Tinker: Call for Community Projects

2025-11-07 · Source: Thinking Machines Lab - Connectionism · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Software Development & Engineering · Depth: Advanced, short

Summary

Thinking Machines has issued a call for community projects to be featured on its Tinker blog, inviting builders and researchers to submit their machine learning projects by November 7, 2025. Tinker is a platform designed for training and customizing models for studies and new applications. The initiative seeks submissions across several categories, including reimplementations of research, original ML research, AI-enabled research in other domains, product prototypes, novel datasets, high-level libraries built on Tinker, and infrastructure contributions. Submissions should include a rigorous write-up with clear evaluations, comparisons to baselines, and preferably an open-source code release. Thinking Machines also provided specific research directions, such as replicating Constitutional AI, exploring RLVR with Noisy Student, on-policy context distillation, RL memory tests, direct RL on pairwise judges, replicating Open Character Training, and developing GANs for joke generation.

Key takeaway

For AI Scientists and Research Scientists developing or customizing machine learning models, you should consider submitting your rigorous, well-documented projects to Tinker for community feature. This offers a valuable opportunity to gain visibility for your work, contribute to the broader ML community, and receive feedback on your experimental methodologies, particularly if your project aligns with the suggested research directions like Constitutional AI or RLVR with Noisy Student.

Key insights

Tinker invites ML projects for publication, emphasizing rigorous evaluation and open-source contributions.

Principles

• Prioritize rigor and clear evaluation over novelty.
• Compare novel methods to simpler baselines.
• Apply multiple analyses to examine each result.

Method

Submit ML research, custom models, or infrastructure contributions with a detailed write-up, clear evaluations, and open-source code to be featured on the Tinker blog.

In practice

• Replicate Constitutional AI from base models.
• Adapt Noisy Student self-distillation to LLMs.
• Compare on-policy vs. off-policy context distillation.

Topics

• Tinker Platform
• Machine Learning Research
• Reinforcement Learning
• Large Language Models
• Model Customization

Code references

• thinking-machines-lab/tinker-project-ideas

Best for: AI Scientist, Research Scientist, AI Researcher, Machine Learning Engineer, AI Student

Related on AIssential

• Announcing Tinker Research and Teaching Grants — Tinker offers grants to expand academic and non-profit access for training open-weight LLMs.
• Tinker: General Availability and Vision Input — Tinker's general availability, new reasoning model, OpenAI API compatibility, and vision input expand its utility for AI development.
• The Zig project's rationale for their firm anti-AI contribution policy — Zig's strict anti-LLM policy prioritizes human contributor development over immediate code contributions.
• Monthly "Is there a tool for..." Post — Centralized community forums effectively manage recurring tool recommendation queries in AI.
• A quick chat with Ines & Matt — Agentic systems can generate "software 2.0" by encoding behavior in data examples, not just explicit code.
• A Community Roadmap for Including Kapampángan in Modern AI Research — Community-driven data creation is essential for including underrepresented languages in AI research and applications.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Thinking Machines Lab - Connectionism.