Open source

ggml-org/llama.cpp b10731

GitHub Releases · github-actions[bot] · 17 days ago

<details open> qwen4exp: support recurrent state rollback (#28123) MTP speculative decoding needs the target state to move back by the number of rejected draft tokens. Without rollback support the context is classified as SEQ_RM_TYPE_FULL and the server serializes the whole recu…

Open on github.com ↗

Source and ranking details

Source adapter: github-releases
Source weight: 3
Points: 0
Age: 412.3 h
Stored score: 0.00004
Current score: 0.00004
First seen: 2026-09-01T08:37:06.014Z

Newsletter

Get practical AI engineering notes

Receive source-checked analysis of models, agents, evaluation, retrieval, and production reliability. Sent only when there is useful work to share.