Spaces:
Runtime error
Apply for community grant: Academic project (gpu)
Hi, I am a second-year undergraduate student at the University of Washington. I recently helped lead a project to develop Superposed Decoding, a new LLM decoding algorithm that interpolates token embeddings to generate $k$ drafts using only one inference pass (under submission to NeurIPS 2024). We would love to demo our technique on HuggingFace spaces, and we believe that our work is important for many draft-based applications, such as GitHub Copilot and VsCode Tabnine. It would be amazing if we could receive 1x A100 or 2x A10 free of charge to use for our demo. Thanks!
Paper: https://arxiv.org/abs/2405.18400
Code: https://github.com/RAIVNLab/SuperposedDecoding
Hi @ethanlshen , we've assigned ZeroGPU to this Space. Please check the compatibility and usage sections of this page so your Space can run on ZeroGPU.