Code

1.18 habana_main merge#307

Code

Closed

kzawora-intel wants to merge 16 commits intov1.18.0HabanaAI/vllm-fork:v1.18.0from habana_mainHabanaAI/vllm-fork:habana_mainCopy head branch name to clipboard

+248-664

Commits

Commits on Sep 12, 2024

• some update to vision model

  Show description for aa4c59c
  xuechendi

  committed
  aa4c59c

  Copy full SHA for aa4c59c

• resolve conflicts

  Show description for 181babf
  xuechendi

  committed
  181babf

  Copy full SHA for 181babf

Commits on Sep 13, 2024

• fix ruff detected format error

  Show description for 1a35da2
  xuechendi

  committed
  1a35da2

  Copy full SHA for 1a35da2

• fix mypy format error

  Show description for 3b710a6
  xuechendi

  committed
  3b710a6

  Copy full SHA for 3b710a6

Commits on Sep 17, 2024

• [Bugfix][Habana_main] fix multi-modal model inference - tested with llava-1.5 (#283)

  Show description for 44c4f93
  michalkuligowski

  authored
  44c4f93

  Copy full SHA for 44c4f93

• Add fake HPU mode to Habana components with dummy habana_frameworks module. (#250)

  Show description for a9de5ba
  

  jmaksymczuk

  and

  kzawora-intel

  authored
  a9de5ba

  Copy full SHA for a9de5ba

• Update documentation on support of fp8 (#288)

  Show description for d39298c
  michalkuligowski

  authored
  d39298c

  Copy full SHA for d39298c

• Reduce default value of VLLM_GRAPH_RESERVED_MEM to 0.1

  kzawora-intel

  committed
  ed19acd

  Copy full SHA for ed19acd

• Removed vllm.hpu directory and changed relevant imports (#291)

  Show description for 6a96d9b
  tzielinski-habana

  authored
  6a96d9b

  Copy full SHA for 6a96d9b

• Reduce default value of VLLM_GRAPH_RESERVED_MEM to 0.1 (#292)

  Show description for 47a89be
  michalkuligowski

  authored
  47a89be

  Copy full SHA for 47a89be

• fix minor logging issue

  schoi-habana

  committed
  18d6339

  Copy full SHA for 18d6339

Commits on Sep 18, 2024

• Fix minor logging issue in habana_model_runner.py (#294)

  Show description for 83b54e9
  michalkuligowski

  authored
  83b54e9

  Copy full SHA for 83b54e9

• Fix blocks number calculation for Flat PA (#269)

  Show description for b62fba8
  iboiko-habana

  authored
  b62fba8

  Copy full SHA for b62fba8

Commits on Sep 20, 2024

• Remove dummy seq group data creation from loop (#301)

  Show description for cd7b1c1
  iboiko-habana

  authored
  cd7b1c1

  Copy full SHA for cd7b1c1

• optimize qwen2 model on Gaudi (#233)

  Show description for 12d7033
  czhu15

  authored
  12d7033

  Copy full SHA for 12d7033

• fix bug: device_str in initialize_ray_cluster requires uppercase string (#297)

  Show description for bc39baa
  hlin99

  authored
  bc39baa

  Copy full SHA for bc39baa