Code

add Glm#33823

Code

Merged

Cyrilvallez merged 75 commits intomainhuggingface/transformers:mainfrom glmhuggingface/transformers:glmCopy head branch name to clipboardOct 18, 2024

+2,975-4

Commits

Commits on Oct 8, 2024

• Create modular_glm.py

  Cyrilvallez

  committed
  6366077

• Update modular_glm.py

  Cyrilvallez

  committed
  fa83dab

• Finalize architecture without all attentions

  Cyrilvallez

  committed
  11d74d9

• Add all attentions modules

  Cyrilvallez

  committed
  a3587d3

• Finalize modular

  Cyrilvallez

  committed
  9430273

• Update given last version

  Cyrilvallez

  committed
  4efd782

• Last update

  Cyrilvallez

  committed
  e9efed1

• Finalize model

  Cyrilvallez

  committed
  a411a38

• Finalize converter

  Cyrilvallez

  committed
  c4d3c21

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  1aba7c4

• style

  Cyrilvallez

  committed
  ea4d61a

• style

  Cyrilvallez

  committed
  b8eed8c

• Create __init__.py

  Cyrilvallez

  committed
  4f4e6f6

• Aff all inits

  Cyrilvallez

  committed
  de2e058

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  9392380

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  0e76b9d

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  02ae570

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  dee65b1

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  65e393d

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  60f0ca0

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  fa58a78

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  a39465a

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  db141cc

• Correct the rotary embeddings

  Cyrilvallez

  committed
  7f2427c

• Remove apply_residual_connection_post_layernorm (always false)

  Cyrilvallez

  committed
  0dc9819

• remove use_rms_norm (always true)

  Cyrilvallez

  committed
  931e28d

• remove past_layer_norm (always true)

  Cyrilvallez

  committed
  49c15d8

• Update __init__.py

  Cyrilvallez

  committed
  fc5cd41

• Update config and license

  Cyrilvallez

  committed
  f7220ba

• start adding tests and doc

  Cyrilvallez

  committed
  7c026f8

• Add doc + style

  Cyrilvallez

  committed
  3331d55

• Update test_modeling_glm.py

  Cyrilvallez

  committed
  a1fe067

• Add dummies

  Cyrilvallez

  committed
  7c26a24

• Apply correct modeling

  Cyrilvallez

  committed
  239e91b

• Refactor attention to follow llama

  Cyrilvallez

  committed
  89624b4

• Update __init__.py

  Cyrilvallez

  committed
  2d887c5

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  20ac604

• Correct bias

  Cyrilvallez

  committed
  cd92f6c

• remove linear_bias and pdrop (never used)

  Cyrilvallez

  committed
  e9c16e8

• apply modular

  Cyrilvallez

  committed
  fcd47ca

• Simplify converter

  Cyrilvallez

  committed
  9c41850

• remove dummies + style

  Cyrilvallez

  committed
  c286c6b

• add model_input_names

  Cyrilvallez

  committed
  63028c9

• Add pretraining_tp to config for when eager attention is used

  Cyrilvallez

  committed
  cce541b

• Update modular to remove all pretraining_tp

  Cyrilvallez

  committed
  307668a

• Update test_modeling_glm.py

  Cyrilvallez

  committed
  acf92e0

• Update the __all__

  Cyrilvallez

  committed
  ca6ba5e

• Update __all__

  Cyrilvallez

  committed
  0c6c8ef

• Update __init__.py

  Cyrilvallez

  committed
  5f76628

• Update test_modeling_glm.py

  Cyrilvallez

  committed
  3bca672

• add revisions

  Cyrilvallez

  committed
  4c8ee11

• Add the correct repos and revisions

  Cyrilvallez

  committed
  82b0757

• style

  Cyrilvallez

  committed
  e394eb8

• Update __init__.py

  Cyrilvallez

  committed
  ad68911

• update exports

  Cyrilvallez

  committed
  fd73ff9

• remove import of modular files

  Cyrilvallez

  committed
  f4494e9

• style

  Cyrilvallez

  committed
  e510503

• Apply Llama changes + refine converter

  Cyrilvallez

  committed
  2a6b0e3

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  7affb64

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  c8762d6

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  4792118

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  6b2f46b

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  e97978b

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  806d035

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  efe17f8

• Update convert_glm_weights_to_hf.py

  Cyrilvallez

  committed
  aab9dcf

• style

  Cyrilvallez

  committed
  4c14a98

• Use new modular converter

  Cyrilvallez

  committed
  409ca3e

• add pretrainedmodel to init

  Cyrilvallez

  committed
  35e9623

• style

  Cyrilvallez

  committed
  684016c

• Update test_modeling_glm.py

  Cyrilvallez

  committed
  764f755

Commits on Oct 18, 2024

• Move config outside modular to please CI about docstrings

  Cyrilvallez

  committed
  2739181

• Add dummies to please CI

  Cyrilvallez

  committed
  1d95488

• Update glm.md

  Cyrilvallez

  committed
  828509c

• Update glm.md

  Cyrilvallez

  committed
  ceec1f6