COMMITS
June 17, 2026
H
[ROCm] GLM-5.2 recipe for MI300 and MI355 (#561)
Hongxia Yang committed
J
update for glm52 (#559)
Jared Wen committed
June 16, 2026
J
Remove DG note (#557)
Jee Jee Li committed
J
Index share (#555)
Jared Wen committed
H
Update Kimi-K2 AMD recipe YAML format (#548)
haic0 committed
June 15, 2026
June 14, 2026
F
MiniMax-M3: set VLLM_USE_BREAKABLE_CUDAGRAPH=0 on AMD MI300X/MI325X/MI355X (#539)
functionstackx committed
R
Add NVFP4 variant for DeepSeek-V4-Pro and DeepSeek-V4-Flash (#541)
Roy Wang committed
R
refactor: enhance logo fetching logic to include hardware vendors (#540)
Roy Wang committed
R
clean up versions (#538)
Roy Wang committed
R
Update MiniMax-M3 YAML (#537)
Roy Wang committed
June 12, 2026
H
R
cleanup: remove outdated dependencies for Kimi-K2.7-Code (#534)
Roy Wang committed
T
Add minimax m3 recipe for NV + AMD (#526)
Tiezhen WANG committed
R
Add moonshotai/Kimi-K2.7-Code recipe (#533)
Roy Wang committed
H
Update Ministral 3 Instruct AMD recipe YAML format (#531)
haic0 committed
R
R
update gpt oss recipe (#524)
Roy Wang committed
June 11, 2026
L
move to docker
Lucas Wilkinson committed
L
Add DiffusionGemma 26B-A4B-IT recipe (#520)
Luciano Martins committed
June 10, 2026
F
cleanup: remove MiniMax fusion config (#521)
Faradawn Yang committed
W
[ROCm] Add the Deepseek-V4-Pro supported on MI355x (#433)
wuhuikx committed
June 7, 2026
F
chore: remove SAFETENSORS_FAST_GPU=1 env var (#513)
Faradawn Yang committed
June 6, 2026
M
Use match-sorter for search: multi-word, separator-insensitive, ranked (#516)
Michael Goin committed
June 5, 2026
L
[Gemma4] Add official QAT W4A16 and mobile quantized variants (#514)
Luciano Martins committed
F
add attention backend recommendation for Minimax 2.5 (#512)
Faradawn Yang committed
M
Add Trillium and Ironwood configurations for Qwens, Llama, and GPT-OSS (#510)
Mateusz Sokół committed
June 4, 2026
F
Nemotron 3 Ultra Recipe (#509)
Faradawn Yang committed
S
Add DGX Spark Qwen3.6 recipe (#480)
Serge Panev committed
June 3, 2026
R
Update google/gemma-4-12B-it: nightly pip support + vLLM PR link (#508)
Roy Wang committed