Author

Naji Alazhar

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.

#artificial intelligence Preprint Oct 2026

Adaptive DHiSparse: Speculative Decoding over a Load-Adaptive KV Hierarchy

Long-context serving runs out of KV memory before the hardware reaches its useful throughput. HiSparse stores history in host memory and fetches only the selected entries, but it uses one cache budget at every load and does not handle draft-model speculative decoding such as DSpark and DFlash2. We present Adaptive DHiS...

Naji Alazhar, Wassel Alazhar · 0 citations