llama.cpp=2.1.1
description=lanyue2.1.0
tcim_version=1.3.0
[recent_commits]
999f6d2fd supprot qwen3 rerank & embedding
f521145a5 add qwen3.5 9b
e8b65d011 sync 23fbfcb1ad6c6f76b230e8895254de785000be46:20260309
b2e2b82ac support --list-devices
beb3652f6 support qwen3.5 prefix cache
72f879e08 updage dadao to v1.2.0. print more info of  houmoNPU
9e73dafb2 sync e43431b3811543efdad896e93a992551cc72ab5a:20260508
update dadao to v1.3.0.
e6811e531 APPSOFT-584 opt mtp performance
ebc9470db Reduce memory footprint. Fix potential out-of-bounds access behaviors.
a44a1b347 support qwen3-next-80b
56199e8cd support new gemma4(support llm kvcache reuse) sliding windows
179b38a34 Optimize the TTFT of ASR & fix the bug of mutibatch decode.
c7d76f9ac support  gemma4-e2b
[last_update]
20260704
