- robotics
+---
+
+# From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction
+
+This repository contains the dataset and resources associated with the paper "[From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction](https://huggingface.co/papers/2510.19654)".
+
+**Project Page:** [https://6550zhao.github.io/Policy-World-Model/](https://6550zhao.github.io/Policy-World-Model/)
+**Code Repository:** [https://github.com/6550Zhao/Policy-World-Model](https://github.com/6550Zhao/Policy-World-Model)
+
+## 📰 News
+
+-   **[2025-09-18] 🎉 Our paper has been accepted to NeurIPS 2025 as a poster! 🎉**
+
+---
+
+## 🖼️ Project Overview
+
+<!-- Project Main Figure Placeholder -->
+<div align="center">
+<img src="https://github.com/6550Zhao/Policy-World-Model/raw/main/assets/paper_figure.png" alt="Policy World Model Overview" width="800">
+</div>
+
+The Policy World Model (PWM) introduces a novel driving paradigm that integrates world modeling and trajectory planning within a unified architecture. It leverages learned world knowledge through an action-free future state forecasting scheme to benefit planning, mimicking human-like anticipatory perception for more reliable performance. The method also features a dynamically enhanced parallel token generation mechanism for efficient video forecasting.
+
+---
+
+## 🚀 Key Features
+
+-   🔗 **Unified Framework**: Integrates world modeling and trajectory planning in a single architecture
+-   🧠 **Human-like Anticipation**: Mimics anticipatory perception through collaborative state-action prediction
+-   ⚡ **Efficient Video Forecasting**: Dynamic parallel token generation with context-guided tokenizer
+-   📊 **State-of-the-Art Performance**: Exceeds existing methods on benchmark datasets
+
+---
+
+## 📊 Results
+
+### Performance Comparison
+
+<!-- Replace with your results table image -->
+<div align="center">
+<img src="https://github.com/6550Zhao/Policy-World-Model/raw/main/assets/results_table.png" alt="Performance Comparison Results" width="800">
+</div>
+
+---
+
+## Sample Usage (Guidlines)
+This guide provides step-by-step instructions for fine-tuning and evaluating the Policy-World-Model (PWM) on NuScenes and NavSim datasets. Ensure you have completed the environment setup and data preparation before proceeding.
+
+### 1. Clone the Repository
+First, clone the repository and navigate to the project root directory:
+```bash
+git clone https://github.com/6550Zhao/Policy-World-Model.git
+cd Policy-World-Model  # Enter the project folder (replace with your actual path if needed)
+```
+### 2. Create & Activate Conda Environment
+Create and activate the environment with the provided environment.yaml:
+```bash
+# Create environment from the configuration file
+conda env create -f environment.yaml
+
+# Activate the PWM environment (name: pwm)
+conda activate pwm
+```
+### 3. Data Preparation
+#### 3.1 Required Data Files
+Download the following resources to run fine-tuning/evaluation:
+
+Pre-trained weights: Download from the official website ([Download](https://huggingface.co/zzzz12334/Policy_World_Model/tree/main)).
+
+Dataset files: Download dataset files from the specified source (for convenience, some required additional dataset files are available for download here: [Download](https://huggingface.co/datasets/talas/pwm_datasets/tree/main)).
+
+#### 3.2 Dataset Directory Structure
+
+Organize your dataset folder as follows (ensure the path matches the configuration in yaml files):
+```
+dataset/
+├── nuscenes/
+│   ├── maps/
+│   ├── samples/
+│   ├── sweeps/
+│   └── ominidrive/ # Download from the provided link
+└── navsim/
+    ├── maps/
+    ├── cache/
+    ├── navsim_logs/
+    ├── sensor_blobs/
+    ├── nuplan_img_logs/ # Download from the provided link
+    ├── nuplan_scene_blobs/
+    │   ├── 10hz_train/
+    │   │   ├── 2021.05.12.19.36.12_veh-35_00005_00204/
+    │   │   ├── 2021.05.12.19.36.12_veh-35_00215_00405/
+    │   │   └── ... (other training scenes)
+    │   ├── 10hz_test/
+    │   └── 10hz_val/
+```
+### 4. Evaluation
+4.1 Evaluate on NuScenes
+Modify the configuration file to enable evaluation mode:
+Open configs/sft_nuscenes/nuscenes.yaml
+Set experiment.eval_only = True (ensure no extra spaces or syntax errors)
+Run the evaluation script:
+```bash
+bash scripts/finetune/nuscenes/run_sft_nusc_no_ego_baseline.sh
+```
+4.2 Evaluate on NavSim
+Modify the configuration file to enable evaluation mode:
+Open configs/sft_nuscenes/nuscenes.yaml
+Set experiment.eval_only = True
+Run the evaluation script:
+```bash
+bash scripts/finetune/navsim/run_sft_navsim_baseline.sh
+```
+### 5. Fine-tuning
+5.1 Fine-tune on NuScenes
+Modify the configuration file to enable training mode:
+Open configs/sft_nuscenes/nuscenes.yaml
+Set experiment.eval_only = False
+Start fine-tuning:
+```bash
+bash scripts/finetune/nuscenes/run_sft_nusc_no_ego_baseline.sh
+```
+5.2 Fine-tune on NavSim
+Modify the configuration file to enable training mode:
+Open configs/sft_nuscenes/nuscenes.yaml
+Set experiment.eval_only = False
+Start fine-tuning:
+```bash
+bash scripts/finetune/navsim/run_sft_navsim_baseline.sh
+```
+
+---
+
+## 📖 Citation
+
+If you find our work useful, please cite:
+
+```bibtex
+@inproceedings{zhao2025pwm,
+  title={From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction},
+  author={Zhao, Zhida and Fu, Talas and Wang, Yifan and Wang, Lijun and Lu, Huchuan},
+  booktitle={Advances in Neural Information Processing Systems},
+  year={2025}
+}
+```
+
+---
+
+## 📄 License
+
+This project is licensed under the MIT License.