Lossfunk: your friendly neighborhood AI lab

Research areas:

We're interested in frontier areas of AI research, including:

• Sample efficient AI
• Reasoning
• Efficiency in AI (training, inference, etc)
• Reinforcement learning
• Non-standard models (e.g. GFlowNets, Diffusion models, etc)
• Non-gradient based methods like program synthesis, evolutionary methods, etc

Here's an open list of research directions we are looking to collaborate on.

If you want to collaborate, please reach out with your proposal. We also provide compute (GPUs/LLM credits) for our collaborations on a case-to-case basis.

Programs:

• A six-week residency for programmers to explore AI deeply

Projects

• IndiaML tracker: papers by Indian teams accepted in top AI/ML conferences

Publications / preprints

• IPO: Your Language Model is Secretly a Preference Classifier [2025]
• Do GFlowNets Transfer? Case Study on the Game of 24/42 [2025]

Contact / follow

• Contact: [email protected]
• X: @lossfunk

-[--->+<]>---.+[----->+++<]>.----.+++.+[--->+<]>++.-[---->+<]>++.---[->++++<]>.-----.[--->+<]>-----.++++++[->++<]>.+[--->+<]>.++++..-------------.-[--->+<]>--.-------.---.[->+++<]>--.

Stay curious :)