Skip to content

Issues: huggingface/transformers

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

993 Open 15,195 Closed
993 Open 15,195 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Fine tuning SAM, 'The input_points must be a 3D tensor. Of shape batch_size, nb_boxes, 4.', ' got torch.Size([2, 4]).'
#34862 opened Nov 21, 2024 by leemorton
1
Offline mode doesn't work with models that require trust_remote_code=True bug
#34855 opened Nov 21, 2024 by VladKha
2 of 4 tasks
QuantizedCache fails with latest versions of optimum-quanto bug
#34848 opened Nov 21, 2024 by SimJeg
1 of 4 tasks
2
model.dequantize() does not remove quantization_config from model.config which effects in error when loading the model bug Good First Issue
#34847 opened Nov 21, 2024 by konradkalita
1 of 4 tasks
1
1
change inputs_embeds to input_embeds Feature request Request for a new feature
#34846 opened Nov 21, 2024 by Luohongzhige
1
When set num_beams in GenerationConfig, stop_strings parameter has no effect bug
#34843 opened Nov 21, 2024 by ZYM66
2 of 4 tasks
1
Urgent: 429 Errors and Potential Rate Limiting on Model Access bug
#34841 opened Nov 21, 2024 by dnuanghsh
2 of 4 tasks
2
CVE-2024-11392/11393/11394 vulnerabilities Feature request Request for a new feature
#34840 opened Nov 21, 2024 by JWYang0329
1
latest transformer has a tensorflow and numpy error bug
#34839 opened Nov 21, 2024 by sleepingcat4
1 of 4 tasks
3
Setting the attention_mask and past_key_values simultaneously will cause the error. bug
#34835 opened Nov 20, 2024 by MikeDean2367
4 tasks done
1
HfArgumentParser error when using LoraConfig dataclass bug
#34834 opened Nov 20, 2024 by JoseSantosAMD
2 of 4 tasks
2
Unable to import SamModel bug
#34831 opened Nov 20, 2024 by leemorton
4 tasks
1
5
Data collator class type integrity is not intact throughout the runtime bug
#34830 opened Nov 20, 2024 by kmehant
2 of 4 tasks
1
Flash attention 2 broke when batch inference bug Multimodal Vision
#34824 opened Nov 20, 2024 by pspdada
2 of 4 tasks
Maybe memory leak occurs after evaluation when using use_liger_kernel. bug
#34822 opened Nov 20, 2024 by upskyy
1 of 4 tasks
5
query_length not found in BLIP-2 config
#34821 opened Nov 20, 2024 by thisisiron
1
HW based PIL images not being handled in pretrained image processor bug
#34820 opened Nov 20, 2024 by alem-147
2 of 4 tasks
4
Qwen2 tests failing due to invalid model identifier bug Tests Related to tests
#34818 opened Nov 20, 2024 by jla524
1 of 4 tasks
1
1
Mamba2 torch_forward reduction dimension possibly incorrect? bug
#34817 opened Nov 19, 2024 by HanGuo97
1 of 4 tasks
5
SeamlessM4TForTextToSpeech.generate not working if generation_config is passed bug
#34811 opened Nov 19, 2024 by ydshieh
3
Flex attention + refactor Feature request Request for a new feature Good Difficult Issue PyTorch Anything PyTorch
#34809 opened Nov 19, 2024 by ArthurZucker
1 of 7 tasks
4
The usage of the "forced_decoder_ids" parameter Feature request Request for a new feature
#34808 opened Nov 19, 2024 by Alsace08
3
Loading bigger models is very slow using AutoModelForCausalLM.from_pretrained bug
#34798 opened Nov 19, 2024 by vibhas-singh
2 of 4 tasks
2
Allow to give the dataset multiprocessing_context Feature request Request for a new feature
#34793 opened Nov 18, 2024 by ierezell
1
CLIP Tokenizer Returns Different DTypes for Numpy versus Pytorch bug
#34790 opened Nov 18, 2024 by ivanthewebber
2 of 4 tasks
3
ProTip! Find all open issues with in progress development work with linked:pr.