vllm
[Feature] Support Pipeline Parallism in torchrun SPMD offline inference for V1
#17827
Merged
Overview
Commits
10
Changes
View On GitHub
Commits
  • Support Pipeline Parallism in torchrun SPMD offline inference
    luccafong committed 1 year ago
  • format
    luccafong committed 1 year ago
  • remove config pipeline_parallel_broadcast_output
    luccafong committed 1 year ago
  • fix backend type
    luccafong committed 1 year ago
  • fix ray path
    luccafong committed 1 year ago
  • fix assertion
    luccafong committed 1 year ago
  • add test case for PP torchrun
    luccafong committed 1 year ago
  • Merge remote-tracking branch 'origin/main' into torch_run_pp
    luccafong committed 1 year ago
  • fix test example
    luccafong committed 1 year ago
  • Merge remote-tracking branch 'origin/main' into torch_run_pp
    luccafong committed 1 year ago
FAQ Terms Privacy Refunds Impressum
Loading