Code

Upgrade to Transformers v4.48#1698

Code

Draft

regisss wants to merge 64 commits intomainhuggingface/optimum-habana:mainfrom transformers_futurehuggingface/optimum-habana:transformers_futureCopy head branch name to clipboard

+2,543-2,002

Commits

Commits on Sep 2, 2024

• Upgrade to commit 74e19e81e2a23809af192532b9b0e7ea202be6f2

  regisss

  committed
  5e88ce7

  Copy full SHA for 5e88ce7

• Merge branch 'main' into transformers_future

  regisss

  committed
  43bc4eb

  Copy full SHA for 43bc4eb

Commits on Sep 3, 2024

• Add specific commit in setup.py

  regisss

  committed
  8eea643

  Copy full SHA for 8eea643

Commits on Sep 6, 2024

• Upgrade to commit e48e5f1f13e05380e24f4f31f5fee07aa6f959eb

  regisss

  committed
  a7be363

  Copy full SHA for a7be363

• Merge branch 'main' into transformers_future

  regisss

  committed
  f0b909a

  Copy full SHA for f0b909a

Commits on Sep 9, 2024

• Fix default cache

  regisss

  committed
  d99f18f

  Copy full SHA for d99f18f

• Merge branch 'main' into transformers_future

  regisss

  committed
  39b7a76

  Copy full SHA for 39b7a76

• Merge branch 'main' into transformers_future

  regisss

  committed
  da66ecf

  Copy full SHA for da66ecf

Commits on Sep 10, 2024

• Merge branch 'main' into transformers_future

  regisss

  committed
  5547767

  Copy full SHA for 5547767

• Merge branch 'main' into transformers_future

  regisss

  committed
  bf89e41

  Copy full SHA for bf89e41

Commits on Sep 24, 2024

• Merge branch 'main' into transformers_future

  regisss

  committed
  98b0da5

  Copy full SHA for 98b0da5

• Upgrade to commit 238b13478df209ab534f2195a397dc64a3930883

  regisss

  committed
  47ad03c

  Copy full SHA for 47ad03c

• Fix

  regisss

  committed
  94c23ba

  Copy full SHA for 94c23ba

Commits on Sep 25, 2024

• Upgrade to v4.45.0

  regisss

  committed
  c19dedd

  Copy full SHA for c19dedd

• Merge branch 'main' into transformers_future

  regisss

  committed
  c12fd7e

  Copy full SHA for c12fd7e

• Fix

  regisss

  committed
  fc399fa

  Copy full SHA for fc399fa

Commits on Sep 26, 2024

• Add bias to gptj (#1363)

  jiminha

  authored
  9216159

  Copy full SHA for 9216159

• Switch roberta from sdpa to eager attn (#1361)

  skaulintel

  authored
  679365a

  Copy full SHA for 679365a

• Update bloom attention forward reshape follwing the transformer change (#1360)

  yeonsily

  authored
  1abd6ee

  Copy full SHA for 1abd6ee

• Workaround for Llava/Llava-next

  regisss

  committed
  8043d2c

  Copy full SHA for 8043d2c

Commits on Sep 28, 2024

• Fix reshape error in mamba (#1369)

  hsubramony

  authored
  047e7ff

  Copy full SHA for 047e7ff

Commits on Sep 30, 2024

• Merge branch 'main' into transformers_future

  regisss

  committed
  f89e03b

  Copy full SHA for f89e03b

• Merge branch 'main' into transformers_future

  regisss

  committed
  2ae546a

  Copy full SHA for 2ae546a

Commits on Oct 1, 2024

• Fix contrastive search

  regisss

  committed
  1b8a3f7

  Copy full SHA for 1b8a3f7

• Fix local variable 'image_features' referenced before assignment (#1383)

  vidyasiv

  authored
  2332afb

  Copy full SHA for 2332afb

Commits on Oct 2, 2024

• Use model.generation_config instead of model.config (#1384)

  Show description for f62ecde
  

  hsubramony

  and

  regisss

  authored
  f62ecde

  Copy full SHA for f62ecde

• Make style

  regisss

  committed
  a8fb8ac

  Copy full SHA for a8fb8ac

Commits on Jan 3, 2025

• Merge branch 'main' into transformers_future

  regisss

  committed
  c7ff331

  Copy full SHA for c7ff331

Commits on Jan 7, 2025

• Upgrade to Transformers v4.47.1

  regisss

  committed
  dd07c16

  Copy full SHA for dd07c16

Commits on Jan 8, 2025

• Fix Transformers version to install

  regisss

  committed
  1924c89

  Copy full SHA for 1924c89

Commits on Jan 9, 2025

• Temporary workaround for GaudiTrainer

  regisss

  committed
  f0926ae

  Copy full SHA for f0926ae

Commits on Jan 10, 2025

• Fixes for text generation

  regisss

  committed
  e50e179

  Copy full SHA for e50e179

• Set eager attention for distilbert, gpt_neox

  regisss

  committed
  c804270

  Copy full SHA for c804270

Commits on Jan 15, 2025

• Upgrade to Transformers v4.48

  regisss

  committed
  0000de5

  Copy full SHA for 0000de5

• Small fixes

  regisss

  committed
  2a3affa

  Copy full SHA for 2a3affa

• Fix integration tests

  regisss

  committed
  064f4c1

  Copy full SHA for 064f4c1

Commits on Jan 16, 2025

• Fixes for text-generation

  regisss

  committed
  21714f7

  Copy full SHA for 21714f7

• Fixes

  regisss

  committed
  1cfd53b

  Copy full SHA for 1cfd53b

• Style

  regisss

  committed
  573cc57

  Copy full SHA for 573cc57

• Again

  regisss

  committed
  a7bc517

  Copy full SHA for a7bc517

• Merge branch 'main' into transformers_future

  regisss

  committed
  3d402ea

  Copy full SHA for 3d402ea

Commits on Jan 28, 2025

• Fix for image2text lora llama test (#1731)

  vidyasiv

  authored
  f69e957

  Copy full SHA for f69e957

Commits on Jan 30, 2025

• Cherry-pick huggingface/transformers#35651

  regisss

  committed
  265e6a1

  Copy full SHA for 265e6a1

• Merge branch 'main' into transformers_future

  regisss

  committed
  6dad1c4

  Copy full SHA for 6dad1c4

Commits on Jan 31, 2025

• Upgrade to Transformers v4.48.2

  regisss

  committed
  32478f5

  Copy full SHA for 32478f5

• Fix deprecated imports following merged changes for DETR and Qwen2-VL

  regisss

  committed
  1b79cf3

  Copy full SHA for 1b79cf3

• Workaround for textual inversion

  regisss

  committed
  c1f30d8

  Copy full SHA for c1f30d8

• Merge branch 'main' into transformers_future

  regisss

  committed
  1460856

  Copy full SHA for 1460856

• Fixes for v4.48 pytest (#1699)

  imangohari1

  authored
  7eadac6

  Copy full SHA for 7eadac6

• fea(): Applied changes in HF #35235 (#1738)

  imangohari1

  authored
  5cee218

  Copy full SHA for 5cee218

• Merge branch 'main' into transformers_future

  regisss

  committed
  417cbee

  Copy full SHA for 417cbee

Commits on Feb 3, 2025

• Merge branch 'main' into transformers_future

  regisss

  committed
  c1e3232

  Copy full SHA for c1e3232

Commits on Feb 4, 2025

• Merge branch 'main' into transformers_future

  regisss

  committed
  0f68bbb

  Copy full SHA for 0f68bbb

Commits on Feb 5, 2025

• Removing HL_DS_DISTRIBUTED_ATTENTION_SEQ_DIM as it's not needed from SynapseAI 1.20 (#1726)

  bhargaveede

  authored
  17943de

  Copy full SHA for 17943de

• Update DS config to align with recommended settings (#1730)

  ckvermaAI

  authored
  d214819

  Copy full SHA for d214819

• Fix graph breaks in Mixtral (#65) (#1705)

  ShengYang1

  authored
  6a520ff

  Copy full SHA for 6a520ff

• Merge branch 'main' into synapse_1_20

  regisss

  committed
  58de6b6

  Copy full SHA for 58de6b6

Commits on Feb 6, 2025

• Add batch dim idx to support latest deepspeed DistributedAttention (#1725)

  bhargaveede

  authored
  bedc041

  Copy full SHA for bedc041

Commits on Feb 7, 2025

• Add _prepare_inputs_for_generation (#1743)

  yafshar

  authored
  ce57e40

  Copy full SHA for ce57e40

• Merge branch 'main' into synapse_1_20

  regisss

  committed
  ef77fac

  Copy full SHA for ef77fac

• Upgrade to v4.48.3

  regisss

  committed
  be34027

  Copy full SHA for be34027

• Fix the issue with --load_quantized_model_with_autoawq (#1747)

  Show description for bd9a60e
  

  schoi-habana

  and

  regisss

  authored
  bd9a60e

  Copy full SHA for bd9a60e

• Merge branch 'main' into synapse_1_20

  regisss

  committed
  fc6a92b

  Copy full SHA for fc6a92b

• Merge branch 'synapse_1_20' into transformers_future

  regisss

  committed
  01bb4af

  Copy full SHA for 01bb4af