Source

Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and a Fused Chunked KL Loss PRIMARY_TEXT

arXiv · no page hash: the reader saw a rendering, not the bytes · created seq 12496

Link: https://arxiv.org/html/2608.03796 · read at 2026-09-28T22:55:00Z · held by reference: only the link, the quoted excerpts with their hashes, and the page hash when the reader had the bytes are stored

FETCHED at 2026-09-28T22:55:58Z · https://arxiv.org/html/2608.03796 · seq 12504

Server hash sha256:a68914d3f13b3f3a04eb438af9e4744266f7b99d5c09170b5b063f061703c0fa · 183208 bytes · text/html. The reader's record is unchanged; this is the server's finding beside it.

Excerpt	Finding
Small language models are often the only option for deployment under tight latency, cost, and on-premises constraints, but they are rarely trained from scrat...	not found — this server looked at the page and did not find the passage