Author

Dhawal Shah

View on Semantic Scholar

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Jul 2026

VarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs

VarRate is presented, a training-free KV codec that assigns each token a variable low-rank budget by its query salience, keeping every token at a nonzero rank, and is the strongest matched-memory compressor.

Shahrzad Esmat, Dhawal Shah, Ali Jannesari · arXiv.org · 0 citations