Code

Add huggingface llava#524

Code

Open

jp1924 wants to merge 46 commits intolinkedin:mainlinkedin/Liger-Kernel:mainfrom jp1924:add_llavajp1924/Liger-Kernel:add_llavaCopy head branch name to clipboard

+952-7

Commits

Commits on Jan 8, 2025

• Add: LLaVa lce_loss

  jp1924

  committed
  f84c03f

  Copy full SHA for f84c03f

• Add: LLaVa monkey_patched

  jp1924

  committed
  4aaf1ef

  Copy full SHA for 4aaf1ef

Commits on Jan 9, 2025

• Add: llava config

  jp1924

  committed
  8a6f8cd

  Copy full SHA for 8a6f8cd

• Add: load_image_processing_config

  jp1924

  committed
  4477479

  Copy full SHA for 4477479

• Add: llava test

  jp1924

  committed
  f77d647

  Copy full SHA for f77d647

• Add: llava

  jp1924

  committed
  53efec3

  Copy full SHA for 53efec3

Commits on Jan 15, 2025

• Add: llava test

  jp1924

  committed
  1fadb0e

  Copy full SHA for 1fadb0e

• Add: llava test

  jp1924

  committed
  2f14cf6

  Copy full SHA for 2f14cf6

Commits on Jan 16, 2025

• Refactor: update llava model forward method and enhance loss handling

  jp1924

  committed
  a3db21c

  Copy full SHA for a3db21c

• debugging

  jp1924

  committed
  79e2102

  Copy full SHA for 79e2102

• Refactor: simplify lce_forward method and improve loss calculation

  jp1924

  committed
  80cfc08

  Copy full SHA for 80cfc08

• Refactor: clean up apply_liger_kernel_to_llava function by removing commented code and simplifying model type checks

  jp1924

  committed
  24ca457

  Copy full SHA for 24ca457

• rm debugging

  jp1924

  committed
  8c181a6

  Copy full SHA for 8c181a6

• Refactor: remove redundant import of llava_lce_forward in apply_liger_kernel_to_llava function

  jp1924

  committed
  24bc56c

  Copy full SHA for 24bc56c

• Refactor: update mini_llava model configuration and add test cases for bfloat16 support

  jp1924

  committed
  42dbe36

  Copy full SHA for 42dbe36

• Fix: loss value error

  jp1924

  committed
  f6ba33f

  Copy full SHA for f6ba33f

• Refactor: add new processor and tokenizer configuration files, remove outdated configurations

  jp1924

  committed
  9223959

  Copy full SHA for 9223959

• Refactor: clean up lce_forward function and update apply_liger_kernel_to_llava for model type handling

  jp1924

  committed
  61476e8

  Copy full SHA for 61476e8

• Add: processor configuration loading function and clean up model configuration parameters

  jp1924

  committed
  22cafbf

  Copy full SHA for 22cafbf

• Merge branch 'main' into add_llava

  jp1924

  authored
  7f61b13

  Copy full SHA for 7f61b13

Commits on Jan 20, 2025

• Fix: typo Qwen2-VL -> LLaVa

  jp1924

  committed
  21bdc0a

  Copy full SHA for 21bdc0a

• Refactor: remove unused input list from run_mini_model function

  jp1924

  committed
  37857ef

  Copy full SHA for 37857ef

• Add model initialization for llava in multimodal tests

  jp1924

  committed
  23403c5

  Copy full SHA for 23403c5

• Add support for deprecated llava forward function and warning for legacy transformer versions

  jp1924

  committed
  5fb2853

  Copy full SHA for 5fb2853

Commits on Jan 21, 2025

• Merge branch 'main' into add_llava

  lancerts

  authored
  8289347

  Copy full SHA for 8289347

Commits on Jan 22, 2025

• Merge branch 'main' into add_llava

  jp1924

  authored
  dbc1cdd

  Copy full SHA for dbc1cdd

Commits on Jan 23, 2025

• Clean: unused module

  jp1924

  committed
  5adf3e2

  Copy full SHA for 5adf3e2

• Update: validate kwargs & model

  jp1924

  committed
  5b72328

  Copy full SHA for 5b72328

• Merge branch 'main' into add_llava

  jp1924

  authored
  9b1f929

  Copy full SHA for 9b1f929

• Fix: incorrect model input

  jp1924

  committed
  35c1788

  Copy full SHA for 35c1788

• Update: enhance documentation for Llava

  jp1924

  committed
  20e87db

  Copy full SHA for 20e87db

• Clean: check model

  jp1924

  committed
  419467e

  Copy full SHA for 419467e

Commits on Jan 24, 2025

• Merge branch 'main' into add_llava

  jp1924

  authored
  a12c95b

  Copy full SHA for a12c95b

• Fix: conflict

  jp1924

  committed
  5801eb9

  Copy full SHA for 5801eb9

• revert change

  jp1924

  committed
  a980a88

  Copy full SHA for a980a88

Commits on Jan 26, 2025

• solve conflict

  jp1924

  committed
  bcca064

  Copy full SHA for bcca064

• solve conflict

  jp1924

  committed
  4fe5307

  Copy full SHA for 4fe5307

• fix conflict

  jp1924

  committed
  5dae0ff

  Copy full SHA for 5dae0ff

• Merge branch 'main' into add_llava

  jp1924

  authored
  dcef62f

  Copy full SHA for dcef62f

• solve conflict

  jp1924

  committed
  eaef787

  Copy full SHA for eaef787

• Add: load_processor_config, load_image_processing_config

  jp1924

  committed
  90587e8

  Copy full SHA for 90587e8

• Add: revert_liger_kernel_to_llava

  jp1924

  committed
  ec94e9c

  Copy full SHA for ec94e9c

Commits on Jan 28, 2025

• apply lint

  jp1924

  committed
  34db41f

  Copy full SHA for 34db41f

• resolve conflict

  jp1924

  committed
  f29dce4

  Copy full SHA for f29dce4

• Merge branch 'main' into add_llava

  jp1924

  authored
  38ef343

  Copy full SHA for 38ef343

Commits on Jan 30, 2025

• Merge branch 'main' into add_llava

  jp1924

  authored
  3f360b5

  Copy full SHA for 3f360b5