huggingface/transformers

Pull Requests Commits

Test hfh v0.15.0.rc0

Wauplin committed 2 years ago

9edc4a42

Skip device placement for past key values in decoder models (#23919)

sgugger committed 2 years ago

Verified fabe17a7

[PushToHub] Make it possible to upload folders (#23920)

NielsRogge committed 2 years ago

Verified 6affd9cd

Update the update metadata job to use upload_folder (#23917)

sgugger committed 2 years ago

Verified 4aa13224

Re-enable squad test (#23912)

sgugger committed 2 years ago

Verified 3ff443a6

remove the extra `accelerator.prepare` (#23914)

pacman100 committed 2 years ago

Verified d13021e3

Bug fix - flip_channel_order for channels first images (#23701)

amyeroberts committed 2 years ago

Verified c608b8fc

Empty circleci config (#23913)

sgugger committed 2 years ago

Verified 0b3d092f

Raise error if loss can't be calculated - ViT MIM (#23872)

amyeroberts committed 2 years ago

Verified 8714b964

add conditional statement for auxiliary loss calculation (#23899)

harisankar95 committed 2 years ago

Verified 404d9253

[`RWKV`] Fix RWKV 4bit (#23910)

younesbelkada committed 2 years ago

Verified c63bfc30

Upgrade safetensors version (#23911)

muellerzr committed 2 years ago

Verified 55451c66

fix: Replace `add_prefix_space` in `get_prompt_ids` with manual space for FastTokenizer compatibility (#23796)

Connor Henderson committed 2 years ago

Verified 7adce8b5

Move import check to before state reset (#23906)

muellerzr committed 2 years ago

Verified 84bac652

[`bnb`] add warning when no linear (#23894)

younesbelkada committed 2 years ago

Verified e42869b0

Unpin numba (#23162)

sanchit-gandhi committed 2 years ago

Verified 8f915c45

ensure banned_mask and indices in same device (#23901)

cauyxy committed 2 years ago

Verified d99f11e8

Support shared tensors (#23871)

thomasw21 committed 2 years ago

Verified d68d6665

Fix Trainer when model is loaded on a different GPU (#23792)

sgugger committed 2 years ago

Verified 68d53bc7

fix(configuration_llama): add `keys_to_ignore_at_inference` to `LlamaConfig` (#23891)

calico-1226 committed 2 years ago

Verified 0963a250

Skip failing test for now

sgugger committed 2 years ago

Verified 00f6ba0e

accelerate deepspeed and gradient accumulation integrate (#23236)

pacman100 committed 2 years ago

Verified a73b1d59

Add TensorFlow implementation of EfficientFormer (#22620)

D-Roberts committed 2 years ago

Verified 88f50a1e

Fix last instances of kbit -> quantized (#23797)

sgugger committed 2 years ago

Verified 9fea71b4

Fix bug leading to missing token in GPTSanJapaneseTokenizer (#23883)

passaglia committed 2 years ago

Verified 38dbbc26

shift torch dynamo handling to accelerate (#23168)

pacman100 committed 2 years ago

Verified 03db5910

move fsdp handling to accelerate (#23158)

pacman100 committed 2 years ago

Verified 0b774074

🌐 [i18n-KO] Translated `pad_truncation.mdx` to Korean (#23823)

sim-so committed 2 years ago

Verified 015829e6

Smangrul/accelerate ddp integrate (#23151)

pacman100 committed 2 years ago

Verified 1cf148a6

Smangrul/accelerate mp integrate (#23148)

pacman100 committed 2 years ago

Verified 9f0646a5

Older