A high-throughput and memory-efficient inference and serving engine for LLMs
main
[KV Offloading] Add tiering metric plumbing (#45959)
Srinivas Krovvidi committed 4 months ago
[CPU][Bugfix][Speculative Decoding] Accept USE_FP64_GUMBEL in CPU recovered-tokens sampler (#46069)
hillel.darshan committed 4 months ago
[Frontend] Split ServingRender into renderer and entrypoint. (#44285)
wang.yuqi committed 4 months ago
[ROCm][P/D] Support MoRIIO heterogeneous TP fan-in (#46332)
Tan Pin Siang committed 4 months ago
fix(security): prevent infinite loop in split_audio with NaN audio sa… (#46463)
Juan Pérez de Algaba committed 4 months ago
[ROCm][Perf] DSv3.2: fuse MLA Q concat+fp8-quant in forward_mqa (#43673)
frida-andersson committed 4 months ago
fix gpt_oss pp>1 with ep (#46441)
Qiming Zhang committed 4 months ago
[Rust Frontend] Align Rust allowed_token_ids validation with Python (#46348)
Reid committed 4 months ago
[Rust Frontend] Support echo for token-ID completion prompts (#46219)
Reid committed 4 months ago
Doc: fix missing GLM-5.x in supported models (#46452)
Sunny Yuan committed 4 months ago
[Rust Frontend] Correct `--reasoning-parser` semantics (#46359)
Bugen Zhao committed 4 months ago
[Bugfix][Model Runner V2] Preserve all allowed_token_ids in the logit bias kernel (#46245)
Ting SUN committed 4 months ago
[BugFix] Omit empty tool_calls from OpenAI chat responses (#44105)
jack committed 4 months ago
[Refactor] Responses API parser state into conversation context (#46030)
Chauncey committed 4 months ago
[DeepEP V2] Fill invalid recv_topk_idx with -1 (#46432)
Woosuk Kwon committed 4 months ago
[ROCm][P/D] Fix MoRIIO WRITE mode for mixed KV layouts (#46290)
Tan Pin Siang committed 4 months ago
[Rust Frontend][Perf] Use dedicated runtime for HTTP/request-processing/ZMQ (#46051)
Bugen Zhao committed 4 months ago
[XPU][CI]fix xpu kv cache layout test (#46429)
Kunshang Ji committed 4 months ago
[v1][kvcache] Honor prefix-cache retention interval for Mamba/linear attention (#45845)
Dao007forever committed 4 months ago
[Quant] Enable modelopt_mixed on Turing (SM75) (#45375)
Mike G committed 4 months ago