Commit History

Автор SHA1 Съобщение Дата
  AlpinDale a5bfc2bc3d VLM: add support for LLaVA-Onevision model (#1100) преди 1 месец
  AlpinDale d44da0332c misc: rename `CudaMemoryProfiler` to `DeviceMemoryProfiler` (#1099) преди 1 месец
  AlpinDale 7ce3174039 VLM: refactor blip models to support composite weight loading (#1098) преди 1 месец
  AlpinDale 91d03c04d2 VLM: refactor composite weight loading logic (#1097) преди 1 месец
  AlpinDale b65449b5ad moe: refactor DBRX experts to support FusedMoE (#1095) преди 1 месец
  AlpinDale ed63c079f7 Triton: remove atomic add op from awq triton (#1094) преди 1 месец
  AlpinDale 651678d2df VLM: use `SequenceData.from_token_counts` to create dummy data (#1093) преди 1 месец
  AlpinDale 7fffa507ff build: build flash attention kernels inside aphrodite (#1085) преди 1 месец
  AlpinDale 3d5b97837f ci: fix the tag for :latest docker преди 1 месец
  AlpinDale d96c363301 api: fix admin key being required for authentication (#1091) преди 1 месец
  AlpinDale 1fac86c325 core: factor out common code in SequenceData (#1083) преди 2 месеца
  AlpinDale ad1205b277 readme: update attributions (#1082) преди 2 месеца
  AlpinDale 193fcee016 chore: check for torch 2.4.0 when registering custom op (#1081) преди 2 месеца
  AlpinDale 86bf2cc4f3 core: rename `PromptInputs,inputs` -> `PromptType,prompt` (#1080) преди 2 месеца
  AlpinDale 766ea79b89 vlm: fix feature size calculation for llava-next models (#1079) преди 2 месеца
  AlpinDale 7b6501bd05 tests: refactor model tests (#1078) преди 2 месеца
  AlpinDale f6df92bde0 fix: unexpected kwarg for the legacy API server (#1076) преди 2 месеца
  AlpinDale 814c850d89 fix: validate `n` in the sampling params (#1075) преди 2 месеца
  AlpinDale 6212072245 api: support LoRA lineage and base model metadata management (#1072) преди 2 месеца
  AlpinDale d9d287a288 rocm: enable multi-step scheduling for rocm (#1071) преди 2 месеца
  AlpinDale ec17b6c4d0 fix: Phi3.5 Mini and MoE LoRA inference (#1070) преди 2 месеца
  AlpinDale acc0c727c8 vlm: add support for molmo vision model (#1069) преди 2 месеца
  AlpinDale 7dd001ec2d build: guard against changes in cuda library name (#1068) преди 2 месеца
  AlpinDale ca7028d5ca sampler: simplify logits resort in _apply_top_k_top_p (#1067) преди 2 месеца
  AlpinDale 61aed092a5 rocm: add support for FP8 KV cache in the custom paged attention kkernels (#1066) преди 2 месеца
  AlpinDale 12b0059b47 api: enable MQAphroditeEngine for embedding models (#1065) преди 2 месеца
  AlpinDale 314fa7f7d9 fix: encoder-decoder models for beam search (#1064) преди 2 месеца
  AlpinDale 34cf9b74f0 api: non-zero exit code if MQ engine startup fails (#1063) преди 2 месеца
  AlpinDale 92cee435e2 rocm: add more quants, fix _scaled_mm call (#1062) преди 2 месеца
  AlpinDale b12d5c0507 readme: add attribution to Ruliad преди 2 месеца