Ravi Ghadia — accepted AI-conference papers

2026

Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking

Ravi Ghadia, Maksim Abraham, Sergei Vorobyov, Max Ryabinin

ICML 2026poster

Efficiently processing long sequences with Transformer models usually requires splitting the computations across accelerators via context parallelism. The dominant approaches in this family of methods, such as Ring Attention or DeepSpeed Ulysses, enable scaling over the context dimension but do not …

Cited by 0 Source ↗Scholar ↗

2025

Dialogue Without Limits: Constant-Sized KV Caches for Extended Response in LLMs

Ravi Ghadia, Avinash Kumar, Gaurav Jain, Prashant J. Nair, Poulami Das

ICML 2025poster

Autoregressive Transformers rely on Key-Value (KV) caching to accelerate inference. However, the linear growth of the KV cache with context length leads to excessive memory consumption and bandwidth constraints. Existing methods drop distant tokens or compress states in a lossy manner, sacrificing a…

Cited by 1 Source ↗PDF ↗Scholar ↗