Commit History

Автор SHA1 Съобщение Дата
  AlpinDale 172dee2573 (2/N) Triton Backend: integrate Triton activation kernels (#1126) преди 3 дни
  AlpinDale 0c17153073 (1/N) Triton Backend: integrate Triton layernorm kernels (#1125) преди 5 дни
  AlpinDale 349a612338 chore: bump bitsandbytes version to latest; enable cuda graphs for 4bit bnb (#1123) преди 5 дни
  AlpinDale ede17d5039 fix: torch.compile dynamo fix (#1122) преди 5 дни
  AlpinDale e294eede32 Quantization: re-enable awq_marlin serialization (#1121) преди 5 дни
  AlpinDale 3e6addcc2c LLM: enable batched inference for llm.chat() API (#1120) преди 5 дни
  AlpinDale fa84f8102e kernels: split marlin kernels for faster compile, fix MoE, temporarily remove HQQ (#1119) преди 5 дни
  AlpinDale 6b75a66c60 fix: unsafe all-reduce sync (#1118) преди 6 дни
  AlpinDale 80be38ca6f chore: expose phi3_v num_crops as an mm_processor_kwargs (#1117) преди 1 седмица
  AlpinDale 18e0b0e932 chore: support loading weights by ID within models (#1116) преди 1 седмица
  AlpinDale 949f974c59 (1/N) XQA: integrate the XQA CUDA kernels within Aphrodite (#1115) преди 1 седмица
  AlpinDale 2cd6ef2d5a misc: skip dumping inputs when unpicklable преди 2 седмици
  AlpinDale be0b0c13ca tests: update scheduler tests (#1113) преди 2 седмици
  AlpinDale 5b03d67abb Core: add output streaming support to multi-step + async (#1112) преди 2 седмици
  AlpinDale b20c4570d2 CI: bump aphrodite-engine to v0.6.6 (#1111) преди 2 седмици
  AlpinDale 7c825e50be fix: correct FP8 support check on Ada+ GPUs by using compressed-tensors (#1110) преди 2 седмици
  AlpinDale 2bb9c9c399 Revert "CI: use self-hosted runner for the build job" преди 2 седмици
  AlpinDale c71d2cf814 CI: use self-hosted runner for the build job преди 2 седмици
  AlpinDale 5c00851691 tests: fix ruff for llava onevision tests преди 2 седмици
  AlpinDale 6d8df254c7 LoRA: skip loading unsupported weight modules (#1109) преди 2 седмици
  AlpinDale f20f5c3491 samplers: improved DRY performance (#1108) преди 2 седмици
  AlpinDale 2dc917fcfd fix: install the headless opencv преди 2 седмици
  AlpinDale eb1ffacf74 Spec Decoding: fix typical acceptance sampler with correct recovered tok IDs (#1106) преди 2 седмици
  AlpinDale 76088aa43a distributed: allow IPv6 in APHRODITE_HOST_IP with ZMQ (#1105) преди 2 седмици
  AlpinDale 69cf654901 LoRA: add assertions for SGMV kernels to avoid incorrect results (#1104) преди 2 седмици
  AlpinDale c90abcc603 VLM: add pipeline parallelism support for Qwen2-VL (#1103) преди 2 седмици
  AlpinDale cc5e185795 VLM: support passing multimodal processor kwargs (#1102) преди 2 седмици
  AlpinDale 1448857bd3 XPU: fix docker build преди 2 седмици
  AlpinDale c36dd3a4b6 build: fix CPU CMake compilation преди 2 седмици
  AlpinDale 98e174b1f4 build: fix cutlass fetch warning преди 2 седмици