Update README.md (#2)

- Update README.md (0d719138d3ff47608159348f19a66ce3fd5fa58c)


Co-authored-by: Avijit Ghosh <evijit@users.noreply.huggingface.co>

README.md

@@ -9,31 +9,50 @@ thumbnail: >-
   https://cdn-uploads.huggingface.co/production/uploads/680ff4388f704be391757780/RXyJ68fc3DZNShdz91nG1.png
 ---
 
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/680ff4388f704be391757780/XCrsh1YyHF1S-u9Q035zf.png)
+<div style="max-width: 100%; overflow: hidden;">
+  <img src="https://cdn-uploads.huggingface.co/production/uploads/680ff4388f704be391757780/XCrsh1YyHF1S-u9Q035zf.png" 
+       alt="Hugging Science" 
+       style="width: 100%; height: 700px;">
+</div>
 
 # Hugging Science  
+**Community-driven open science for the age of AI**
 
+## About
 
-**Community-driven open science for the age of AI**  
+Many of the deepest challenges in advancing AI for scientific discovery are not purely technical—they are social and organizational. Progress is often limited not by algorithms or computational power, but by how effectively we coordinate efforts, share resources, and collaborate across disciplinary boundaries. Hugging Science brings together a global community of researchers, developers, and practitioners committed to accelerating breakthroughs in physics, biology, chemistry, neuroscience, and beyond through open collaboration.
 
-Many of the deepest challenges in science are not purely technical, but social and organizational. Progress is often limited not by knowledge itself, but by how we coordinate, share, and collaborate. Hugging Science was created to bring together a global community of researchers, developers, and enthusiasts who believe that by working openly, we can accelerate breakthroughs in physics, drug development, neuroscience, biology, chemistry, and beyond.  
+Our vision is grounded in the argument presented in [our position paper](https://arxiv.org/abs/2509.06580): democratizing AI for science requires treating it as a collective social project where equitable participation and sustainable collaboration are prerequisites for technical progress.
 
-This vision is outlined in [our preprint](https://arxiv.org/html/2509.06580v1), which argues for democratizing scientific progress by empowering distributed contributions. Instead of concentrating resources in a few places, we can organize large-scale efforts around open challenges, shared benchmarks, and transparent workflows.  
+## What We Do  
 
-Hugging Science is both a community and an experiment: can we turn fragmented efforts into collective action, and in doing so, make real advances in scientific discovery? We invite anyone—whether from machine learning, the natural sciences, or just curiosity—to join us.  
+- **Launch collaborative challenges and open problem calls** to identify and mobilize collective effort around upstream computational bottlenecks with broad applicability across scientific domains—such as efficient PDE solvers, multi-scale coupling, and high-dimensional sampling—rather than fragmenting resources across narrow, domain-specific applications
 
-## What We Do  
-- Launch collaborative challenges & problem calls  
-- Build open toolkits, benchmarks, and workflows  
-- Support cross-disciplinary exchange & education  
-- Nurture a vibrant community at the interface of domain science and machine learning
-- Learn together 
+- **Build open toolkits, benchmarks, and workflows** that address data fragmentation through standardized formats and shared evaluation metrics, making it easier for researchers at institutions of all resource levels to collaborate and build on each other's work
+
+- **Support cross-disciplinary exchange and education** by creating resources that bridge the communication gap between domain scientists who prioritize mechanistic understanding and ML researchers who focus on predictive performance, enabling more effective collaboration
+
+- **Nurture a community** that values contributions to data curation, infrastructure development, and educational resources alongside algorithmic innovation—recognizing that datasets and infrastructure often have far greater long-term impact than individual models
+
+- **Learn together** through open discussion of both technical advances and the social and institutional barriers that constrain progress, working to align incentives and build sustainable practices for scientific AI
 
 ## Join Us  
-- [Discord](https://discord.com/invite/VYkdEVjJ5J) — real-time collaboration & discussion  
-- [GitHub](https://github.com/hugging-science) — open source projects & contributions
 
-Not sure where to start? That's okay! Join the Discord to get a lay of the land. 
+**Discord**: [Join our server](https://discord.com/invite/VYkdEVjJ5J) for real-time collaboration, discussion, and community coordination.
+
+**GitHub**: [Explore and contribute](https://github.com/hugging-science) to our open-source projects and initiatives.
+
+**New to the community?** Join the Discord to connect with other members and explore current initiatives. We'll help you find your place.
 
+**Have a project idea?** You don't need permission to start building. If you'd like community support, computational resources, or collaborative connections, reach out via Discord or GitHub.
+
+## Learn More
+
+Read our position paper: [AI for Scientific Discovery is a Social Problem](https://arxiv.org/abs/2509.06580)
+
+**Explore the [Science Release Heatmap](https://huggingface.co/spaces/hugging-science/science-release-map)** to visualize open models, datasets, and applications from organizations contributing to AI4Science, tracking release patterns across materials science, biology, physics, chemistry, and more!
+
+
+---
 
-Already have an idea? You don't need permission from us--you can just do things. If you want support from us on a cool project, you're already working on. Reach out!
+*Hugging Science is an open community effort. We believe that by working openly and collaboratively, we can make AI-enabled scientific discovery more accessible, equitable, and impactful.*