A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 12, 13:30 UTC
main
[responsesAPI][2] parse ResponseFunctionToolCallOutputItem (#29383)
Andrew Xia committed 12 months ago
Fix RoPE related failures in Transformers nightly tests (#29333)
Harry Mellor committed 12 months ago
Scheduled removal of `override_pooler_config` and `disable_log_requests` (#29402)
Harry Mellor committed 12 months ago
[Hybrid Allocator] Support KV cache groups with different block_size (#29143)
Yifan Qiao committed 12 months ago
[Kernel] Add NVFP4 MoE CUTLASS support for SM120 (#29242)
Michael Goin committed 12 months ago
[UX] Put CUDA attention backend selection log into one line (#29337)
Michael Goin committed 12 months ago
[Feature]: Improve GGUF loading from HuggingFace user experience like repo_id:quant_type (#29137)
Injae Ryou committed 12 months ago
Revert back to torch.equal over torch.allclose from #28819 (#29086)
Eldar Kurtić committed 12 months ago
[Bugfix] Fix logic for choosing default prefix caching setting (#29393)
Thomas Parnell committed 12 months ago
Scheduled removal of `CompilationConfig.use_inductor` (#29323)
Harry Mellor committed 12 months ago
Make Transformers Nightly tests soft-fail and enable all tests (#29401)
Harry Mellor committed 12 months ago
[Bugfix] Fix overallocation in MM profiling (#29386)
Roger Wang committed 12 months ago
[Core] Generalize Encoder-Decoder `seq_lens` computation to avoid Whisper hardcoded logic (#29268)
Nicolò Lucchesi committed 12 months ago
[Misc] Suppress log outputs when constructing the default vllm config. (#29291)
wang.yuqi committed 12 months ago
[CI] Resettle pooling entrypoints tests. (#29370)
wang.yuqi committed 12 months ago
[Frontend] Respect Chat Completion parallel_tool_calls param (#26233)
Ben Browning committed 12 months ago
[responsesAPI] refactor construct_input_messages (#29359)
Andrew Xia committed 12 months ago
[BugFix] bad_words filtering ineffective when n > 1 (#29313)
Avishek Goswami committed 12 months ago
[Misc] Streamline unique id generation (#29375)
Nick Hill committed 12 months ago
[Bugfix] If chunked_prefill is disabled, end the scheduling early. (#28911)
wang.yuqi committed 12 months ago