url string | repository_url string | labels_url string | comments_url string | events_url string | html_url string | id int64 | node_id string | number int64 | title string | user dict | labels list | state string | locked bool | assignee dict | assignees list | milestone null | comments list | created_at timestamp[ms] | updated_at timestamp[ms] | closed_at timestamp[ms] | author_association string | type dict | active_lock_reason null | draft bool | pull_request dict | body string | closed_by dict | reactions dict | timeline_url string | performed_via_github_app null | state_reason string | sub_issues_summary dict | issue_dependencies_summary dict | is_pull_request bool | is_closed bool |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://api.github.com/repos/huggingface/transformers/issues/39336 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39336/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39336/comments | https://api.github.com/repos/huggingface/transformers/issues/39336/events | https://github.com/huggingface/transformers/issues/39336 | 3,219,035,401 | I_kwDOCUB6oc6_3pUJ | 39,336 | TypeError: GenerationMixin._extract_past_from_model_output() got an unexpected keyword argument 'standardize_cache_format' | {
"login": "FurkanGozukara",
"id": 19240467,
"node_id": "MDQ6VXNlcjE5MjQwNDY3",
"avatar_url": "https://avatars.githubusercontent.com/u/19240467?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/FurkanGozukara",
"html_url": "https://github.com/FurkanGozukara",
"followers_url": "https://api.github.com/users/FurkanGozukara/followers",
"following_url": "https://api.github.com/users/FurkanGozukara/following{/other_user}",
"gists_url": "https://api.github.com/users/FurkanGozukara/gists{/gist_id}",
"starred_url": "https://api.github.com/users/FurkanGozukara/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/FurkanGozukara/subscriptions",
"organizations_url": "https://api.github.com/users/FurkanGozukara/orgs",
"repos_url": "https://api.github.com/users/FurkanGozukara/repos",
"events_url": "https://api.github.com/users/FurkanGozukara/events{/privacy}",
"received_events_url": "https://api.github.com/users/FurkanGozukara/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-10T11:49:02 | 2025-08-18T08:03:13 | 2025-08-18T08:03:13 | NONE | null | null | null | null | I am using CogVLM2 video captioning model
It works latest with transformers==4.43.4
with transformers==4.44.0 and forward I get below error
but I need to use latest version of transformers since currently 4bit quantization fails on some gpus and platforms
how can i fix this issue?
`TypeError: GenerationMixin._extract_past_from_model_output() got an unexpected keyword argument 'standardize_cache_format'`
```
14:23:32 - INFO - Final video tensor shape for CogVLM processing: torch.Size([3, 24, 720, 1280])
14:23:35 - ERROR - Error during auto-captioning: GenerationMixin._extract_past_from_model_output() got an unexpected keyword argument 'standardize_cache_format'
Traceback (most recent call last):
File "E:\Ultimate_Video_Processing_v1\STAR\logic\cogvlm_utils.py", line 679, in auto_caption
outputs_tensor = local_model_ref.generate(**inputs_on_device, **gen_kwargs)
File "E:\Ultimate_Video_Processing_v1\venv\lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "E:\Ultimate_Video_Processing_v1\venv\lib\site-packages\transformers\generation\utils.py", line 2024, in generate
result = self._sample(
File "E:\Ultimate_Video_Processing_v1\venv\lib\site-packages\transformers\generation\utils.py", line 3032, in _sample
model_kwargs = self._update_model_kwargs_for_generation(
File "E:\Ultimate_Video_Processing_v1\STAR\models\modules\transformers_modules\cogvlm2-video-llama3-chat\modeling_cogvlm.py", line 726, in _update_model_kwargs_for_generation
cache_name, cache = self._extract_past_from_model_output(
TypeError: GenerationMixin._extract_past_from_model_output() got an unexpected keyword argument 'standardize_cache_format'
```
@amyeroberts, @qubvel @SunMarc @MekkCyber
the error i am getting is below with 4.43.1 on B200 when doing 4bit quant. interesting same code same libraries on my rtx 5090 on windows working without errors
fp16 has no issues
```
11:45:10 - INFO - Preparing to load model from: /workspace/STAR/models/cogvlm2-video-llama3-chat with quant: 4, dtype: torch.bfloat16, device: cuda, device_map: auto, low_cpu_mem: True
11:45:10 - INFO - Starting model loading - this operation cannot be interrupted once started
/workspace/venv/lib/python3.10/site-packages/torchvision/transforms/_functional_video.py:6: UserWarning: The 'torchvision.transforms._functional_video' module is deprecated since 0.12 and will be removed in the future. Please use the 'torchvision.transforms.functional' module instead.
warnings.warn(
/workspace/venv/lib/python3.10/site-packages/torchvision/transforms/_transforms_video.py:22: UserWarning: The 'torchvision.transforms._transforms_video' module is deprecated since 0.12 and will be removed in the future. Please use the 'torchvision.transforms' module instead.
warnings.warn(
Loading checkpoint shards: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 6/6 [01:18<00:00, 13.07s/steps]
11:46:30 - ERROR - Failed to load CogVLM2 model from path: /workspace/STAR/models/cogvlm2-video-llama3-chat
11:46:30 - ERROR - Exception type: ValueError
11:46:30 - ERROR - Exception details: `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.
Traceback (most recent call last):
File "/workspace/STAR/logic/cogvlm_utils.py", line 160, in load_cogvlm_model
raise model_loading_result["error"]
File "/workspace/STAR/logic/cogvlm_utils.py", line 122, in load_model_thread
model = AutoModelForCausalLM.from_pretrained(
File "/workspace/venv/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 559, in from_pretrained
return model_class.from_pretrained(
File "/workspace/venv/lib/python3.10/site-packages/transformers/modeling_utils.py", line 4000, in from_pretrained
dispatch_model(model, **device_map_kwargs)
File "/workspace/venv/lib/python3.10/site-packages/accelerate/big_modeling.py", line 502, in dispatch_model
model.to(device)
File "/workspace/venv/lib/python3.10/site-packages/transformers/modeling_utils.py", line 2849, in to
raise ValueError(
ValueError: `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.
11:46:30 - ERROR - Error during auto-captioning: 'Could not load CogVLM2 model (check logs for details): `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.'
Traceback (most recent call last):
File "/workspace/STAR/logic/cogvlm_utils.py", line 160, in load_cogvlm_model
raise model_loading_result["error"]
File "/workspace/STAR/logic/cogvlm_utils.py", line 122, in load_model_thread
model = AutoModelForCausalLM.from_pretrained(
File "/workspace/venv/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 559, in from_pretrained
return model_class.from_pretrained(
File "/workspace/venv/lib/python3.10/site-packages/transformers/modeling_utils.py", line 4000, in from_pretrained
dispatch_model(model, **device_map_kwargs)
File "/workspace/venv/lib/python3.10/site-packages/accelerate/big_modeling.py", line 502, in dispatch_model
model.to(device)
File "/workspace/venv/lib/python3.10/site-packages/transformers/modeling_utils.py", line 2849, in to
raise ValueError(
ValueError: `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/workspace/STAR/logic/cogvlm_utils.py", line 478, in auto_caption
local_model_ref, local_tokenizer_ref = load_cogvlm_model(quantization, cogvlm_device, cog_vlm_model_path, logger)
File "/workspace/STAR/logic/cogvlm_utils.py", line 243, in load_cogvlm_model
raise gr.Error(f"Could not load CogVLM2 model (check logs for details): {str(e)[:200]}")
gradio.exceptions.Error: 'Could not load CogVLM2 model (check logs for details): `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.'
11:46:30 - INFO - Unloading CogVLM2 model with strategy: full
```
| {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39336/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39336/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39335 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39335/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39335/comments | https://api.github.com/repos/huggingface/transformers/issues/39335/events | https://github.com/huggingface/transformers/pull/39335 | 3,218,928,455 | PR_kwDOCUB6oc6eSMTy | 39,335 | [BLIP] remove cache from Qformer | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T11:15:24 | 2025-07-14T07:20:02 | 2025-07-14T07:20:02 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39335",
"html_url": "https://github.com/huggingface/transformers/pull/39335",
"diff_url": "https://github.com/huggingface/transformers/pull/39335.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39335.patch",
"merged_at": "2025-07-14T07:20:02"
} | # What does this PR do?
Qformer doesn't use cache and is not generative, this PR deprecates it for one version. We'll remove all usage of cache in non-generative models after 4.55 :)
| {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39335/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39335/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39334 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39334/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39334/comments | https://api.github.com/repos/huggingface/transformers/issues/39334/events | https://github.com/huggingface/transformers/issues/39334 | 3,218,910,250 | I_kwDOCUB6oc6_3Kwq | 39,334 | FlashAttention2 ImportError: undefined symbol with flash_attn_2_cuda when loading Phi-4-Multimodal | {
"login": "tanuj-rai",
"id": 84439872,
"node_id": "MDQ6VXNlcjg0NDM5ODcy",
"avatar_url": "https://avatars.githubusercontent.com/u/84439872?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/tanuj-rai",
"html_url": "https://github.com/tanuj-rai",
"followers_url": "https://api.github.com/users/tanuj-rai/followers",
"following_url": "https://api.github.com/users/tanuj-rai/following{/other_user}",
"gists_url": "https://api.github.com/users/tanuj-rai/gists{/gist_id}",
"starred_url": "https://api.github.com/users/tanuj-rai/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/tanuj-rai/subscriptions",
"organizations_url": "https://api.github.com/users/tanuj-rai/orgs",
"repos_url": "https://api.github.com/users/tanuj-rai/repos",
"events_url": "https://api.github.com/users/tanuj-rai/events{/privacy}",
"received_events_url": "https://api.github.com/users/tanuj-rai/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-10T11:10:35 | 2025-07-10T12:41:35 | 2025-07-10T12:41:35 | CONTRIBUTOR | null | null | null | null | ### System Info
Transformers version: 4.53.1
PyTorch version: 2.6.0+cu124
Python version: 3.11.13
CUDA version: 12.4
GPU: Tesla T4
### Who can help?
```
ImportError Traceback (most recent call last)
[/tmp/ipython-input-9-1994488020.py](https://localhost:8080/#) in <cell line: 0>()
2
3 import torch
----> 4 from transformers import AutoModelForCausalLM, AutoProcessor
5 import soundfile as sf
6
10 frames
[/usr/local/lib/python3.11/dist-packages/flash_attn/flash_attn_interface.py](https://localhost:8080/#) in <module>
13 from .flash_attn_triton_amd import interface_fa as flash_attn_gpu
14 else:
---> 15 import flash_attn_2_cuda as flash_attn_gpu
16
17 # isort: on
ImportError: /usr/local/lib/python3.11/dist-packages/flash_attn_2_cuda.cpython-311-x86_64-linux-gnu.so: undefined symbol: _ZN3c105ErrorC2ENS_14SourceLocationENSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE
`
### Information
- [ ] The official example scripts
- [ ] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
```
import torch
from transformers import AutoProcessor, AutoModelForCausalLM
import soundfile as sf
model_name = "microsoft/Phi-4-multimodal-instruct"
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", torch_dtype=torch.float16, trust_remote_code=True)
audio, sr = sf.read("your_audio.wav")
messages = [{
"role": "user",
"content": [
{"type": "audio", "array": audio, "sampling_rate": sr},
{"type": "text", "text": "Transcribe the audio."}
],
}]
inputs = processor.apply_chat_template(messages, tokenize=True, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=200)
response = processor.batch_decode(out[:, inputs["input_ids"].shape[1]:], skip_special_tokens=True)[0]
print("Model response:", response)
```
### Expected behavior
The code should work without error. | {
"login": "tanuj-rai",
"id": 84439872,
"node_id": "MDQ6VXNlcjg0NDM5ODcy",
"avatar_url": "https://avatars.githubusercontent.com/u/84439872?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/tanuj-rai",
"html_url": "https://github.com/tanuj-rai",
"followers_url": "https://api.github.com/users/tanuj-rai/followers",
"following_url": "https://api.github.com/users/tanuj-rai/following{/other_user}",
"gists_url": "https://api.github.com/users/tanuj-rai/gists{/gist_id}",
"starred_url": "https://api.github.com/users/tanuj-rai/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/tanuj-rai/subscriptions",
"organizations_url": "https://api.github.com/users/tanuj-rai/orgs",
"repos_url": "https://api.github.com/users/tanuj-rai/repos",
"events_url": "https://api.github.com/users/tanuj-rai/events{/privacy}",
"received_events_url": "https://api.github.com/users/tanuj-rai/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39334/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39334/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39333 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39333/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39333/comments | https://api.github.com/repos/huggingface/transformers/issues/39333/events | https://github.com/huggingface/transformers/pull/39333 | 3,218,731,982 | PR_kwDOCUB6oc6eRgO1 | 39,333 | [Qwen2.5-VL] Fix torch.finfo() TypeError for integer attention_mask_tensor | {
"login": "dsnsabari",
"id": 46018083,
"node_id": "MDQ6VXNlcjQ2MDE4MDgz",
"avatar_url": "https://avatars.githubusercontent.com/u/46018083?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dsnsabari",
"html_url": "https://github.com/dsnsabari",
"followers_url": "https://api.github.com/users/dsnsabari/followers",
"following_url": "https://api.github.com/users/dsnsabari/following{/other_user}",
"gists_url": "https://api.github.com/users/dsnsabari/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dsnsabari/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dsnsabari/subscriptions",
"organizations_url": "https://api.github.com/users/dsnsabari/orgs",
"repos_url": "https://api.github.com/users/dsnsabari/repos",
"events_url": "https://api.github.com/users/dsnsabari/events{/privacy}",
"received_events_url": "https://api.github.com/users/dsnsabari/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T10:15:20 | 2025-07-14T07:47:40 | 2025-07-14T07:47:39 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39333",
"html_url": "https://github.com/huggingface/transformers/pull/39333",
"diff_url": "https://github.com/huggingface/transformers/pull/39333.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39333.patch",
"merged_at": "2025-07-14T07:47:39"
} | What does this PR do?
This PR fixes a critical TypeError in the Qwen2.5-VL model that occurs when processing attention masks with integer dtypes. The error torch.finfo() requires a floating point input type was preventing model inference when using frameworks like Unsloth.
Problem
The issue occurs in modeling_qwen2_5_vl.py at line 1292 where torch.finfo() is called on attention mask tensors that may have integer dtypes:
pythonattention_mask_tensor = attention_mask_tensor / torch.finfo(attention_mask_tensor.dtype).min
Since torch.finfo() only supports floating-point types, this crashes when the attention mask tensor has an integer dtype.
Solution
Added dtype checking to use the appropriate function:
torch.iinfo() for integer dtypes
torch.finfo() for floating-point dtypes
This maintains backward compatibility while fixing the crash for integer attention masks.
Impact
β
Fixes model crashes during vision inference with Unsloth
β
Enables support for integer dtype attention masks
β
Maintains existing functionality for float dtype tensors
β
No breaking changes to existing code
<!-- Remove if not applicable -->
Fixes # (issue)
Before submitting
This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
Did you write any new necessary tests?
Additional Testing Details
β
Tested with integer attention mask dtypes (torch.int32, torch.int64)
β
Tested with floating-point attention mask dtypes (torch.float16, torch.float32)
β
Verified compatibility with Unsloth framework
β
Confirmed no regression in existing Qwen2.5-VL functionality
β
Added unit tests covering both dtype paths
Who can review?
@amyeroberts @qubvel - This is a vision model fix for Qwen2.5-VL dtype compatibility
Priority: High - This is a blocking issue that prevents model inference in common usage scenarios with Unsloth and other frameworks that may use integer attention masks.
Backward Compatibility: β
Fully backward compatible - no changes to existing API or behavior for floating-point tensors. | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39333/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39333/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39332 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39332/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39332/comments | https://api.github.com/repos/huggingface/transformers/issues/39332/events | https://github.com/huggingface/transformers/pull/39332 | 3,218,454,375 | PR_kwDOCUB6oc6eQjrm | 39,332 | Remove conditional generation in image-to-text pipelines | {
"login": "jiangwangyi",
"id": 39762734,
"node_id": "MDQ6VXNlcjM5NzYyNzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/39762734?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiangwangyi",
"html_url": "https://github.com/jiangwangyi",
"followers_url": "https://api.github.com/users/jiangwangyi/followers",
"following_url": "https://api.github.com/users/jiangwangyi/following{/other_user}",
"gists_url": "https://api.github.com/users/jiangwangyi/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiangwangyi/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiangwangyi/subscriptions",
"organizations_url": "https://api.github.com/users/jiangwangyi/orgs",
"repos_url": "https://api.github.com/users/jiangwangyi/repos",
"events_url": "https://api.github.com/users/jiangwangyi/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiangwangyi/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T08:46:34 | 2025-09-11T07:16:12 | 2025-09-11T07:16:12 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39332",
"html_url": "https://github.com/huggingface/transformers/pull/39332",
"diff_url": "https://github.com/huggingface/transformers/pull/39332.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39332.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
This PR removes the conditional generation implementation from `ImageToTextPipeline`.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
@amyeroberts, @qubvel | {
"login": "jiangwangyi",
"id": 39762734,
"node_id": "MDQ6VXNlcjM5NzYyNzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/39762734?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiangwangyi",
"html_url": "https://github.com/jiangwangyi",
"followers_url": "https://api.github.com/users/jiangwangyi/followers",
"following_url": "https://api.github.com/users/jiangwangyi/following{/other_user}",
"gists_url": "https://api.github.com/users/jiangwangyi/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiangwangyi/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiangwangyi/subscriptions",
"organizations_url": "https://api.github.com/users/jiangwangyi/orgs",
"repos_url": "https://api.github.com/users/jiangwangyi/repos",
"events_url": "https://api.github.com/users/jiangwangyi/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiangwangyi/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39332/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39332/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39331 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39331/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39331/comments | https://api.github.com/repos/huggingface/transformers/issues/39331/events | https://github.com/huggingface/transformers/pull/39331 | 3,218,422,619 | PR_kwDOCUB6oc6eQctW | 39,331 | Remove conditional generation in ImageToTextPipeline | {
"login": "jiangwangyi",
"id": 39762734,
"node_id": "MDQ6VXNlcjM5NzYyNzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/39762734?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiangwangyi",
"html_url": "https://github.com/jiangwangyi",
"followers_url": "https://api.github.com/users/jiangwangyi/followers",
"following_url": "https://api.github.com/users/jiangwangyi/following{/other_user}",
"gists_url": "https://api.github.com/users/jiangwangyi/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiangwangyi/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiangwangyi/subscriptions",
"organizations_url": "https://api.github.com/users/jiangwangyi/orgs",
"repos_url": "https://api.github.com/users/jiangwangyi/repos",
"events_url": "https://api.github.com/users/jiangwangyi/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiangwangyi/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T08:35:56 | 2025-07-10T08:41:23 | 2025-07-10T08:37:55 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39331",
"html_url": "https://github.com/huggingface/transformers/pull/39331",
"diff_url": "https://github.com/huggingface/transformers/pull/39331.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39331.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Conditional generation has been implemented in ImageTextToTextPipeline, so remove related contents in ImageToTextPipeline.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
@amyeroberts @qubvel | {
"login": "jiangwangyi",
"id": 39762734,
"node_id": "MDQ6VXNlcjM5NzYyNzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/39762734?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiangwangyi",
"html_url": "https://github.com/jiangwangyi",
"followers_url": "https://api.github.com/users/jiangwangyi/followers",
"following_url": "https://api.github.com/users/jiangwangyi/following{/other_user}",
"gists_url": "https://api.github.com/users/jiangwangyi/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiangwangyi/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiangwangyi/subscriptions",
"organizations_url": "https://api.github.com/users/jiangwangyi/orgs",
"repos_url": "https://api.github.com/users/jiangwangyi/repos",
"events_url": "https://api.github.com/users/jiangwangyi/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiangwangyi/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39331/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39331/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39330 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39330/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39330/comments | https://api.github.com/repos/huggingface/transformers/issues/39330/events | https://github.com/huggingface/transformers/pull/39330 | 3,218,336,239 | PR_kwDOCUB6oc6eQJ2J | 39,330 | Remove deprecated audio utils functions | {
"login": "jiangwangyi",
"id": 39762734,
"node_id": "MDQ6VXNlcjM5NzYyNzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/39762734?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiangwangyi",
"html_url": "https://github.com/jiangwangyi",
"followers_url": "https://api.github.com/users/jiangwangyi/followers",
"following_url": "https://api.github.com/users/jiangwangyi/following{/other_user}",
"gists_url": "https://api.github.com/users/jiangwangyi/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiangwangyi/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiangwangyi/subscriptions",
"organizations_url": "https://api.github.com/users/jiangwangyi/orgs",
"repos_url": "https://api.github.com/users/jiangwangyi/repos",
"events_url": "https://api.github.com/users/jiangwangyi/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiangwangyi/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T08:06:55 | 2025-07-15T14:03:02 | 2025-07-15T14:02:25 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39330",
"html_url": "https://github.com/huggingface/transformers/pull/39330",
"diff_url": "https://github.com/huggingface/transformers/pull/39330.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39330.patch",
"merged_at": "2025-07-15T14:02:25"
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Remove deprecated functions in `src/transformers/audio_utils.py`.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
@ArthurZucker @amyeroberts @qubvel
| {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39330/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39330/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39329 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39329/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39329/comments | https://api.github.com/repos/huggingface/transformers/issues/39329/events | https://github.com/huggingface/transformers/issues/39329 | 3,218,281,277 | I_kwDOCUB6oc6_0xM9 | 39,329 | Adding support for Gemma 3n GGUFs | {
"login": "minhnguyenhoang",
"id": 27799472,
"node_id": "MDQ6VXNlcjI3Nzk5NDcy",
"avatar_url": "https://avatars.githubusercontent.com/u/27799472?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/minhnguyenhoang",
"html_url": "https://github.com/minhnguyenhoang",
"followers_url": "https://api.github.com/users/minhnguyenhoang/followers",
"following_url": "https://api.github.com/users/minhnguyenhoang/following{/other_user}",
"gists_url": "https://api.github.com/users/minhnguyenhoang/gists{/gist_id}",
"starred_url": "https://api.github.com/users/minhnguyenhoang/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/minhnguyenhoang/subscriptions",
"organizations_url": "https://api.github.com/users/minhnguyenhoang/orgs",
"repos_url": "https://api.github.com/users/minhnguyenhoang/repos",
"events_url": "https://api.github.com/users/minhnguyenhoang/events{/privacy}",
"received_events_url": "https://api.github.com/users/minhnguyenhoang/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 2648621985,
"node_id": "MDU6TGFiZWwyNjQ4NjIxOTg1",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Feature%20request",
"name": "Feature request",
"color": "FBCA04",
"default": false,
"description": "Request for a new feature"
}
] | open | false | null | [] | null | [] | 2025-07-10T07:46:49 | 2025-07-24T09:58:08 | null | NONE | null | null | null | null | ### Feature request
There are now quantised GGUF versions of Gemma 3n models available on Hugging Face (e.g. https://huggingface.co/unsloth/gemma-3n-E2B-it-GGUF). Currently they can be run with other backends, but I'd like to have support for them with the `transformers` library.
If I run:
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
model_id = "unsloth/gemma-3n-E2B-it-GGUF"
gguf_file = "gemma-3n-E2B-it-Q4_K_M.gguf"
tokenizer = AutoTokenizer.from_pretrained(model_id, gguf_file=gguf_file)
model = AutoModelForCausalLM.from_pretrained(
model_id,
gguf_file=gguf_file,
torch_dtype=torch.bfloat16,
low_cpu_mem_usage=True)
```
It will return a `ValueError: GGUF model with architecture gemma3n is not supported yet.`
Side note: in my real implementation of this code I actually have the model downloaded manually from HF then used. However, the end result should be the same and replicable - the error doesn't suggest that my code has a bug.
### Motivation
I want to use the transformers backend to inference with this model, but I also want to use the GGUF models instead of the tensor version.
### Your contribution
I have minimal experience in AI/ML in general, but if there are resources/examples on previous works similar to this I'm willing to have a try at submitting a PR. | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39329/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39329/timeline | null | null | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | false |
https://api.github.com/repos/huggingface/transformers/issues/39327 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39327/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39327/comments | https://api.github.com/repos/huggingface/transformers/issues/39327/events | https://github.com/huggingface/transformers/pull/39327 | 3,217,961,482 | PR_kwDOCUB6oc6eO4yD | 39,327 | Fix `get_relative_import_files` for module import | {
"login": "MayDomine",
"id": 57244158,
"node_id": "MDQ6VXNlcjU3MjQ0MTU4",
"avatar_url": "https://avatars.githubusercontent.com/u/57244158?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MayDomine",
"html_url": "https://github.com/MayDomine",
"followers_url": "https://api.github.com/users/MayDomine/followers",
"following_url": "https://api.github.com/users/MayDomine/following{/other_user}",
"gists_url": "https://api.github.com/users/MayDomine/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MayDomine/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MayDomine/subscriptions",
"organizations_url": "https://api.github.com/users/MayDomine/orgs",
"repos_url": "https://api.github.com/users/MayDomine/repos",
"events_url": "https://api.github.com/users/MayDomine/events{/privacy}",
"received_events_url": "https://api.github.com/users/MayDomine/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T05:32:03 | 2025-07-10T05:48:44 | 2025-07-10T05:48:44 | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39327",
"html_url": "https://github.com/huggingface/transformers/pull/39327",
"diff_url": "https://github.com/huggingface/transformers/pull/39327.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39327.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes # (issue)
`get_relative_import_files` from dynamic_module_utils.py will crash when the `object_file` is a module instead a python file. This issue will make custom `save_function` for `save_pretrained` function of `PretrainedModel` crash.
This PR fix this function to make it possible to handle module import.
https://github.com/huggingface/transformers/blob/bc161d5d06ba3b40eaca99ce42734b4f1fecfaa2/src/transformers/dynamic_module_utils.py#L127C2-L135C63
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "MayDomine",
"id": 57244158,
"node_id": "MDQ6VXNlcjU3MjQ0MTU4",
"avatar_url": "https://avatars.githubusercontent.com/u/57244158?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MayDomine",
"html_url": "https://github.com/MayDomine",
"followers_url": "https://api.github.com/users/MayDomine/followers",
"following_url": "https://api.github.com/users/MayDomine/following{/other_user}",
"gists_url": "https://api.github.com/users/MayDomine/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MayDomine/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MayDomine/subscriptions",
"organizations_url": "https://api.github.com/users/MayDomine/orgs",
"repos_url": "https://api.github.com/users/MayDomine/repos",
"events_url": "https://api.github.com/users/MayDomine/events{/privacy}",
"received_events_url": "https://api.github.com/users/MayDomine/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39327/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39327/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39326 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39326/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39326/comments | https://api.github.com/repos/huggingface/transformers/issues/39326/events | https://github.com/huggingface/transformers/issues/39326 | 3,217,840,227 | I_kwDOCUB6oc6_zFhj | 39,326 | TypeError in Qwen2_5_VLForConditionalGeneration (torch.finfo misuse) | {
"login": "dsnsabari",
"id": 46018083,
"node_id": "MDQ6VXNlcjQ2MDE4MDgz",
"avatar_url": "https://avatars.githubusercontent.com/u/46018083?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dsnsabari",
"html_url": "https://github.com/dsnsabari",
"followers_url": "https://api.github.com/users/dsnsabari/followers",
"following_url": "https://api.github.com/users/dsnsabari/following{/other_user}",
"gists_url": "https://api.github.com/users/dsnsabari/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dsnsabari/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dsnsabari/subscriptions",
"organizations_url": "https://api.github.com/users/dsnsabari/orgs",
"repos_url": "https://api.github.com/users/dsnsabari/repos",
"events_url": "https://api.github.com/users/dsnsabari/events{/privacy}",
"received_events_url": "https://api.github.com/users/dsnsabari/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-10T04:24:55 | 2025-07-10T04:30:43 | 2025-07-10T04:30:43 | CONTRIBUTOR | null | null | null | null | null | {
"login": "dsnsabari",
"id": 46018083,
"node_id": "MDQ6VXNlcjQ2MDE4MDgz",
"avatar_url": "https://avatars.githubusercontent.com/u/46018083?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dsnsabari",
"html_url": "https://github.com/dsnsabari",
"followers_url": "https://api.github.com/users/dsnsabari/followers",
"following_url": "https://api.github.com/users/dsnsabari/following{/other_user}",
"gists_url": "https://api.github.com/users/dsnsabari/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dsnsabari/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dsnsabari/subscriptions",
"organizations_url": "https://api.github.com/users/dsnsabari/orgs",
"repos_url": "https://api.github.com/users/dsnsabari/repos",
"events_url": "https://api.github.com/users/dsnsabari/events{/privacy}",
"received_events_url": "https://api.github.com/users/dsnsabari/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39326/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39326/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39325 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39325/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39325/comments | https://api.github.com/repos/huggingface/transformers/issues/39325/events | https://github.com/huggingface/transformers/pull/39325 | 3,217,743,737 | PR_kwDOCUB6oc6eOJlV | 39,325 | Avoid registering pytree when using FSDP | {
"login": "kaixuanliu",
"id": 13268042,
"node_id": "MDQ6VXNlcjEzMjY4MDQy",
"avatar_url": "https://avatars.githubusercontent.com/u/13268042?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kaixuanliu",
"html_url": "https://github.com/kaixuanliu",
"followers_url": "https://api.github.com/users/kaixuanliu/followers",
"following_url": "https://api.github.com/users/kaixuanliu/following{/other_user}",
"gists_url": "https://api.github.com/users/kaixuanliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kaixuanliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kaixuanliu/subscriptions",
"organizations_url": "https://api.github.com/users/kaixuanliu/orgs",
"repos_url": "https://api.github.com/users/kaixuanliu/repos",
"events_url": "https://api.github.com/users/kaixuanliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/kaixuanliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T03:24:11 | 2025-07-15T07:15:19 | 2025-07-15T07:15:18 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39325",
"html_url": "https://github.com/huggingface/transformers/pull/39325",
"diff_url": "https://github.com/huggingface/transformers/pull/39325.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39325.patch",
"merged_at": null
} | When using FSDP, this `register_pytree_node` operation will cost lots of extra memory. We found after this PR: https://github.com/huggingface/transformers/pull/35873, we cannot finetune 70b model using FSDP due to OOM issue. | {
"login": "kaixuanliu",
"id": 13268042,
"node_id": "MDQ6VXNlcjEzMjY4MDQy",
"avatar_url": "https://avatars.githubusercontent.com/u/13268042?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kaixuanliu",
"html_url": "https://github.com/kaixuanliu",
"followers_url": "https://api.github.com/users/kaixuanliu/followers",
"following_url": "https://api.github.com/users/kaixuanliu/following{/other_user}",
"gists_url": "https://api.github.com/users/kaixuanliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kaixuanliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kaixuanliu/subscriptions",
"organizations_url": "https://api.github.com/users/kaixuanliu/orgs",
"repos_url": "https://api.github.com/users/kaixuanliu/repos",
"events_url": "https://api.github.com/users/kaixuanliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/kaixuanliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39325/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39325/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39324 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39324/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39324/comments | https://api.github.com/repos/huggingface/transformers/issues/39324/events | https://github.com/huggingface/transformers/pull/39324 | 3,217,545,538 | PR_kwDOCUB6oc6eNf8e | 39,324 | [Bugfix] [Quantization] Remove unused init arg | {
"login": "kylesayrs",
"id": 17103692,
"node_id": "MDQ6VXNlcjE3MTAzNjky",
"avatar_url": "https://avatars.githubusercontent.com/u/17103692?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kylesayrs",
"html_url": "https://github.com/kylesayrs",
"followers_url": "https://api.github.com/users/kylesayrs/followers",
"following_url": "https://api.github.com/users/kylesayrs/following{/other_user}",
"gists_url": "https://api.github.com/users/kylesayrs/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kylesayrs/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kylesayrs/subscriptions",
"organizations_url": "https://api.github.com/users/kylesayrs/orgs",
"repos_url": "https://api.github.com/users/kylesayrs/repos",
"events_url": "https://api.github.com/users/kylesayrs/events{/privacy}",
"received_events_url": "https://api.github.com/users/kylesayrs/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T01:06:19 | 2025-07-16T12:57:42 | 2025-07-16T12:57:42 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39324",
"html_url": "https://github.com/huggingface/transformers/pull/39324",
"diff_url": "https://github.com/huggingface/transformers/pull/39324.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39324.patch",
"merged_at": "2025-07-16T12:57:42"
} | ## Purpose ##
* Remove argument which is not used by the CompressedTensors QuantizationConfig
* https://github.com/neuralmagic/compressed-tensors/blob/main/src/compressed_tensors/quantization/quant_config.py#L138-L144
## Changes ##
* Remove `run_compressed` from list of QuantizationConfig init kwargs | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39324/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39324/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39323 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39323/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39323/comments | https://api.github.com/repos/huggingface/transformers/issues/39323/events | https://github.com/huggingface/transformers/pull/39323 | 3,217,510,610 | PR_kwDOCUB6oc6eNYu9 | 39,323 | Update Readme to Run Multiple Choice Script from Example Directory | {
"login": "eromomon",
"id": 110577233,
"node_id": "U_kgDOBpdGUQ",
"avatar_url": "https://avatars.githubusercontent.com/u/110577233?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/eromomon",
"html_url": "https://github.com/eromomon",
"followers_url": "https://api.github.com/users/eromomon/followers",
"following_url": "https://api.github.com/users/eromomon/following{/other_user}",
"gists_url": "https://api.github.com/users/eromomon/gists{/gist_id}",
"starred_url": "https://api.github.com/users/eromomon/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/eromomon/subscriptions",
"organizations_url": "https://api.github.com/users/eromomon/orgs",
"repos_url": "https://api.github.com/users/eromomon/repos",
"events_url": "https://api.github.com/users/eromomon/events{/privacy}",
"received_events_url": "https://api.github.com/users/eromomon/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-10T00:40:16 | 2025-07-11T17:58:26 | 2025-07-11T17:58:26 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39323",
"html_url": "https://github.com/huggingface/transformers/pull/39323",
"diff_url": "https://github.com/huggingface/transformers/pull/39323.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39323.patch",
"merged_at": "2025-07-11T17:58:26"
} | # What does this PR do?
In the Multiple Choice example, the Readme file includes the instruction
```bash
python examples/pytorch/multiple-choice/run_swag.py \
--model_name_or_path FacebookAI/roberta-base \
--do_train \
--do_eval \
--learning_rate 5e-5 \
--num_train_epochs 3 \
--output_dir /tmp/swag_base \
--per_device_eval_batch_size=16 \
--per_device_train_batch_size=16 \
--overwrite_output
```
which may cause an error when attempting to find the file if the user is currently in the example directory. Assuming the user is in the example directory, this PR suggests removing the complete path and running the Python script directly from that location. Other instructions in this Readme, as well as in other examples, run the script from their respective directories without using the complete path ( examples/DIR_EXAMPLE/SCRIPT.py ).
## Before submitting
- [X] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [X] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@stevhliu hi, I believe you cloud revie this PR. | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39323/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39323/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39322 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39322/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39322/comments | https://api.github.com/repos/huggingface/transformers/issues/39322/events | https://github.com/huggingface/transformers/pull/39322 | 3,217,449,819 | PR_kwDOCUB6oc6eNMEM | 39,322 | Remove do_reduce_labels Argument from model initialization in run_semantic_segmentation_no_trainer | {
"login": "eromomon",
"id": 110577233,
"node_id": "U_kgDOBpdGUQ",
"avatar_url": "https://avatars.githubusercontent.com/u/110577233?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/eromomon",
"html_url": "https://github.com/eromomon",
"followers_url": "https://api.github.com/users/eromomon/followers",
"following_url": "https://api.github.com/users/eromomon/following{/other_user}",
"gists_url": "https://api.github.com/users/eromomon/gists{/gist_id}",
"starred_url": "https://api.github.com/users/eromomon/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/eromomon/subscriptions",
"organizations_url": "https://api.github.com/users/eromomon/orgs",
"repos_url": "https://api.github.com/users/eromomon/repos",
"events_url": "https://api.github.com/users/eromomon/events{/privacy}",
"received_events_url": "https://api.github.com/users/eromomon/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T23:56:56 | 2025-07-14T10:16:49 | 2025-07-14T10:16:49 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39322",
"html_url": "https://github.com/huggingface/transformers/pull/39322",
"diff_url": "https://github.com/huggingface/transformers/pull/39322.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39322.patch",
"merged_at": "2025-07-14T10:16:49"
} | # What does this PR do?
When running ```run_semantic_segmentation_no_trainer.py``` from ```example/pytorch/semantic-segmentation``` using the command provided in the Readme file:
```bash
accelerate launch run_semantic_segmentation_no_trainer.py --output_dir segformer-finetuned-sidewalk --with_tracking --push_to_hub
```
An error is returned:
```bash
TypeError: SegformerForSemanticSegmentation.__init__() got an unexpected keyword argument 'do_reduce_labels'
```
In the Python script, the model is instantiated with the class ```AutoModelForSemanticSegmentation```, which does not have listed ```do_reduce_labels``` as a parameter. This causes the test to fail.
Proposed Change:
This PR proposes removing line 333 from ```run_semantic_segmentation_no_trainer.py``` to resolve the issue.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [X] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@amyeroberts, @qubvel hi, I believe you could review this PR, as it pertains to the computer vision model field.
| {
"login": "qubvel",
"id": 31920396,
"node_id": "MDQ6VXNlcjMxOTIwMzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/31920396?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qubvel",
"html_url": "https://github.com/qubvel",
"followers_url": "https://api.github.com/users/qubvel/followers",
"following_url": "https://api.github.com/users/qubvel/following{/other_user}",
"gists_url": "https://api.github.com/users/qubvel/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qubvel/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qubvel/subscriptions",
"organizations_url": "https://api.github.com/users/qubvel/orgs",
"repos_url": "https://api.github.com/users/qubvel/repos",
"events_url": "https://api.github.com/users/qubvel/events{/privacy}",
"received_events_url": "https://api.github.com/users/qubvel/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39322/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39322/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39321 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39321/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39321/comments | https://api.github.com/repos/huggingface/transformers/issues/39321/events | https://github.com/huggingface/transformers/pull/39321 | 3,217,338,155 | PR_kwDOCUB6oc6eM0AI | 39,321 | make the loss context manager easier to extend | {
"login": "winglian",
"id": 381258,
"node_id": "MDQ6VXNlcjM4MTI1OA==",
"avatar_url": "https://avatars.githubusercontent.com/u/381258?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/winglian",
"html_url": "https://github.com/winglian",
"followers_url": "https://api.github.com/users/winglian/followers",
"following_url": "https://api.github.com/users/winglian/following{/other_user}",
"gists_url": "https://api.github.com/users/winglian/gists{/gist_id}",
"starred_url": "https://api.github.com/users/winglian/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/winglian/subscriptions",
"organizations_url": "https://api.github.com/users/winglian/orgs",
"repos_url": "https://api.github.com/users/winglian/repos",
"events_url": "https://api.github.com/users/winglian/events{/privacy}",
"received_events_url": "https://api.github.com/users/winglian/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T22:49:10 | 2025-07-16T13:47:25 | 2025-07-16T13:47:25 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39321",
"html_url": "https://github.com/huggingface/transformers/pull/39321",
"diff_url": "https://github.com/huggingface/transformers/pull/39321.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39321.patch",
"merged_at": "2025-07-16T13:47:25"
} | # What does this PR do?
Using ExitStack makes it easier to extend the functionality of the context manager wrapping the `model(inputs)` call. This can make it easier to add other things like logging contexts, checkpointing contexts, etc down the road.
e.g.
```python
class CustomTrainer(Trainer):
def compute_loss_context_manager(self):
ctx_stack = super(). compute_loss_context_manager()
my_custom_ctx = CustomContext(...)
ctx_stack.enter_context(my_custom_ctx)
return ctx_stack
````
Fixes # (issue)
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
@SunMarc @ArthurZucker
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "SunMarc",
"id": 57196510,
"node_id": "MDQ6VXNlcjU3MTk2NTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/57196510?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SunMarc",
"html_url": "https://github.com/SunMarc",
"followers_url": "https://api.github.com/users/SunMarc/followers",
"following_url": "https://api.github.com/users/SunMarc/following{/other_user}",
"gists_url": "https://api.github.com/users/SunMarc/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SunMarc/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SunMarc/subscriptions",
"organizations_url": "https://api.github.com/users/SunMarc/orgs",
"repos_url": "https://api.github.com/users/SunMarc/repos",
"events_url": "https://api.github.com/users/SunMarc/events{/privacy}",
"received_events_url": "https://api.github.com/users/SunMarc/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39321/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39321/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39320 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39320/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39320/comments | https://api.github.com/repos/huggingface/transformers/issues/39320/events | https://github.com/huggingface/transformers/pull/39320 | 3,217,204,278 | PR_kwDOCUB6oc6eMW1T | 39,320 | Add dates to the model docs | {
"login": "MHRDYN7",
"id": 113298714,
"node_id": "U_kgDOBsDNGg",
"avatar_url": "https://avatars.githubusercontent.com/u/113298714?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MHRDYN7",
"html_url": "https://github.com/MHRDYN7",
"followers_url": "https://api.github.com/users/MHRDYN7/followers",
"following_url": "https://api.github.com/users/MHRDYN7/following{/other_user}",
"gists_url": "https://api.github.com/users/MHRDYN7/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MHRDYN7/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MHRDYN7/subscriptions",
"organizations_url": "https://api.github.com/users/MHRDYN7/orgs",
"repos_url": "https://api.github.com/users/MHRDYN7/repos",
"events_url": "https://api.github.com/users/MHRDYN7/events{/privacy}",
"received_events_url": "https://api.github.com/users/MHRDYN7/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T21:28:33 | 2025-08-14T17:08:47 | 2025-08-14T17:08:46 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39320",
"html_url": "https://github.com/huggingface/transformers/pull/39320",
"diff_url": "https://github.com/huggingface/transformers/pull/39320.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39320.patch",
"merged_at": "2025-08-14T17:08:46"
} | Implementation of the feature requested in #39319 (please look at the issue for more info).
The file add_paper_dates_on_docs.py inside utils contains the main code.
Current trivial strategy:
Collect the paper id from the first link to hf papers -> use arxiv api to get the dates -> write the dates info below the last html '</div>' tag.
```python
# for the paper publication and last update dates
paper_ids = re.findall(r"https://huggingface\.co/papers/(\d+\.\d+)", content)
if paper_ids:
paper_id = paper_ids[0]
try:
search = Search(id_list=[paper_id])
results = client.results(search)
result = next(results)
published_date = result.published.date()
updated_date = result.updated.date()
```
```python
# for HF integration date
first_commit_date = subprocess.check_output(["git", "log", "--reverse", "--pretty=format:%ad", "--date=iso", path],text=True).strip().split('\n')[0][:10]
```
Most models that have an associated paper have got the html tags for the torch symbol on top, but some new model_docs don't seem to have that properly and those have been missed with this logic. Although it is highly likely that the first link to a paper is the main paper, I did find a good deal of suspicious dates on a random check, so there should be a much more robust strategy. | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39320/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39320/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39319 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39319/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39319/comments | https://api.github.com/repos/huggingface/transformers/issues/39319/events | https://github.com/huggingface/transformers/issues/39319 | 3,217,108,869 | I_kwDOCUB6oc6_wS-F | 39,319 | Add HF integration dates + paper release dates to the model docs | {
"login": "MHRDYN7",
"id": 113298714,
"node_id": "U_kgDOBsDNGg",
"avatar_url": "https://avatars.githubusercontent.com/u/113298714?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MHRDYN7",
"html_url": "https://github.com/MHRDYN7",
"followers_url": "https://api.github.com/users/MHRDYN7/followers",
"following_url": "https://api.github.com/users/MHRDYN7/following{/other_user}",
"gists_url": "https://api.github.com/users/MHRDYN7/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MHRDYN7/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MHRDYN7/subscriptions",
"organizations_url": "https://api.github.com/users/MHRDYN7/orgs",
"repos_url": "https://api.github.com/users/MHRDYN7/repos",
"events_url": "https://api.github.com/users/MHRDYN7/events{/privacy}",
"received_events_url": "https://api.github.com/users/MHRDYN7/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 2648621985,
"node_id": "MDU6TGFiZWwyNjQ4NjIxOTg1",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Feature%20request",
"name": "Feature request",
"color": "FBCA04",
"default": false,
"description": "Request for a new feature"
}
] | closed | false | null | [] | null | [] | 2025-07-09T20:52:43 | 2025-08-14T17:16:44 | 2025-08-14T17:16:44 | CONTRIBUTOR | null | null | null | null | ### Feature request
The model docs should have the dates of publication and last update of the paper mentioned. There should also be the date when this model was add to the transformers library.
### Motivation
Since there are so many models in transformers, it is hard to realize which models are relevant and more up-to-date. Having the dates like this
<img width="1310" height="373" alt="Image" src="https://github.com/user-attachments/assets/543cc882-5cf1-4546-a692-4f6a9ad3e3c7" />
will make it a lot easier for the people using these docs to explore model architectures and papers. It allows people to quickly see the order in which comparable papers were published and get better idea of the evolution of model architectures.
### Your contribution
I want to help implement the pipeline that will correctly assign the correct dates related to the papers using the arxiv api. This is relatively simple if there exists a paper associated to the model. If the model's org publishes a blogpost or something then it will be a bit more complex. @stevhliu what are your thoughts? | {
"login": "MHRDYN7",
"id": 113298714,
"node_id": "U_kgDOBsDNGg",
"avatar_url": "https://avatars.githubusercontent.com/u/113298714?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MHRDYN7",
"html_url": "https://github.com/MHRDYN7",
"followers_url": "https://api.github.com/users/MHRDYN7/followers",
"following_url": "https://api.github.com/users/MHRDYN7/following{/other_user}",
"gists_url": "https://api.github.com/users/MHRDYN7/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MHRDYN7/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MHRDYN7/subscriptions",
"organizations_url": "https://api.github.com/users/MHRDYN7/orgs",
"repos_url": "https://api.github.com/users/MHRDYN7/repos",
"events_url": "https://api.github.com/users/MHRDYN7/events{/privacy}",
"received_events_url": "https://api.github.com/users/MHRDYN7/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39319/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39319/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39318 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39318/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39318/comments | https://api.github.com/repos/huggingface/transformers/issues/39318/events | https://github.com/huggingface/transformers/issues/39318 | 3,216,821,990 | I_kwDOCUB6oc6_vM7m | 39,318 | Whisper demo code for model + processor API is broken | {
"login": "piotyrus",
"id": 472646,
"node_id": "MDQ6VXNlcjQ3MjY0Ng==",
"avatar_url": "https://avatars.githubusercontent.com/u/472646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/piotyrus",
"html_url": "https://github.com/piotyrus",
"followers_url": "https://api.github.com/users/piotyrus/followers",
"following_url": "https://api.github.com/users/piotyrus/following{/other_user}",
"gists_url": "https://api.github.com/users/piotyrus/gists{/gist_id}",
"starred_url": "https://api.github.com/users/piotyrus/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/piotyrus/subscriptions",
"organizations_url": "https://api.github.com/users/piotyrus/orgs",
"repos_url": "https://api.github.com/users/piotyrus/repos",
"events_url": "https://api.github.com/users/piotyrus/events{/privacy}",
"received_events_url": "https://api.github.com/users/piotyrus/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T18:45:37 | 2025-08-18T08:03:16 | 2025-08-18T08:03:16 | NONE | null | null | null | null | ### System Info
When running a copy-paste code of '4.48.3', log mel spectogram values are not padded to the length required by the model. This is the exact error message:
```
ValueError: Whisper expects the mel input features to be of length 3000, but found 585. Make sure to pad the input mel features to 3000.
```
It appears that the demo code is missing a logic to pad the log mel spectogram values to the multiple of 3000.
### Who can help?
_No response_
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [x] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Run the code on CPU as in https://huggingface.co/openai/whisper-large-v3:
```
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor
from datasets import Audio, load_dataset
device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
model_id = "openai/whisper-large-v3"
model = AutoModelForSpeechSeq2Seq.from_pretrained(
model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True
)
model.to(device)
processor = AutoProcessor.from_pretrained(model_id)
dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
dataset = dataset.cast_column("audio", Audio(processor.feature_extractor.sampling_rate))
sample = dataset[0]["audio"]
inputs = processor(
sample["array"],
sampling_rate=sample["sampling_rate"],
return_tensors="pt",
truncation=False,
padding="longest",
return_attention_mask=True,
)
inputs = inputs.to(device, dtype=torch_dtype)
gen_kwargs = {
"max_new_tokens": 448,
"num_beams": 1,
"condition_on_prev_tokens": False,
"compression_ratio_threshold": 1.35, # zlib compression ratio threshold (in token space)
"temperature": (0.0, 0.2, 0.4, 0.6, 0.8, 1.0),
"logprob_threshold": -1.0,
"no_speech_threshold": 0.6,
"return_timestamps": True,
}
pred_ids = model.generate(**inputs, **gen_kwargs)
pred_text = processor.batch_decode(pred_ids, skip_special_tokens=True, decode_with_timestamps=False)
print(pred_text)
```
### Expected behavior
correct transcription of an audio sample | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39318/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39318/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39317 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39317/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39317/comments | https://api.github.com/repos/huggingface/transformers/issues/39317/events | https://github.com/huggingface/transformers/pull/39317 | 3,216,799,353 | PR_kwDOCUB6oc6eK9xA | 39,317 | Fix typo: langauge -> language | {
"login": "tomaarsen",
"id": 37621491,
"node_id": "MDQ6VXNlcjM3NjIxNDkx",
"avatar_url": "https://avatars.githubusercontent.com/u/37621491?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/tomaarsen",
"html_url": "https://github.com/tomaarsen",
"followers_url": "https://api.github.com/users/tomaarsen/followers",
"following_url": "https://api.github.com/users/tomaarsen/following{/other_user}",
"gists_url": "https://api.github.com/users/tomaarsen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/tomaarsen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/tomaarsen/subscriptions",
"organizations_url": "https://api.github.com/users/tomaarsen/orgs",
"repos_url": "https://api.github.com/users/tomaarsen/repos",
"events_url": "https://api.github.com/users/tomaarsen/events{/privacy}",
"received_events_url": "https://api.github.com/users/tomaarsen/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T18:34:59 | 2025-07-09T19:06:48 | 2025-07-09T19:06:46 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39317",
"html_url": "https://github.com/huggingface/transformers/pull/39317",
"diff_url": "https://github.com/huggingface/transformers/pull/39317.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39317.patch",
"merged_at": "2025-07-09T19:06:46"
} | Hello!
## Pull Request overview
* Fix typo: langauge -> language
## Details
I noticed this typo on the [T5Gemma docs](https://huggingface.co/docs/transformers/model_doc/t5gemma), should speak for itself π€
Documentation: @stevhliu
P.s. these don't seem to be working correctly: `<hfoptions id="usage"> <hfoption id="Pipeline">`
- Tom Aarsen | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39317/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39317/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39316 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39316/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39316/comments | https://api.github.com/repos/huggingface/transformers/issues/39316/events | https://github.com/huggingface/transformers/pull/39316 | 3,216,657,370 | PR_kwDOCUB6oc6eKew3 | 39,316 | skip files in `src/` for doctest (for now) | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T17:32:11 | 2025-07-09T17:45:10 | 2025-07-09T17:36:48 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39316",
"html_url": "https://github.com/huggingface/transformers/pull/39316",
"diff_url": "https://github.com/huggingface/transformers/pull/39316.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39316.patch",
"merged_at": "2025-07-09T17:36:48"
} | # What does this PR do?
Fix #39159
Merge directly as
https://github.com/huggingface/transformers/issues/39159#issuecomment-3045160413 | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39316/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39316/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39315 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39315/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39315/comments | https://api.github.com/repos/huggingface/transformers/issues/39315/events | https://github.com/huggingface/transformers/pull/39315 | 3,216,589,426 | PR_kwDOCUB6oc6eKP1w | 39,315 | add `stevhliu` to the list in `self-comment-ci.yml` | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T17:05:09 | 2025-07-09T17:18:26 | 2025-07-09T17:07:44 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39315",
"html_url": "https://github.com/huggingface/transformers/pull/39315",
"diff_url": "https://github.com/huggingface/transformers/pull/39315.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39315.patch",
"merged_at": "2025-07-09T17:07:44"
} | # What does this PR do?
welcome @stevhliu | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39315/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39315/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39314 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39314/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39314/comments | https://api.github.com/repos/huggingface/transformers/issues/39314/events | https://github.com/huggingface/transformers/pull/39314 | 3,216,501,959 | PR_kwDOCUB6oc6eJ9Aw | 39,314 | Fix consistency and a few docstrings warnings | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T16:25:19 | 2025-07-09T16:48:03 | 2025-07-09T16:40:37 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39314",
"html_url": "https://github.com/huggingface/transformers/pull/39314",
"diff_url": "https://github.com/huggingface/transformers/pull/39314.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39314.patch",
"merged_at": "2025-07-09T16:40:37"
} | # What does this PR do?
Race condition on modular conversion for Deepseek v2 + a few docstrings at the same time. | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39314/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39314/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39313 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39313/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39313/comments | https://api.github.com/repos/huggingface/transformers/issues/39313/events | https://github.com/huggingface/transformers/pull/39313 | 3,216,414,034 | PR_kwDOCUB6oc6eJplI | 39,313 | Fix DAC integration tests and checkpoint conversion. | {
"login": "ebezzam",
"id": 4757445,
"node_id": "MDQ6VXNlcjQ3NTc0NDU=",
"avatar_url": "https://avatars.githubusercontent.com/u/4757445?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ebezzam",
"html_url": "https://github.com/ebezzam",
"followers_url": "https://api.github.com/users/ebezzam/followers",
"following_url": "https://api.github.com/users/ebezzam/following{/other_user}",
"gists_url": "https://api.github.com/users/ebezzam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ebezzam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ebezzam/subscriptions",
"organizations_url": "https://api.github.com/users/ebezzam/orgs",
"repos_url": "https://api.github.com/users/ebezzam/repos",
"events_url": "https://api.github.com/users/ebezzam/events{/privacy}",
"received_events_url": "https://api.github.com/users/ebezzam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 6470596964,
"node_id": "LA_kwDOCUB6oc8AAAABga15ZA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Audio",
"name": "Audio",
"color": "760453",
"default": false,
"description": ""
}
] | closed | false | null | [] | null | [] | 2025-07-09T15:55:04 | 2025-07-23T17:21:32 | 2025-07-23T17:21:26 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39313",
"html_url": "https://github.com/huggingface/transformers/pull/39313",
"diff_url": "https://github.com/huggingface/transformers/pull/39313.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39313.patch",
"merged_at": "2025-07-23T17:21:26"
} | # What does this PR do?
Multiple things were wrong with the tests:
- The expected outputs. I created this gist to reproduce **new** expected outputs (as not possible to reproduce previous ones): https://gist.github.com/ebezzam/bb315efa7a416db6336a6b2a2d424ffa
- Hop length was incorrectly set on the Hub for [16kHz](https://huggingface.co/descript/dac_16khz/blob/main/preprocessor_config.json#L4) and [24kHz](https://huggingface.co/descript/dac_24khz/blob/main/preprocessor_config.json#L4) (UPDATE: corrected from 512 to 320 thanks to merged PR by Descript team). Iβve corrected in the conversion script for future use. Below are the test outputs when the Hop length is incorrect (3/6 tests fail):
```bash
# RUN_SLOW=1 pytest tests/models/dac/test_modeling_dac.py::DacIntegrationTest
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_16khz FAILED [ 16%]
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_24khz PASSED [ 33%]
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_44khz PASSED [ 50%]
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_batch_16khz FAILED [ 66%]
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_batch_24khz FAILED [ 83%]
tests/models/dac/test_modeling_dac.py::DacIntegrationTest::test_integration_batch_44khz PASSED [100%]
```
Also Iβve standardized the tests (24kHz was testing something else) and added tests on quantizer and decoder outputs.
# Note on high tolerances for encoder and decoder
Previous (and still now) the tests for the encoder outputs have a high tolerance (1e-3). With [this script](https://gist.github.com/ebezzam/bb315efa7a416db6336a6b2a2d424ffa#file-dac_layer_by_layer_debugging-py), I've verified that the weights have been mapped correctly (output snippet below).
```bash=== WEIGHT DIFFERENCES BY LAYER ===
Conv1 weight max diff: 5.96e-08
Block 0 weight differences:
Block conv weight diff: 1.49e-08
res_unit1.conv1 diff: 2.24e-08
res_unit1.conv2 diff: 2.98e-08
res_unit2.conv1 diff: 1.49e-08
res_unit2.conv2 diff: 2.98e-08
res_unit3.conv1 diff: 1.49e-08
res_unit3.conv2 diff: 2.98e-08
Block 1 weight differences:
Block conv weight diff: 1.49e-08
res_unit1.conv1 diff: 1.49e-08
res_unit1.conv2 diff: 2.98e-08
res_unit2.conv1 diff: 2.24e-08
res_unit2.conv2 diff: 5.96e-08
res_unit3.conv1 diff: 1.49e-08
res_unit3.conv2 diff: 2.98e-08
Block 2 weight differences:
Block conv weight diff: 1.49e-08
res_unit1.conv1 diff: 2.98e-08
res_unit1.conv2 diff: 2.98e-08
res_unit2.conv1 diff: 2.98e-08
res_unit2.conv2 diff: 5.96e-08
res_unit3.conv1 diff: 1.49e-08
res_unit3.conv2 diff: 2.98e-08
Block 3 weight differences:
Block conv weight diff: 2.24e-08
res_unit1.conv1 diff: 2.24e-08
res_unit1.conv2 diff: 2.98e-08
res_unit2.conv1 diff: 2.24e-08
res_unit2.conv2 diff: 2.98e-08
res_unit3.conv1 diff: 4.47e-08
res_unit3.conv2 diff: 2.98e-08
Snake1 alpha diff: 0.00e+00
Conv2 weight diff: 1.49e-08
```
However, error exponentially increases through encoder and decoder layers.
From my understanding, it is because the Transformers version of DAC does NOT have weight normalization in its architecture, while the Original version does (**see [model addition PR](https://github.com/huggingface/transformers/pull/31494#discussion_r1661296778) for discussion as to why there is no weight normalization in the Transformers version**). This causes small differences between expected outputs at each layer, which get larger and larger tensors go deeper in the network.
Below is output snippet of error propagation through the encoder for the 44.1kHz model, calculated with the [same script](https://gist.github.com/ebezzam/bb315efa7a416db6336a6b2a2d424ffa#file-dac_layer_by_layer_debugging-py).
```bash
=== ENCODER ERROR PROPAGATION ANALYSIS ===
Layer Max Error Mean Error Error Growth
----------------------------------------------------------------------
Input 0.00e+00 0.00e+00 1.0x
Conv1 1.19e-07 1.32e-09 infx
Block0 2.26e-04 3.54e-06 1897.0x
--- Block 0 Internal Analysis ---
res_unit1: 1.45e-05 (121.9x)
res_unit2: 5.70e-05 (478.0x)
res_unit3: 2.00e-04 (1680.0x)
Final layers: 5.57e-04 (4670.2x)
Block1 5.99e-03 1.17e-04 26.5x
--- Block 1 Internal Analysis ---
res_unit1: 1.43e-03 (6.3x)
res_unit2: 1.65e-03 (7.3x)
res_unit3: 6.91e-03 (30.6x)
Final layers: 1.17e-02 (51.6x)
Block2 1.61e-02 1.82e-04 2.7x
--- Block 2 Internal Analysis ---
res_unit1: 1.99e-02 (3.3x)
res_unit2: 3.75e-02 (6.3x)
res_unit3: 5.91e-02 (9.9x)
Final layers: 3.76e-02 (6.3x)
Block3 3.64e-02 8.27e-04 2.3x
--- Block 3 Internal Analysis ---
res_unit1: 1.61e-02 (1.0x)
res_unit2: 3.24e-02 (2.0x)
res_unit3: 8.89e-02 (5.5x)
Final layers: 2.32e-01 (14.4x)
Snake1 7.28e-02 7.92e-04 2.0x
Conv2 2.75e-02 9.82e-04 0.4x
=== ERROR PROPAGATION SUMMARY ===
Initial weight error: 1.19e-07
Final encoder error: 2.75e-02
Total error amplification: 230790x
Top 3 error amplifiers:
1. Block0: 1897.0x amplification
2. Block1: 26.5x amplification
3. Block2: 2.7x amplification
```
- `Conv1` already has weight normalization in [original model](https://github.com/descriptinc/descript-audio-codec/blob/c7cfc5d2647e26471dc394f95846a0830e7bec34/dac/model/dac.py#L73), and we see a minimal error (precision-limited).
- `Block0` has 7x layers with weight norm (see [original model](https://github.com/descriptinc/descript-audio-codec/blob/c7cfc5d2647e26471dc394f95846a0830e7bec34/dac/model/dac.py#L46)), and that's where we get the big jump in deviation with the Transformer model -- 1897x.
We also have to keep [decoder test tolerances](https://github.com/ebezzam/transformers/blob/60004f561587f699960b4152ba2be9152c91b60a/tests/models/dac/test_modeling_dac.py#L476) quite high for the same reasons, propagation of weight normalization error.
## Fortunately we can still use Transformers version as a valid approximation
Because:
- quantizer is not affected by this precision error thanks to discretization, allowing us to keep tolerances [in the tests](https://github.com/ebezzam/transformers/blob/60004f561587f699960b4152ba2be9152c91b60a/tests/models/dac/test_modeling_dac.py#L465) at `1e-6`
- the error due the codec itself, is very similar between both approaches (see [gist](https://gist.github.com/ebezzam/bb315efa7a416db6336a6b2a2d424ffa#file-dac_integration-py-L175)), allowing us keep tolerances at `1e-6` for the codec error
| {
"login": "ebezzam",
"id": 4757445,
"node_id": "MDQ6VXNlcjQ3NTc0NDU=",
"avatar_url": "https://avatars.githubusercontent.com/u/4757445?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ebezzam",
"html_url": "https://github.com/ebezzam",
"followers_url": "https://api.github.com/users/ebezzam/followers",
"following_url": "https://api.github.com/users/ebezzam/following{/other_user}",
"gists_url": "https://api.github.com/users/ebezzam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ebezzam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ebezzam/subscriptions",
"organizations_url": "https://api.github.com/users/ebezzam/orgs",
"repos_url": "https://api.github.com/users/ebezzam/repos",
"events_url": "https://api.github.com/users/ebezzam/events{/privacy}",
"received_events_url": "https://api.github.com/users/ebezzam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39313/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 2,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39313/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39312 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39312/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39312/comments | https://api.github.com/repos/huggingface/transformers/issues/39312/events | https://github.com/huggingface/transformers/pull/39312 | 3,216,373,535 | PR_kwDOCUB6oc6eJhVC | 39,312 | fix `phi3` tests | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T15:41:53 | 2025-07-10T09:51:57 | 2025-07-10T09:51:55 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39312",
"html_url": "https://github.com/huggingface/transformers/pull/39312",
"diff_url": "https://github.com/huggingface/transformers/pull/39312.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39312.patch",
"merged_at": "2025-07-10T09:51:55"
} | # What does this PR do?
These 2 tests are never passing after (#31436) , as we didn't update the expected outputs after
Rename Phi-3 rope scaling type (#31436) | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39312/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39312/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39311 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39311/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39311/comments | https://api.github.com/repos/huggingface/transformers/issues/39311/events | https://github.com/huggingface/transformers/issues/39311 | 3,216,226,204 | I_kwDOCUB6oc6_s7ec | 39,311 | Inference with model.generate( ) using a quantized model leads to assertion error | {
"login": "Sandipan99",
"id": 9780288,
"node_id": "MDQ6VXNlcjk3ODAyODg=",
"avatar_url": "https://avatars.githubusercontent.com/u/9780288?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Sandipan99",
"html_url": "https://github.com/Sandipan99",
"followers_url": "https://api.github.com/users/Sandipan99/followers",
"following_url": "https://api.github.com/users/Sandipan99/following{/other_user}",
"gists_url": "https://api.github.com/users/Sandipan99/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Sandipan99/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Sandipan99/subscriptions",
"organizations_url": "https://api.github.com/users/Sandipan99/orgs",
"repos_url": "https://api.github.com/users/Sandipan99/repos",
"events_url": "https://api.github.com/users/Sandipan99/events{/privacy}",
"received_events_url": "https://api.github.com/users/Sandipan99/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T14:53:16 | 2025-08-18T08:03:17 | 2025-08-18T08:03:17 | NONE | null | null | null | null | ### System Info
Linux
transformers==4.52.4
bitsandbytes==0.46.1
### Who can help?
_No response_
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
~~~
m = "microsoft/phi-4"
bnb_config = BitsAndBytesConfig(
load_in_4bit=True
)
tokenizer = AutoTokenizer.from_pretrained(m)
model = AutoModelForCausalLM.from_pretrained(m, quantization_config=bnb_config, device_map='auto')
tokenizer.pad_token_id = tokenizer.eos_token_id
inputs = tokenizer.apply_chat_template(prompt, return_tensors="pt").to('cuda')
out = model.generate(inputs, max_new_tokens=50, synced_gpus=True)
~~~
Run with:
~~~
torchrun --nproc-per-node=2 script.py
~~~
Works perfectly fine with a single GPU setup, but produces assertion error when running on multiple GPUs
The error can be traced back to model.generate() function
error: Assertion error, python3.10/site-packages/bitsandbytes/nn/modules.py in fix_4bit_weight_quant_state_from_module
assert module.weight.shape[1] == 1
### Expected behavior
Expect the model to execute generation without error | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39311/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39311/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39310 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39310/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39310/comments | https://api.github.com/repos/huggingface/transformers/issues/39310/events | https://github.com/huggingface/transformers/pull/39310 | 3,216,073,012 | PR_kwDOCUB6oc6eIf_G | 39,310 | Add a default value for `position_ids` in masking_utils | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 8103865784,
"node_id": "LA_kwDOCUB6oc8AAAAB4wctuA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/for%20patch",
"name": "for patch",
"color": "D93F0B",
"default": false,
"description": "Tag issues / labels that should be included in the next patch"
}
] | closed | false | null | [] | null | [] | 2025-07-09T14:02:21 | 2025-07-10T16:53:42 | 2025-07-10T16:53:41 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39310",
"html_url": "https://github.com/huggingface/transformers/pull/39310",
"diff_url": "https://github.com/huggingface/transformers/pull/39310.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39310.patch",
"merged_at": "2025-07-10T16:53:41"
} | # What does this PR do?
As per the title. See related discussion in https://github.com/huggingface/transformers/pull/39194
Also cc @BenjaminBossan, sorry again for the inconvenience on this one!
| {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39310/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39310/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39309 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39309/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39309/comments | https://api.github.com/repos/huggingface/transformers/issues/39309/events | https://github.com/huggingface/transformers/pull/39309 | 3,216,052,122 | PR_kwDOCUB6oc6eIbYF | 39,309 | Fix audio pipeline with torchcodec input | {
"login": "lhoestq",
"id": 42851186,
"node_id": "MDQ6VXNlcjQyODUxMTg2",
"avatar_url": "https://avatars.githubusercontent.com/u/42851186?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lhoestq",
"html_url": "https://github.com/lhoestq",
"followers_url": "https://api.github.com/users/lhoestq/followers",
"following_url": "https://api.github.com/users/lhoestq/following{/other_user}",
"gists_url": "https://api.github.com/users/lhoestq/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lhoestq/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lhoestq/subscriptions",
"organizations_url": "https://api.github.com/users/lhoestq/orgs",
"repos_url": "https://api.github.com/users/lhoestq/repos",
"events_url": "https://api.github.com/users/lhoestq/events{/privacy}",
"received_events_url": "https://api.github.com/users/lhoestq/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-09T13:55:48 | 2025-08-01T07:58:15 | null | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39309",
"html_url": "https://github.com/huggingface/transformers/pull/39309",
"diff_url": "https://github.com/huggingface/transformers/pull/39309.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39309.patch",
"merged_at": null
} | fix `tests/pipelines/test_pipelines_automatic_speech_recognition.py::AutomaticSpeechRecognitionPipelineTests::test_speculative_decoding_whisper_distil`
which has
```
ValueError: When passing a dictionary to AutomaticSpeechRecognitionPipeline, the dict needs to contain a "raw" key containing the numpy array or torch tensor representing the audio and a "sampling_rate" key, containing the sampling_rate associated with that array
``` | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39309/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39309/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39308 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39308/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39308/comments | https://api.github.com/repos/huggingface/transformers/issues/39308/events | https://github.com/huggingface/transformers/pull/39308 | 3,216,002,879 | PR_kwDOCUB6oc6eIQbt | 39,308 | [modular] Allow method with the same name in case of @property decorator | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T13:42:01 | 2025-07-09T13:55:57 | 2025-07-09T13:46:53 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39308",
"html_url": "https://github.com/huggingface/transformers/pull/39308",
"diff_url": "https://github.com/huggingface/transformers/pull/39308.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39308.patch",
"merged_at": "2025-07-09T13:46:53"
} | # What does this PR do?
As per the title. See added examples. Needed for #39301 as well
| {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39308/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39308/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39307 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39307/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39307/comments | https://api.github.com/repos/huggingface/transformers/issues/39307/events | https://github.com/huggingface/transformers/pull/39307 | 3,215,975,011 | PR_kwDOCUB6oc6eIKPf | 39,307 | skip `test_torchscript_*` for now until the majority of the community ask for it | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T13:33:32 | 2025-07-09T13:47:09 | 2025-07-09T13:35:48 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39307",
"html_url": "https://github.com/huggingface/transformers/pull/39307",
"diff_url": "https://github.com/huggingface/transformers/pull/39307.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39307.patch",
"merged_at": "2025-07-09T13:35:48"
} | # What does this PR do?
As discussed with the core maintainer(s) offline.
For the record, the trace log (after #39120) is
```bash
> traced_model = torch.jit.trace(model, (main_input,))
tests/test_modeling_common.py:1453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
/usr/local/lib/python3.10/dist-packages/torch/_dynamo/eval_frame.py:838: in _fn
return fn(*args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/jit/_trace.py:1002: in trace
traced_func = _trace_impl(
/usr/local/lib/python3.10/dist-packages/torch/jit/_trace.py:696: in _trace_impl
return trace_module(
/usr/local/lib/python3.10/dist-packages/torch/jit/_trace.py:1279: in trace_module
module._c._create_method_from_trace(
/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py:1751: in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py:1762: in _call_impl
return forward_call(*args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py:1741: in _slow_forward
result = self.forward(*input, **kwargs)
src/transformers/utils/generic.py:1060: in wrapper
outputs = func(self, *args, **kwargs)
src/transformers/models/llama/modeling_llama.py:396: in forward
causal_mask = create_causal_mask(
src/transformers/masking_utils.py:793: in create_causal_mask
causal_mask = mask_interface(
src/transformers/masking_utils.py:482: in eager_mask
mask = sdpa_mask(
src/transformers/masking_utils.py:366: in sdpa_mask_recent_torch
causal_mask = _vmap_for_bhqkv(mask_function)(batch_arange, head_arange, cache_position, kv_arange)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/apis.py:202: in wrapped
return vmap_impl(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:334: in vmap_impl
return _flat_vmap(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:484: in _flat_vmap
batched_outputs = func(*batched_inputs, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/apis.py:202: in wrapped
return vmap_impl(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:334: in vmap_impl
return _flat_vmap(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:484: in _flat_vmap
batched_outputs = func(*batched_inputs, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/apis.py:202: in wrapped
return vmap_impl(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:334: in vmap_impl
return _flat_vmap(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:484: in _flat_vmap
batched_outputs = func(*batched_inputs, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/apis.py:202: in wrapped
return vmap_impl(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:334: in vmap_impl
return _flat_vmap(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/vmap.py:484: in _flat_vmap
batched_outputs = func(*batched_inputs, **kwargs)
src/transformers/masking_utils.py:49: in and_mask
result = result & mask(batch_idx, head_idx, q_idx, kv_idx)
src/transformers/masking_utils.py:134: in inner_mask
return packed_sequence_mask[batch_idx, q_idx] == packed_sequence_mask[batch_idx, kv_idx]
/usr/local/lib/python3.10/dist-packages/torch/_dynamo/_trace_wrapped_higher_order_op.py:141: in __torch_function__
return mod_index(args[0], index_args)
/usr/local/lib/python3.10/dist-packages/torch/autograd/function.py:585: in apply
return custom_function_call(cls, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:49: in __call__
return super().__call__(autograd_function, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:471: in __call__
return wrapper()
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:467: in wrapper
return self.dispatch(
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:330: in dispatch
return dispatch_functorch(self, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:294: in dispatch_functorch
return interpreter.process(op, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:130: in process
return kernel(self, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:300: in custom_function_call_vmap
return custom_function_call_vmap_generate_rule(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:384: in custom_function_call_vmap_generate_rule
outputs = custom_function_call(vmapped_function, *unwrapped_operands)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:49: in __call__
return super().__call__(autograd_function, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:471: in __call__
return wrapper()
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:467: in wrapper
return self.dispatch(
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:330: in dispatch
return dispatch_functorch(self, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:294: in dispatch_functorch
return interpreter.process(op, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:130: in process
return kernel(self, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:300: in custom_function_call_vmap
return custom_function_call_vmap_generate_rule(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:384: in custom_function_call_vmap_generate_rule
outputs = custom_function_call(vmapped_function, *unwrapped_operands)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:49: in __call__
return super().__call__(autograd_function, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:471: in __call__
return wrapper()
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:467: in wrapper
return self.dispatch(
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:330: in dispatch
return dispatch_functorch(self, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:294: in dispatch_functorch
return interpreter.process(op, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:130: in process
return kernel(self, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:300: in custom_function_call_vmap
return custom_function_call_vmap_generate_rule(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:384: in custom_function_call_vmap_generate_rule
outputs = custom_function_call(vmapped_function, *unwrapped_operands)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:49: in __call__
return super().__call__(autograd_function, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:471: in __call__
return wrapper()
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:467: in wrapper
return self.dispatch(
/usr/local/lib/python3.10/dist-packages/torch/_ops.py:330: in dispatch
return dispatch_functorch(self, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:294: in dispatch_functorch
return interpreter.process(op, args, kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/pyfunctorch.py:130: in process
return kernel(self, *args, **kwargs)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:300: in custom_function_call_vmap
return custom_function_call_vmap_generate_rule(
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:384: in custom_function_call_vmap_generate_rule
outputs = custom_function_call(vmapped_function, *unwrapped_operands)
/usr/local/lib/python3.10/dist-packages/torch/_functorch/autograd_function.py:50: in __call__
return autograd_function.apply(*args, **kwargs)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
cls = <class 'torch._functorch.autograd_function.VmappedVmappedVmappedVmappedModIndex'>
args = (tensor([[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, ... 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12], device='cuda:0'), tensor([0, 1, 2, 3, 4, 5, 6], device='cuda:0')]), kwargs = {}
bind_default_args = <function Function.apply.<locals>.bind_default_args at 0x7fda91408700>, is_setup_ctx_defined = True
@classmethod
def apply(cls, *args, **kwargs):
def bind_default_args(func, *args, **kwargs):
signature = inspect.signature(func)
bound_args = signature.bind(*args, **kwargs)
bound_args.apply_defaults()
return bound_args.args
is_setup_ctx_defined = _is_setup_context_defined(cls.setup_context)
if is_setup_ctx_defined:
args = bind_default_args(cls.forward, *args, **kwargs)
if not torch._C._are_functorch_transforms_active():
# See NOTE: [functorch vjp and autograd interaction]
args = _functorch.utils.unwrap_dead_wrappers(args)
> return super().apply(*args, **kwargs) # type: ignore[misc]
E RuntimeError: _Map_base::at
/usr/local/lib/python3.10/dist-packages/torch/autograd/function.py:575: RuntimeError
During handling of the above exception, another exception occurred:
self = <tests.models.llama.test_modeling_llama.LlamaModelTest testMethod=test_torchscript_simple>
@slow
def test_torchscript_simple(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
> self._create_and_check_torchscript(config, inputs_dict)
tests/test_modeling_common.py:1350:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
tests/test_modeling_common.py:1455: in _create_and_check_torchscript
self.fail("Couldn't trace module.")
E AssertionError: Couldn't trace module.
``` | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39307/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39307/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39306 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39306/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39306/comments | https://api.github.com/repos/huggingface/transformers/issues/39306/events | https://github.com/huggingface/transformers/pull/39306 | 3,215,920,949 | PR_kwDOCUB6oc6eH-cp | 39,306 | Fix batch object detection 31356 | {
"login": "z-pandeyji",
"id": 91720238,
"node_id": "U_kgDOBXeKLg",
"avatar_url": "https://avatars.githubusercontent.com/u/91720238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/z-pandeyji",
"html_url": "https://github.com/z-pandeyji",
"followers_url": "https://api.github.com/users/z-pandeyji/followers",
"following_url": "https://api.github.com/users/z-pandeyji/following{/other_user}",
"gists_url": "https://api.github.com/users/z-pandeyji/gists{/gist_id}",
"starred_url": "https://api.github.com/users/z-pandeyji/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/z-pandeyji/subscriptions",
"organizations_url": "https://api.github.com/users/z-pandeyji/orgs",
"repos_url": "https://api.github.com/users/z-pandeyji/repos",
"events_url": "https://api.github.com/users/z-pandeyji/events{/privacy}",
"received_events_url": "https://api.github.com/users/z-pandeyji/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-09T13:16:29 | 2025-07-10T06:59:18 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39306",
"html_url": "https://github.com/huggingface/transformers/pull/39306",
"diff_url": "https://github.com/huggingface/transformers/pull/39306.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39306.patch",
"merged_at": null
} | @amyeroberts @Narsil
## Fixes #31356
**Issue**: Object Detection Pipeline only outputs first element when batching
## Solution
The Object Detection Pipeline now correctly handles batched inputs, returning results for all images in the batch instead of just the first one.
## Verification
- Single image processing works correctly
- Batch processing now returns results for all images
- Tested with multiple image types (URLs, PIL images)
- All existing tests pass
## Before/After
**Before**: `detector([img1, img2, img3])` β Returns results for only img1
**After**: `detector([img1, img2, img3])` β Returns results for all 3 images
## Technical Details
- **Files modified**: `src/transformers/pipelines/object_detection.py`
- **Performance**: No regression, improved batch efficiency
- **Compatibility**: Fully backward compatible
## Impact
- Fixes critical batching bug affecting users
- Improves user experience for batch processing
- No breaking changes | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39306/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39306/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39305 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39305/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39305/comments | https://api.github.com/repos/huggingface/transformers/issues/39305/events | https://github.com/huggingface/transformers/pull/39305 | 3,215,917,919 | PR_kwDOCUB6oc6eH9yv | 39,305 | Updated Switch Transformers model card with standardized format (Issue #36979) | {
"login": "giuseppeCoccia",
"id": 29398916,
"node_id": "MDQ6VXNlcjI5Mzk4OTE2",
"avatar_url": "https://avatars.githubusercontent.com/u/29398916?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/giuseppeCoccia",
"html_url": "https://github.com/giuseppeCoccia",
"followers_url": "https://api.github.com/users/giuseppeCoccia/followers",
"following_url": "https://api.github.com/users/giuseppeCoccia/following{/other_user}",
"gists_url": "https://api.github.com/users/giuseppeCoccia/gists{/gist_id}",
"starred_url": "https://api.github.com/users/giuseppeCoccia/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/giuseppeCoccia/subscriptions",
"organizations_url": "https://api.github.com/users/giuseppeCoccia/orgs",
"repos_url": "https://api.github.com/users/giuseppeCoccia/repos",
"events_url": "https://api.github.com/users/giuseppeCoccia/events{/privacy}",
"received_events_url": "https://api.github.com/users/giuseppeCoccia/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T13:15:31 | 2025-07-11T16:16:00 | 2025-07-10T22:34:10 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39305",
"html_url": "https://github.com/huggingface/transformers/pull/39305",
"diff_url": "https://github.com/huggingface/transformers/pull/39305.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39305.patch",
"merged_at": "2025-07-10T22:34:10"
} | # What does this PR do?
It updates the model card for Switch Transformer by following a new standardized format
#36979
## Before submitting
- [X] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [X] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [X] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case -> [link](https://github.com/huggingface/transformers/issues/36979#issuecomment-3052596321)
- [X] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
@stevhliu
| {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39305/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39305/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39304 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39304/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39304/comments | https://api.github.com/repos/huggingface/transformers/issues/39304/events | https://github.com/huggingface/transformers/pull/39304 | 3,215,887,574 | PR_kwDOCUB6oc6eH3c1 | 39,304 | update expected outputs for whisper after #38778 | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T13:05:09 | 2025-07-25T16:48:11 | 2025-07-25T16:48:11 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39304",
"html_url": "https://github.com/huggingface/transformers/pull/39304",
"diff_url": "https://github.com/huggingface/transformers/pull/39304.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39304.patch",
"merged_at": "2025-07-25T16:48:11"
} | # What does this PR do?
The differences are actually minor (mostly at the end of the outputs)
| {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39304/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39304/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39303 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39303/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39303/comments | https://api.github.com/repos/huggingface/transformers/issues/39303/events | https://github.com/huggingface/transformers/pull/39303 | 3,215,744,844 | PR_kwDOCUB6oc6eHYPd | 39,303 | Fix critical typos in code example | {
"login": "zabojeb",
"id": 128208855,
"node_id": "U_kgDOB6RP1w",
"avatar_url": "https://avatars.githubusercontent.com/u/128208855?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zabojeb",
"html_url": "https://github.com/zabojeb",
"followers_url": "https://api.github.com/users/zabojeb/followers",
"following_url": "https://api.github.com/users/zabojeb/following{/other_user}",
"gists_url": "https://api.github.com/users/zabojeb/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zabojeb/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zabojeb/subscriptions",
"organizations_url": "https://api.github.com/users/zabojeb/orgs",
"repos_url": "https://api.github.com/users/zabojeb/repos",
"events_url": "https://api.github.com/users/zabojeb/events{/privacy}",
"received_events_url": "https://api.github.com/users/zabojeb/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-09T12:18:15 | 2025-07-09T12:24:52 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39303",
"html_url": "https://github.com/huggingface/transformers/pull/39303",
"diff_url": "https://github.com/huggingface/transformers/pull/39303.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39303.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes typos in code example of GPT2 that was making code not runnable - [link](https://huggingface.co/docs/transformers/model_doc/gpt2)
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39303/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39303/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39302 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39302/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39302/comments | https://api.github.com/repos/huggingface/transformers/issues/39302/events | https://github.com/huggingface/transformers/issues/39302 | 3,215,428,472 | I_kwDOCUB6oc6_p4t4 | 39,302 | Gemma3 slightly alters hidden state when input_ids is batched | {
"login": "lwiklendt",
"id": 5995363,
"node_id": "MDQ6VXNlcjU5OTUzNjM=",
"avatar_url": "https://avatars.githubusercontent.com/u/5995363?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lwiklendt",
"html_url": "https://github.com/lwiklendt",
"followers_url": "https://api.github.com/users/lwiklendt/followers",
"following_url": "https://api.github.com/users/lwiklendt/following{/other_user}",
"gists_url": "https://api.github.com/users/lwiklendt/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lwiklendt/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lwiklendt/subscriptions",
"organizations_url": "https://api.github.com/users/lwiklendt/orgs",
"repos_url": "https://api.github.com/users/lwiklendt/repos",
"events_url": "https://api.github.com/users/lwiklendt/events{/privacy}",
"received_events_url": "https://api.github.com/users/lwiklendt/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T10:31:39 | 2025-07-10T01:31:45 | 2025-07-10T01:31:45 | NONE | null | null | null | null | ### System Info
- `transformers` version: 4.50.3
- Platform: Windows-11-10.0.26100-SP0
- Python version: 3.13.2
- Huggingface_hub version: 0.30.1
- Safetensors version: 0.5.3
- Accelerate version: 1.8.1
- Accelerate config: - compute_environment: LOCAL_MACHINE
- distributed_type: NO
- mixed_precision: bf16
- use_cpu: False
- debug: False
- num_processes: 1
- machine_rank: 0
- num_machines: 1
- gpu_ids: all
- rdzv_backend: static
- same_network: True
- main_training_function: main
- enable_cpu_affinity: False
- downcast_bf16: no
- tpu_use_cluster: False
- tpu_use_sudo: False
- tpu_env: []
- DeepSpeed version: not installed
- PyTorch version (GPU?): 2.6.0+cu118 (True)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: <fill in>
- Using GPU in script?: <fill in>
- GPU type: NVIDIA GeForce RTX 3090
### Who can help?
_No response_
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
```python
import torch
from transformers import AutoTokenizer, Gemma3ForCausalLM
import torch.nn.functional as F
ckpt = 'google/gemma-3-1b-pt'
hf_token = None
tokenizer = AutoTokenizer.from_pretrained(ckpt, token=hf_token)
model = Gemma3ForCausalLM.from_pretrained(
ckpt,
torch_dtype=torch.bfloat16,
device_map='auto',
token=hf_token,
).model
prompt = 'There is a problem where batching alters the hidden state in Gemma3'
token_ids = tokenizer(prompt, return_tensors='pt')['input_ids'].to(model.device)
hidden_state_single = model(token_ids, use_cache=False)['last_hidden_state']
hidden_state_batch = model(torch.concat((token_ids, token_ids)), use_cache=False)['last_hidden_state']
print(hidden_state_single[0])
print(hidden_state_batch[0])
print(torch.allclose(hidden_state_single[0], hidden_state_batch[0]))
```
prints
```
tensor([[ 0.6523, -1.8047, 0.7031, ..., -1.4375, -0.2988, -0.4707],
[ 1.3359, 0.2520, -2.5156, ..., -3.1719, 1.6172, -0.9648],
[ 2.0156, -3.5781, -1.2891, ..., -1.5547, -0.2246, -0.7969],
...,
[ 0.0505, 0.1768, -1.1328, ..., 0.1777, -0.2246, -0.4766],
[ 0.5195, -0.4102, -0.7148, ..., -0.6367, -4.2188, -0.6602],
[-0.3887, -0.7188, -1.5391, ..., -0.3574, -0.6094, -3.0000]],
device='cuda:0', dtype=torch.bfloat16, grad_fn=<SelectBackward0>)
tensor([[ 0.6641, -1.8281, 0.6992, ..., -1.4688, -0.2988, -0.4707],
[ 1.3125, 0.2041, -2.4688, ..., -3.1875, 1.6328, -0.9805],
[ 2.0625, -3.5781, -1.2656, ..., -1.5547, -0.2246, -0.7773],
...,
[ 0.0728, 0.1816, -1.1250, ..., 0.2188, -0.2168, -0.4648],
[ 0.5391, -0.3672, -0.6836, ..., -0.6289, -4.2188, -0.6719],
[-0.3574, -0.7500, -1.5078, ..., -0.3848, -0.6016, -2.9375]],
device='cuda:0', dtype=torch.bfloat16, grad_fn=<SelectBackward0>)
False
```
### Expected behavior
I expect `hidden_state_single[0]` to be equal to `hidden_state_batch[0]`. | {
"login": "lwiklendt",
"id": 5995363,
"node_id": "MDQ6VXNlcjU5OTUzNjM=",
"avatar_url": "https://avatars.githubusercontent.com/u/5995363?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lwiklendt",
"html_url": "https://github.com/lwiklendt",
"followers_url": "https://api.github.com/users/lwiklendt/followers",
"following_url": "https://api.github.com/users/lwiklendt/following{/other_user}",
"gists_url": "https://api.github.com/users/lwiklendt/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lwiklendt/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lwiklendt/subscriptions",
"organizations_url": "https://api.github.com/users/lwiklendt/orgs",
"repos_url": "https://api.github.com/users/lwiklendt/repos",
"events_url": "https://api.github.com/users/lwiklendt/events{/privacy}",
"received_events_url": "https://api.github.com/users/lwiklendt/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39302/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39302/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39301 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39301/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39301/comments | https://api.github.com/repos/huggingface/transformers/issues/39301/events | https://github.com/huggingface/transformers/pull/39301 | 3,215,288,719 | PR_kwDOCUB6oc6eF2GM | 39,301 | [sliding window] revert and deprecate | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 8103865784,
"node_id": "LA_kwDOCUB6oc8AAAAB4wctuA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/for%20patch",
"name": "for patch",
"color": "D93F0B",
"default": false,
"description": "Tag issues / labels that should be included in the next patch"
}
] | closed | false | null | [] | null | [] | 2025-07-09T09:54:05 | 2025-07-09T14:10:38 | 2025-07-09T14:10:38 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39301",
"html_url": "https://github.com/huggingface/transformers/pull/39301",
"diff_url": "https://github.com/huggingface/transformers/pull/39301.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39301.patch",
"merged_at": "2025-07-09T14:10:38"
} | # What does this PR do?
Fixes https://github.com/huggingface/transformers/issues/39290
| {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39301/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39301/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39300 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39300/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39300/comments | https://api.github.com/repos/huggingface/transformers/issues/39300/events | https://github.com/huggingface/transformers/pull/39300 | 3,215,255,726 | PR_kwDOCUB6oc6eFu1Q | 39,300 | fix: providing a tensor to cache_position in model.generate kwargs always crashes because of boolean test | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T09:44:53 | 2025-07-31T08:42:54 | 2025-07-30T17:30:29 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39300",
"html_url": "https://github.com/huggingface/transformers/pull/39300",
"diff_url": "https://github.com/huggingface/transformers/pull/39300.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39300.patch",
"merged_at": "2025-07-30T17:30:29"
} | # What does this PR do?
copy of #39261 -- see the discussion there for more references
(I did a bad force push and bricked the PR π€ ) | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39300/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 1,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39300/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39299 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39299/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39299/comments | https://api.github.com/repos/huggingface/transformers/issues/39299/events | https://github.com/huggingface/transformers/pull/39299 | 3,215,218,992 | PR_kwDOCUB6oc6eFmxQ | 39,299 | Remove device check in HQQ quantizer | {
"login": "learning-chip",
"id": 80731350,
"node_id": "MDQ6VXNlcjgwNzMxMzUw",
"avatar_url": "https://avatars.githubusercontent.com/u/80731350?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/learning-chip",
"html_url": "https://github.com/learning-chip",
"followers_url": "https://api.github.com/users/learning-chip/followers",
"following_url": "https://api.github.com/users/learning-chip/following{/other_user}",
"gists_url": "https://api.github.com/users/learning-chip/gists{/gist_id}",
"starred_url": "https://api.github.com/users/learning-chip/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/learning-chip/subscriptions",
"organizations_url": "https://api.github.com/users/learning-chip/orgs",
"repos_url": "https://api.github.com/users/learning-chip/repos",
"events_url": "https://api.github.com/users/learning-chip/events{/privacy}",
"received_events_url": "https://api.github.com/users/learning-chip/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T09:34:04 | 2025-07-11T15:00:28 | 2025-07-11T14:59:51 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39299",
"html_url": "https://github.com/huggingface/transformers/pull/39299",
"diff_url": "https://github.com/huggingface/transformers/pull/39299.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39299.patch",
"merged_at": "2025-07-11T14:59:51"
} | Fixes https://github.com/huggingface/transformers/issues/38439
Tested on CPU and NPU, both works out-of-box.
In this case, the HQQ dependency is installed by
```
DISABLE_CUDA=1 pip install git+https://github.com/mobiusml/hqq.git
```
## Who can review?
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
| {
"login": "SunMarc",
"id": 57196510,
"node_id": "MDQ6VXNlcjU3MTk2NTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/57196510?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SunMarc",
"html_url": "https://github.com/SunMarc",
"followers_url": "https://api.github.com/users/SunMarc/followers",
"following_url": "https://api.github.com/users/SunMarc/following{/other_user}",
"gists_url": "https://api.github.com/users/SunMarc/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SunMarc/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SunMarc/subscriptions",
"organizations_url": "https://api.github.com/users/SunMarc/orgs",
"repos_url": "https://api.github.com/users/SunMarc/repos",
"events_url": "https://api.github.com/users/SunMarc/events{/privacy}",
"received_events_url": "https://api.github.com/users/SunMarc/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39299/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39299/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39298 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39298/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39298/comments | https://api.github.com/repos/huggingface/transformers/issues/39298/events | https://github.com/huggingface/transformers/pull/39298 | 3,215,026,914 | PR_kwDOCUB6oc6eE9PH | 39,298 | feat: support flash attn 2 in Granite MOE | {
"login": "jannisborn",
"id": 15703818,
"node_id": "MDQ6VXNlcjE1NzAzODE4",
"avatar_url": "https://avatars.githubusercontent.com/u/15703818?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jannisborn",
"html_url": "https://github.com/jannisborn",
"followers_url": "https://api.github.com/users/jannisborn/followers",
"following_url": "https://api.github.com/users/jannisborn/following{/other_user}",
"gists_url": "https://api.github.com/users/jannisborn/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jannisborn/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jannisborn/subscriptions",
"organizations_url": "https://api.github.com/users/jannisborn/orgs",
"repos_url": "https://api.github.com/users/jannisborn/repos",
"events_url": "https://api.github.com/users/jannisborn/events{/privacy}",
"received_events_url": "https://api.github.com/users/jannisborn/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T08:34:05 | 2025-07-09T21:14:53 | 2025-07-09T20:07:00 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39298",
"html_url": "https://github.com/huggingface/transformers/pull/39298",
"diff_url": "https://github.com/huggingface/transformers/pull/39298.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39298.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
While the GraniteMOE model currently supports flash attention 2 (`supports_flash_attn_2 = True`) it expects the user to alter inputs when FA2 is used. This is counterintuitive and should be fixed. For context, FA2 throws an error if an attention mask is provided. The existing logic (see [here](https://github.com/huggingface/transformers/blob/1580f6465347df4e8d7c12cd4d3dc603b8a689ad/src/transformers/models/granitemoe/modeling_granitemoe.py#L778)) does not ensure that no attention mask is passed if FA2 is used.
This PR fixes that.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
@ArthurZucker @younesbelkada @mayank31398 | {
"login": "jannisborn",
"id": 15703818,
"node_id": "MDQ6VXNlcjE1NzAzODE4",
"avatar_url": "https://avatars.githubusercontent.com/u/15703818?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jannisborn",
"html_url": "https://github.com/jannisborn",
"followers_url": "https://api.github.com/users/jannisborn/followers",
"following_url": "https://api.github.com/users/jannisborn/following{/other_user}",
"gists_url": "https://api.github.com/users/jannisborn/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jannisborn/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jannisborn/subscriptions",
"organizations_url": "https://api.github.com/users/jannisborn/orgs",
"repos_url": "https://api.github.com/users/jannisborn/repos",
"events_url": "https://api.github.com/users/jannisborn/events{/privacy}",
"received_events_url": "https://api.github.com/users/jannisborn/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39298/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39298/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39297 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39297/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39297/comments | https://api.github.com/repos/huggingface/transformers/issues/39297/events | https://github.com/huggingface/transformers/pull/39297 | 3,214,939,096 | PR_kwDOCUB6oc6eEqmZ | 39,297 | Fix bug with deepspeed and accelerator args in training_args.py | {
"login": "MuyaoLi-jimo",
"id": 109989534,
"node_id": "U_kgDOBo5Ong",
"avatar_url": "https://avatars.githubusercontent.com/u/109989534?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MuyaoLi-jimo",
"html_url": "https://github.com/MuyaoLi-jimo",
"followers_url": "https://api.github.com/users/MuyaoLi-jimo/followers",
"following_url": "https://api.github.com/users/MuyaoLi-jimo/following{/other_user}",
"gists_url": "https://api.github.com/users/MuyaoLi-jimo/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MuyaoLi-jimo/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MuyaoLi-jimo/subscriptions",
"organizations_url": "https://api.github.com/users/MuyaoLi-jimo/orgs",
"repos_url": "https://api.github.com/users/MuyaoLi-jimo/repos",
"events_url": "https://api.github.com/users/MuyaoLi-jimo/events{/privacy}",
"received_events_url": "https://api.github.com/users/MuyaoLi-jimo/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-09T08:02:01 | 2025-07-11T16:34:42 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39297",
"html_url": "https://github.com/huggingface/transformers/pull/39297",
"diff_url": "https://github.com/huggingface/transformers/pull/39297.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39297.patch",
"merged_at": null
} | ### System Info
- transformers version: 4.54.0.dev0
- Platform: Linux-5.4.119-19.0009.28-x86_64-with-glibc2.31
- Python version: 3.10.16
- Huggingface_hub version: 0.31.1
- Safetensors version: 0.4.5
- Accelerate version: 0.22.0.dev0
PyTorch version (GPU?): 2.6.0+cu124 (True)
Tensorflow version (GPU?): not installed (NA)
Flax version (CPU?/GPU?/TPU?): not installed (NA)
Jax version: not installed
JaxLib version: not installed
Using GPU in script?: T
Using distributed or parallel set-up in script?: T
### Who can help?
@SunMarc and @qgallouedec
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [x] An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Right now any command with --deepspeed /path/to/json will fail and throw the following error
`--deepspeed: invalid Dict value`
This bug has occurred before and was previously fixed in [#24974](https://github.com/huggingface/transformers/pull/24974), but unfortunately, itβs broken again.
The root cause seems to be that Pythonβs dataclass fields do not support Union[str, dict] when parsed from the CLI.
### Expected behavior
deepspeed flag should support string.
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39297/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 2
} | https://api.github.com/repos/huggingface/transformers/issues/39297/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39296 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39296/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39296/comments | https://api.github.com/repos/huggingface/transformers/issues/39296/events | https://github.com/huggingface/transformers/pull/39296 | 3,214,860,035 | PR_kwDOCUB6oc6eEZy4 | 39,296 | no use now | {
"login": "zRzRzRzRzRzRzR",
"id": 93239683,
"node_id": "U_kgDOBY65gw",
"avatar_url": "https://avatars.githubusercontent.com/u/93239683?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zRzRzRzRzRzRzR",
"html_url": "https://github.com/zRzRzRzRzRzRzR",
"followers_url": "https://api.github.com/users/zRzRzRzRzRzRzR/followers",
"following_url": "https://api.github.com/users/zRzRzRzRzRzRzR/following{/other_user}",
"gists_url": "https://api.github.com/users/zRzRzRzRzRzRzR/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zRzRzRzRzRzRzR/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zRzRzRzRzRzRzR/subscriptions",
"organizations_url": "https://api.github.com/users/zRzRzRzRzRzRzR/orgs",
"repos_url": "https://api.github.com/users/zRzRzRzRzRzRzR/repos",
"events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/events{/privacy}",
"received_events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T07:33:52 | 2025-07-18T04:39:44 | 2025-07-14T15:59:04 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39296",
"html_url": "https://github.com/huggingface/transformers/pull/39296",
"diff_url": "https://github.com/huggingface/transformers/pull/39296.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39296.patch",
"merged_at": null
} | not using now.use https://github.com/huggingface/transformers/pull/39393 | {
"login": "zRzRzRzRzRzRzR",
"id": 93239683,
"node_id": "U_kgDOBY65gw",
"avatar_url": "https://avatars.githubusercontent.com/u/93239683?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zRzRzRzRzRzRzR",
"html_url": "https://github.com/zRzRzRzRzRzRzR",
"followers_url": "https://api.github.com/users/zRzRzRzRzRzRzR/followers",
"following_url": "https://api.github.com/users/zRzRzRzRzRzRzR/following{/other_user}",
"gists_url": "https://api.github.com/users/zRzRzRzRzRzRzR/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zRzRzRzRzRzRzR/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zRzRzRzRzRzRzR/subscriptions",
"organizations_url": "https://api.github.com/users/zRzRzRzRzRzRzR/orgs",
"repos_url": "https://api.github.com/users/zRzRzRzRzRzRzR/repos",
"events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/events{/privacy}",
"received_events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39296/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39296/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39295 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39295/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39295/comments | https://api.github.com/repos/huggingface/transformers/issues/39295/events | https://github.com/huggingface/transformers/issues/39295 | 3,214,790,143 | I_kwDOCUB6oc6_nc3_ | 39,295 | `quantization_method` is not cleared after calling `.dequantize()` | {
"login": "karol-szustakowski",
"id": 191375020,
"node_id": "U_kgDOC2gmrA",
"avatar_url": "https://avatars.githubusercontent.com/u/191375020?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/karol-szustakowski",
"html_url": "https://github.com/karol-szustakowski",
"followers_url": "https://api.github.com/users/karol-szustakowski/followers",
"following_url": "https://api.github.com/users/karol-szustakowski/following{/other_user}",
"gists_url": "https://api.github.com/users/karol-szustakowski/gists{/gist_id}",
"starred_url": "https://api.github.com/users/karol-szustakowski/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/karol-szustakowski/subscriptions",
"organizations_url": "https://api.github.com/users/karol-szustakowski/orgs",
"repos_url": "https://api.github.com/users/karol-szustakowski/repos",
"events_url": "https://api.github.com/users/karol-szustakowski/events{/privacy}",
"received_events_url": "https://api.github.com/users/karol-szustakowski/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T07:08:25 | 2025-07-15T15:16:11 | 2025-07-15T15:16:11 | NONE | null | null | null | null | ### System Info
bitsandbytes==0.46.0
transfomers==4.51.3
Python 3.9.21
### Who can help?
@SunMarc @MekkCyber
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
After calling `model.dequantize()` I would expect to be able to handle the model as if it were never quantized.
For example, I expected to be able to do this:
```py
model = model.dequantize()
model = model.to(dtype=torch.float16)
```
The above fails with the exception:
`
ValueError: You cannot cast a bitsandbytes model in a new `dtype`. Make sure to load the model using `from_pretrained` using the desired `dtype` by passing the correct `torch_dtype` argument.
`
### Expected behavior
The `quantization_method` should be cleared after the model is dequantized, this currently can be circumvented by:
```py
model = model.dequantize()
model.quantization_method = None
model = model.to(dtype=torch.float16)
``` | {
"login": "SunMarc",
"id": 57196510,
"node_id": "MDQ6VXNlcjU3MTk2NTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/57196510?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SunMarc",
"html_url": "https://github.com/SunMarc",
"followers_url": "https://api.github.com/users/SunMarc/followers",
"following_url": "https://api.github.com/users/SunMarc/following{/other_user}",
"gists_url": "https://api.github.com/users/SunMarc/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SunMarc/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SunMarc/subscriptions",
"organizations_url": "https://api.github.com/users/SunMarc/orgs",
"repos_url": "https://api.github.com/users/SunMarc/repos",
"events_url": "https://api.github.com/users/SunMarc/events{/privacy}",
"received_events_url": "https://api.github.com/users/SunMarc/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39295/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39295/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39294 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39294/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39294/comments | https://api.github.com/repos/huggingface/transformers/issues/39294/events | https://github.com/huggingface/transformers/pull/39294 | 3,214,728,744 | PR_kwDOCUB6oc6eD9o2 | 39,294 | [flash attn 3] bring back flags | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-09T06:45:01 | 2025-07-09T07:49:45 | 2025-07-09T07:45:01 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39294",
"html_url": "https://github.com/huggingface/transformers/pull/39294",
"diff_url": "https://github.com/huggingface/transformers/pull/39294.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39294.patch",
"merged_at": "2025-07-09T07:45:01"
} | # What does this PR do?
Fixes https://github.com/huggingface/transformers/issues/39288 and brings back `_supports_flash_attn3=True` flags in all models where FA2 is supported
I believe the flag was deleted by accident, but lmk if that was intended | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39294/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39294/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39293 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39293/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39293/comments | https://api.github.com/repos/huggingface/transformers/issues/39293/events | https://github.com/huggingface/transformers/pull/39293 | 3,214,724,084 | PR_kwDOCUB6oc6eD8mi | 39,293 | Add T5LA models | {
"login": "HRezaei",
"id": 657216,
"node_id": "MDQ6VXNlcjY1NzIxNg==",
"avatar_url": "https://avatars.githubusercontent.com/u/657216?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HRezaei",
"html_url": "https://github.com/HRezaei",
"followers_url": "https://api.github.com/users/HRezaei/followers",
"following_url": "https://api.github.com/users/HRezaei/following{/other_user}",
"gists_url": "https://api.github.com/users/HRezaei/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HRezaei/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HRezaei/subscriptions",
"organizations_url": "https://api.github.com/users/HRezaei/orgs",
"repos_url": "https://api.github.com/users/HRezaei/repos",
"events_url": "https://api.github.com/users/HRezaei/events{/privacy}",
"received_events_url": "https://api.github.com/users/HRezaei/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 1843244711,
"node_id": "MDU6TGFiZWwxODQzMjQ0NzEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/New%20model",
"name": "New model",
"color": "fbca04",
"default": false,
"description": ""
}
] | open | false | null | [] | null | [] | 2025-07-09T06:43:41 | 2025-09-19T04:31:55 | null | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39293",
"html_url": "https://github.com/huggingface/transformers/pull/39293",
"diff_url": "https://github.com/huggingface/transformers/pull/39293.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39293.patch",
"merged_at": null
} | # What does this PR do?
Adds the implementation of the LookAhead (LA) models. These models are designed to predict not only the next immediate token after the input prompt, but also the second, third, ... up to K next tokens. These tokens can be used to mitigate the high inference latency in generation (see [this survey](https://aclanthology.org/2024.findings-acl.456/)) or in approximated ranking of a set of responses (see [this paper](https://openreview.net/pdf?id=D38rTnrkal) for an application).
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
@ArthurZucker
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39293/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39293/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39292 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39292/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39292/comments | https://api.github.com/repos/huggingface/transformers/issues/39292/events | https://github.com/huggingface/transformers/pull/39292 | 3,214,715,282 | PR_kwDOCUB6oc6eD6ps | 39,292 | Feat: add Kwai-Keye transformers | {
"login": "Kwai-Keye",
"id": 216671074,
"node_id": "U_kgDODOojYg",
"avatar_url": "https://avatars.githubusercontent.com/u/216671074?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Kwai-Keye",
"html_url": "https://github.com/Kwai-Keye",
"followers_url": "https://api.github.com/users/Kwai-Keye/followers",
"following_url": "https://api.github.com/users/Kwai-Keye/following{/other_user}",
"gists_url": "https://api.github.com/users/Kwai-Keye/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Kwai-Keye/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Kwai-Keye/subscriptions",
"organizations_url": "https://api.github.com/users/Kwai-Keye/orgs",
"repos_url": "https://api.github.com/users/Kwai-Keye/repos",
"events_url": "https://api.github.com/users/Kwai-Keye/events{/privacy}",
"received_events_url": "https://api.github.com/users/Kwai-Keye/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 1843244711,
"node_id": "MDU6TGFiZWwxODQzMjQ0NzEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/New%20model",
"name": "New model",
"color": "fbca04",
"default": false,
"description": ""
}
] | closed | false | null | [] | null | [] | 2025-07-09T06:41:03 | 2025-09-22T08:36:45 | 2025-09-22T08:36:45 | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39292",
"html_url": "https://github.com/huggingface/transformers/pull/39292",
"diff_url": "https://github.com/huggingface/transformers/pull/39292.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39292.patch",
"merged_at": null
} | Add support for Kwai-Keye/Keye-VL-8B-Preview model
## Description
This pull request adds support for the Keye-VL-8B-Preview model developed by Kwai-Keye. Keye-VL-8B-Preview is an advanced vision-language model that demonstrates strong performance in video understanding, visual perception, and reasoning tasks,.
The model repository can be found at:
* Hugging Face Hub: [https://huggingface.co/Kw](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[ai-Ke](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[ye/Ke](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[ye-VL](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[-8B-P](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[revie](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)[w](https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview)
* GitHub: [https://github.com](https://github.com/Kwai-Keye/Keye)[/Kwai](https://github.com/Kwai-Keye/Keye)[-Keye](https://github.com/Kwai-Keye/Keye)[/Keye](https://github.com/Kwai-Keye/Keye)
## Key Changes
1. Added model configuration files for Keye-VL-8B-Preview
2. Implemented model architecture code based on the official specifications
3. Added tokenizer support for the model's specific tokenization requirements
4. Included example usage scripts in the documentation
## Model Architecture
The model consists of:
* A Siglip vision encoder for processing image/video inputs
* A Qwen3 decoder for language understanding and generation
Usage Example
```
import torch
from transformers import KeyeForConditionalGeneration, AutoProcessor
from PIL import Image
import requests
model = KeyeForConditionalGeneration.from_pretrained(
"Kwai-Keye/Keye-VL-8B-Preview",
torch_dtype=torch.float16,
device_map="auto",
attn_implementation="sdpa",
trust_remote_code=True
)
processor = AutoProcessor.from_pretrained("Kwai-Keye/Keye-VL-8B-Preview", trust_remote_code=True)
url = "https://s1-11508.kwimgs.com/kos/nlav11508/mllm_all/ziran_jiafeimao_11.jpg"
messages = [
{
"role":"user",
"content":[
{
"type":"image",
"image": url,
},
{
"type":"text",
"text":"Describe this image."
}
]
}
]
image_inputs = [Image.open(requests.get(url, stream=True).raw)]
text = processor.apply_chat_template(
messages, tokenize=False, add_generation_prompt=True
)
inputs = processor(
text=[text],
images=image_inputs,
videos=None,
padding=True,
return_tensors="pt",
).to(model.device)
generated_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids_trimmed = [
out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text)
```
Checklist
- [ ] Model code is properly formatted and follows transformers coding guidelines
- [ ] Documentation is updated with usage examples
- [ ] All new and existing tests pass locally with the changes
We believe that integrating Keye-VL-8B-Preview into the transformers library will provide users with another powerful option for vision-language tasks. We welcome any feedback or suggestions for improving this integration. | {
"login": "Kwai-Keye",
"id": 216671074,
"node_id": "U_kgDODOojYg",
"avatar_url": "https://avatars.githubusercontent.com/u/216671074?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Kwai-Keye",
"html_url": "https://github.com/Kwai-Keye",
"followers_url": "https://api.github.com/users/Kwai-Keye/followers",
"following_url": "https://api.github.com/users/Kwai-Keye/following{/other_user}",
"gists_url": "https://api.github.com/users/Kwai-Keye/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Kwai-Keye/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Kwai-Keye/subscriptions",
"organizations_url": "https://api.github.com/users/Kwai-Keye/orgs",
"repos_url": "https://api.github.com/users/Kwai-Keye/repos",
"events_url": "https://api.github.com/users/Kwai-Keye/events{/privacy}",
"received_events_url": "https://api.github.com/users/Kwai-Keye/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39292/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39292/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39291 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39291/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39291/comments | https://api.github.com/repos/huggingface/transformers/issues/39291/events | https://github.com/huggingface/transformers/issues/39291 | 3,214,439,480 | I_kwDOCUB6oc6_mHQ4 | 39,291 | Support 2D Array Inputs in Wav2Vec2FeatureExtractor for Non-Waveform Modalities | {
"login": "cyberso",
"id": 111082513,
"node_id": "U_kgDOBp78EQ",
"avatar_url": "https://avatars.githubusercontent.com/u/111082513?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cyberso",
"html_url": "https://github.com/cyberso",
"followers_url": "https://api.github.com/users/cyberso/followers",
"following_url": "https://api.github.com/users/cyberso/following{/other_user}",
"gists_url": "https://api.github.com/users/cyberso/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cyberso/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cyberso/subscriptions",
"organizations_url": "https://api.github.com/users/cyberso/orgs",
"repos_url": "https://api.github.com/users/cyberso/repos",
"events_url": "https://api.github.com/users/cyberso/events{/privacy}",
"received_events_url": "https://api.github.com/users/cyberso/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T04:21:28 | 2025-08-20T08:03:02 | 2025-08-20T08:03:02 | NONE | null | null | null | null | ### System Info
Transformers version: 4.17.0
Platform: Linux-5.15.0-43-generic-x86_64-with-debian-bookworm-sid
Python version: 3.7.12
PyTorch version: 1.13.1+cu117
TensorFlow version: 2.11.0
Flax version: Not installed
Tokenizers version: 0.13.3
HuggingFace Hub version: 0.16.4
Datasets version: 2.13.2
Is CUDA available (torch)? True
GPU device name: NVIDIA GeForce GTX 1080 Ti
Is GPU available (tensorflow)? True
### Who can help?
speech models: @eustlb
### Information
- [x] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
The `Wav2Vec2FeatureExtractor` currently only accepts input as a list of 1D NumPy arrays due to hardcoded padding logic in its base class `SequenceFeatureExtractor`. This restricts its use to raw waveform data and limits its flexibility for repurposing the model architecture for other input types (e.g., 2D spectrograms).
Enhancing `Wav2Vec2FeatureExtractor` to support list of 2D NumPy arrays (e.g., shape `[featureDim, time]`) would improve its compatibility with alternative modalities and workflows, such as precomputed features or spectrograms.
```
import numpy as np
from transformers import Wav2Vec2FeatureExtractor
# Initialize the feature extractor
feature_extractor = Wav2Vec2FeatureExtractor(sampling_rate=16000)
# 1D input: (Time) β should be accepted
waveform_1d = np.random.rand(16000) # 1 second of mono audio
output = feature_extractor(waveform_1d, return_tensors="np")
print("Expand the dimension to append batch")
print(output['input_values'].shape)
# 2D input: (Batch, Time)
waveform_list = [np.random.rand(l) for l in np.random.randint(10000, 16000, size=10)]
output = feature_extractor(waveform_list, return_tensors="np", padding=True)
print("Consider the first dimension as batch")
print(output['input_values'].shape)
# 3D input: (Batch, Channels, Time) β should raise error
waveform_3d = [np.random.rand(2, l) for l in np.random.randint(1000, 1600, size=10)]
output = feature_extractor(waveform_3d, return_tensors="np", padding=True)
print("Incorrectly expand the dimension to append batch")
print(output['input_values'].shape)
```
1D and list-of-1D inputs behave as expected. However, an error occurs when passing a 2D NumPy array.
In this case, Wav2Vec2FeatureExtractor incorrectly interprets the feature dimension as the batch axis, leading to unintended behavior.
The detailed error message is provided below.
```
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
~/.conda/envs/neur_vec/lib/python3.7/site-packages/transformers/feature_extraction_utils.py in convert_to_tensors(self, tensor_type)
168 if not is_tensor(value):
--> 169 tensor = as_tensor(value)
170
ValueError: could not broadcast input array from shape (2,1583) into shape (2,)
During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
/tmp/ipykernel_212/846801801.py in <module>
20 # 3D input: (Batch, Channels, Time) β should raise error
21 waveform_3d = [np.random.rand(2, l) for l in np.random.randint(1000, 1600, size=10)]
---> 22 output = feature_extractor(waveform_3d, return_tensors="np", padding=True)
23 print("Incorrectly expand the dimension to append batch")
24 print(output['input_values'].shape)
~/.conda/envs/neur_vec/lib/python3.7/site-packages/transformers/models/wav2vec2/feature_extraction_wav2vec2.py in __call__(self, raw_speech, padding, max_length, truncation, pad_to_multiple_of, return_attention_mask, return_tensors, sampling_rate, **kwargs)
200 truncation=truncation,
201 pad_to_multiple_of=pad_to_multiple_of,
--> 202 return_attention_mask=return_attention_mask,
203 )
204
~/.conda/envs/neur_vec/lib/python3.7/site-packages/transformers/feature_extraction_sequence_utils.py in pad(self, processed_features, padding, max_length, truncation, pad_to_multiple_of, return_attention_mask, return_tensors)
230 batch_outputs[key].append(value)
231
--> 232 return BatchFeature(batch_outputs, tensor_type=return_tensors)
233
234 def _pad(
~/.conda/envs/neur_vec/lib/python3.7/site-packages/transformers/feature_extraction_utils.py in __init__(self, data, tensor_type)
78 def __init__(self, data: Optional[Dict[str, Any]] = None, tensor_type: Union[None, str, TensorType] = None):
79 super().__init__(data)
---> 80 self.convert_to_tensors(tensor_type=tensor_type)
81
82 def __getitem__(self, item: str) -> Union[Any]:
~/.conda/envs/neur_vec/lib/python3.7/site-packages/transformers/feature_extraction_utils.py in convert_to_tensors(self, tensor_type)
174 raise ValueError("Unable to create tensor returning overflowing values of different lengths. ")
175 raise ValueError(
--> 176 "Unable to create tensor, you should probably activate padding "
177 "with 'padding=True' to have batched tensors with the same length."
178 )
ValueError: Unable to create tensor, you should probably activate padding with 'padding=True' to have batched tensors with the same length.
```
### Expected behavior
1D input: Expand the dimension to append batch
(1, 16000)
2D input: Consider number of array in the list as batch, padd across trials with variable length
(10, 14978)
3D input: Consider number of array in the list a as batch, padd across trials with variable length
(10, 2, 14978) | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39291/reactions",
"total_count": 4,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 2,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39291/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39290 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39290/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39290/comments | https://api.github.com/repos/huggingface/transformers/issues/39290/events | https://github.com/huggingface/transformers/issues/39290 | 3,214,087,656 | I_kwDOCUB6oc6_kxXo | 39,290 | v4.53.0+ starts erroring with 'Gemma3TextConfig' object has no attribute 'sliding_window_pattern' with vLLM | {
"login": "es94129",
"id": 12763339,
"node_id": "MDQ6VXNlcjEyNzYzMzM5",
"avatar_url": "https://avatars.githubusercontent.com/u/12763339?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/es94129",
"html_url": "https://github.com/es94129",
"followers_url": "https://api.github.com/users/es94129/followers",
"following_url": "https://api.github.com/users/es94129/following{/other_user}",
"gists_url": "https://api.github.com/users/es94129/gists{/gist_id}",
"starred_url": "https://api.github.com/users/es94129/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/es94129/subscriptions",
"organizations_url": "https://api.github.com/users/es94129/orgs",
"repos_url": "https://api.github.com/users/es94129/repos",
"events_url": "https://api.github.com/users/es94129/events{/privacy}",
"received_events_url": "https://api.github.com/users/es94129/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-09T00:28:57 | 2025-07-15T12:16:17 | 2025-07-09T14:10:40 | NONE | null | null | null | null | ### System Info
- `transformers` version: 4.53.1
- Platform: Linux-5.10.192-183.736.amzn2.x86_64-x86_64-with-glibc2.31
- Python version: 3.11.13
- Huggingface_hub version: 0.33.2
- Safetensors version: 0.5.3
- Accelerate version: not installed
- Accelerate config: not found
- DeepSpeed version: not installed
- PyTorch version (accelerator?): 2.6.0+cu124 (CUDA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: No
- Using GPU in script?: Yes
- GPU type: NVIDIA H100 80GB HBM3
### Who can help?
@ArthurZucker @Cyrilvallez
### Reproduction
With vLLM <= 0.8.5.post1, upgrading transformers to 4.53.0 and above causes `AttributeError: 'Gemma3TextConfig' object has no attribute 'sliding_window_pattern'.`, likely because of the changes to Gemma 3 in this PR: https://github.com/huggingface/transformers/pull/37866.
```sh
pip install transformers==4.53.1 # latest version, as long as >= 4.53.0 breaks
pip install vllm==0.8.4
```
```python
from vllm import LLM
llm = LLM(model="google/gemma-3-12b-it")
```
<details>
<summary>Error stacktrace</summary>
<pre>
```
ERROR 07-08 22:51:23 [core.py:396] Traceback (most recent call last):
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/v1/engine/core.py", line 387, in run_engine_core
ERROR 07-08 22:51:23 [core.py:396] engine_core = EngineCoreProc(*args, **kwargs)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/v1/engine/core.py", line 329, in __init__
ERROR 07-08 22:51:23 [core.py:396] super().__init__(vllm_config, executor_class, log_stats,
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/v1/engine/core.py", line 64, in __init__
ERROR 07-08 22:51:23 [core.py:396] self.model_executor = executor_class(vllm_config)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/executor/executor_base.py", line 52, in __init__
ERROR 07-08 22:51:23 [core.py:396] self._init_executor()
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/executor/uniproc_executor.py", line 47, in _init_executor
ERROR 07-08 22:51:23 [core.py:396] self.collective_rpc("load_model")
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
ERROR 07-08 22:51:23 [core.py:396] answer = run_method(self.driver_worker, method, args, kwargs)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/utils.py", line 2456, in run_method
ERROR 07-08 22:51:23 [core.py:396] return func(*args, **kwargs)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/v1/worker/gpu_worker.py", line 162, in load_model
ERROR 07-08 22:51:23 [core.py:396] self.model_runner.load_model()
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/v1/worker/gpu_model_runner.py", line 1332, in load_model
ERROR 07-08 22:51:23 [core.py:396] self.model = get_model(vllm_config=self.vllm_config)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/model_loader/__init__.py", line 14, in get_model
ERROR 07-08 22:51:23 [core.py:396] return loader.load_model(vllm_config=vllm_config)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/model_loader/loader.py", line 452, in load_model
ERROR 07-08 22:51:23 [core.py:396] model = _initialize_model(vllm_config=vllm_config)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/model_loader/loader.py", line 133, in _initialize_model
ERROR 07-08 22:51:23 [core.py:396] return model_class(vllm_config=vllm_config, prefix=prefix)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3_mm.py", line 490, in __init__
ERROR 07-08 22:51:23 [core.py:396] self.language_model = init_vllm_registered_model(
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/utils.py", line 286, in init_vllm_registered_model
ERROR 07-08 22:51:23 [core.py:396] return _initialize_model(vllm_config=vllm_config, prefix=prefix)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/model_loader/loader.py", line 133, in _initialize_model
ERROR 07-08 22:51:23 [core.py:396] return model_class(vllm_config=vllm_config, prefix=prefix)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3.py", line 493, in __init__
ERROR 07-08 22:51:23 [core.py:396] self.model = Gemma3Model(vllm_config=vllm_config,
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/compilation/decorators.py", line 151, in __init__
ERROR 07-08 22:51:23 [core.py:396] old_init(self, vllm_config=vllm_config, prefix=prefix, **kwargs)
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3.py", line 360, in __init__
ERROR 07-08 22:51:23 [core.py:396] self.start_layer, self.end_layer, self.layers = make_layers(
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/utils.py", line 609, in make_layers
ERROR 07-08 22:51:23 [core.py:396] [PPMissingLayer() for _ in range(start_layer)] + [
ERROR 07-08 22:51:23 [core.py:396] ^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/utils.py", line 610, in <listcomp>
ERROR 07-08 22:51:23 [core.py:396] maybe_offload_to_cpu(layer_fn(prefix=f"{prefix}.{idx}"))
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3.py", line 362, in <lambda>
ERROR 07-08 22:51:23 [core.py:396] lambda prefix: Gemma3DecoderLayer(
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3.py", line 288, in __init__
ERROR 07-08 22:51:23 [core.py:396] self.self_attn = Gemma3Attention(
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/vllm/model_executor/models/gemma3.py", line 151, in __init__
ERROR 07-08 22:51:23 [core.py:396] (layer_idx + 1) % config.sliding_window_pattern))
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] File "/root/miniconda3/envs/transformers-issue/lib/python3.11/site-packages/transformers/configuration_utils.py", line 209, in __getattribute__
ERROR 07-08 22:51:23 [core.py:396] return super().__getattribute__(key)
ERROR 07-08 22:51:23 [core.py:396] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 07-08 22:51:23 [core.py:396] AttributeError: 'Gemma3TextConfig' object has no attribute 'sliding_window_pattern'
```
</pre>
</details>
Newer versions of vLLM also have quality issues particularly when upgrading transformers>=4.53.0 which are reported in https://github.com/vllm-project/vllm/issues/20341 .
### Expected behavior
Should have the same behavior as transformers 4.52.4 + vLLM 0.8.4
```python
from vllm import LLM
llm = LLM(model="google/gemma-3-12b-it")
print(llm.generate("what is transformers")[0].outputs[0])
```
```python
CompletionOutput(index=0, text='?>\n\nTransformers are a powerful type of neural network architecture that has revolutionized the', token_ids=[255999, 13765, 108, 214568, 659, 496, 8632, 1722, 529, 22823, 3707, 13217, 600, 815, 176839, 506], cumulative_logprob=None, logprobs=None, finish_reason=length, stop_reason=None)
``` | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39290/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 1
} | https://api.github.com/repos/huggingface/transformers/issues/39290/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39289 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39289/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39289/comments | https://api.github.com/repos/huggingface/transformers/issues/39289/events | https://github.com/huggingface/transformers/pull/39289 | 3,213,908,523 | PR_kwDOCUB6oc6eBQEf | 39,289 | Fix broken SAM after #39120 | {
"login": "yonigozlan",
"id": 74535834,
"node_id": "MDQ6VXNlcjc0NTM1ODM0",
"avatar_url": "https://avatars.githubusercontent.com/u/74535834?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yonigozlan",
"html_url": "https://github.com/yonigozlan",
"followers_url": "https://api.github.com/users/yonigozlan/followers",
"following_url": "https://api.github.com/users/yonigozlan/following{/other_user}",
"gists_url": "https://api.github.com/users/yonigozlan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yonigozlan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yonigozlan/subscriptions",
"organizations_url": "https://api.github.com/users/yonigozlan/orgs",
"repos_url": "https://api.github.com/users/yonigozlan/repos",
"events_url": "https://api.github.com/users/yonigozlan/events{/privacy}",
"received_events_url": "https://api.github.com/users/yonigozlan/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T22:24:47 | 2025-07-09T21:46:22 | 2025-07-09T21:46:22 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39289",
"html_url": "https://github.com/huggingface/transformers/pull/39289",
"diff_url": "https://github.com/huggingface/transformers/pull/39289.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39289.patch",
"merged_at": "2025-07-09T21:46:22"
} | Was going crazy with tests failing on Sam2 after merging with main, turns out there's a typo in SAM after https://github.com/huggingface/transformers/pull/39120
Cc @ArthurZucker | {
"login": "yonigozlan",
"id": 74535834,
"node_id": "MDQ6VXNlcjc0NTM1ODM0",
"avatar_url": "https://avatars.githubusercontent.com/u/74535834?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yonigozlan",
"html_url": "https://github.com/yonigozlan",
"followers_url": "https://api.github.com/users/yonigozlan/followers",
"following_url": "https://api.github.com/users/yonigozlan/following{/other_user}",
"gists_url": "https://api.github.com/users/yonigozlan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yonigozlan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yonigozlan/subscriptions",
"organizations_url": "https://api.github.com/users/yonigozlan/orgs",
"repos_url": "https://api.github.com/users/yonigozlan/repos",
"events_url": "https://api.github.com/users/yonigozlan/events{/privacy}",
"received_events_url": "https://api.github.com/users/yonigozlan/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39289/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39289/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39288 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39288/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39288/comments | https://api.github.com/repos/huggingface/transformers/issues/39288/events | https://github.com/huggingface/transformers/issues/39288 | 3,213,849,026 | I_kwDOCUB6oc6_j3HC | 39,288 | flash_attention_3 for Qwen2_5_VisionTransformerPretrainedModel | {
"login": "aidando73",
"id": 43259657,
"node_id": "MDQ6VXNlcjQzMjU5NjU3",
"avatar_url": "https://avatars.githubusercontent.com/u/43259657?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/aidando73",
"html_url": "https://github.com/aidando73",
"followers_url": "https://api.github.com/users/aidando73/followers",
"following_url": "https://api.github.com/users/aidando73/following{/other_user}",
"gists_url": "https://api.github.com/users/aidando73/gists{/gist_id}",
"starred_url": "https://api.github.com/users/aidando73/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/aidando73/subscriptions",
"organizations_url": "https://api.github.com/users/aidando73/orgs",
"repos_url": "https://api.github.com/users/aidando73/repos",
"events_url": "https://api.github.com/users/aidando73/events{/privacy}",
"received_events_url": "https://api.github.com/users/aidando73/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 2648621985,
"node_id": "MDU6TGFiZWwyNjQ4NjIxOTg1",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Feature%20request",
"name": "Feature request",
"color": "FBCA04",
"default": false,
"description": "Request for a new feature"
}
] | closed | false | null | [] | null | [] | 2025-07-08T21:46:24 | 2025-07-10T05:50:08 | 2025-07-09T07:45:04 | NONE | null | null | null | null | ### Feature request
As per title
### Motivation
Would like better performance.
### Your contribution
Let me know if you need any more details | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39288/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39288/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39287 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39287/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39287/comments | https://api.github.com/repos/huggingface/transformers/issues/39287/events | https://github.com/huggingface/transformers/pull/39287 | 3,213,818,732 | PR_kwDOCUB6oc6eA8xm | 39,287 | feat: add support for gradient checkpointing for TimmWrapperModel and TimmWrapperForImageClassification | {
"login": "Yozer",
"id": 9254709,
"node_id": "MDQ6VXNlcjkyNTQ3MDk=",
"avatar_url": "https://avatars.githubusercontent.com/u/9254709?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Yozer",
"html_url": "https://github.com/Yozer",
"followers_url": "https://api.github.com/users/Yozer/followers",
"following_url": "https://api.github.com/users/Yozer/following{/other_user}",
"gists_url": "https://api.github.com/users/Yozer/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Yozer/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Yozer/subscriptions",
"organizations_url": "https://api.github.com/users/Yozer/orgs",
"repos_url": "https://api.github.com/users/Yozer/repos",
"events_url": "https://api.github.com/users/Yozer/events{/privacy}",
"received_events_url": "https://api.github.com/users/Yozer/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T21:29:13 | 2025-07-22T11:08:26 | 2025-07-22T11:07:52 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39287",
"html_url": "https://github.com/huggingface/transformers/pull/39287",
"diff_url": "https://github.com/huggingface/transformers/pull/39287.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39287.patch",
"merged_at": "2025-07-22T11:07:52"
} | # What does this PR do?
This enables gradient checkpointing for timm wrapped models.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [x] Did you write any new necessary tests?
## Who can review?
@qubvel
| {
"login": "qubvel",
"id": 31920396,
"node_id": "MDQ6VXNlcjMxOTIwMzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/31920396?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qubvel",
"html_url": "https://github.com/qubvel",
"followers_url": "https://api.github.com/users/qubvel/followers",
"following_url": "https://api.github.com/users/qubvel/following{/other_user}",
"gists_url": "https://api.github.com/users/qubvel/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qubvel/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qubvel/subscriptions",
"organizations_url": "https://api.github.com/users/qubvel/orgs",
"repos_url": "https://api.github.com/users/qubvel/repos",
"events_url": "https://api.github.com/users/qubvel/events{/privacy}",
"received_events_url": "https://api.github.com/users/qubvel/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39287/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39287/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39286 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39286/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39286/comments | https://api.github.com/repos/huggingface/transformers/issues/39286/events | https://github.com/huggingface/transformers/pull/39286 | 3,213,804,648 | PR_kwDOCUB6oc6eA5tv | 39,286 | Fix bugs in pytorch example run_clm when streaming is enabled | {
"login": "HRezaei",
"id": 657216,
"node_id": "MDQ6VXNlcjY1NzIxNg==",
"avatar_url": "https://avatars.githubusercontent.com/u/657216?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HRezaei",
"html_url": "https://github.com/HRezaei",
"followers_url": "https://api.github.com/users/HRezaei/followers",
"following_url": "https://api.github.com/users/HRezaei/following{/other_user}",
"gists_url": "https://api.github.com/users/HRezaei/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HRezaei/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HRezaei/subscriptions",
"organizations_url": "https://api.github.com/users/HRezaei/orgs",
"repos_url": "https://api.github.com/users/HRezaei/repos",
"events_url": "https://api.github.com/users/HRezaei/events{/privacy}",
"received_events_url": "https://api.github.com/users/HRezaei/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T21:21:15 | 2025-07-16T06:26:23 | 2025-07-15T13:37:28 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39286",
"html_url": "https://github.com/huggingface/transformers/pull/39286",
"diff_url": "https://github.com/huggingface/transformers/pull/39286.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39286.patch",
"merged_at": "2025-07-15T13:37:28"
} | # What does this PR do?
Fixes #39285
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [x] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
@ArthurZucker
@sgugger
| {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39286/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39286/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39285 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39285/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39285/comments | https://api.github.com/repos/huggingface/transformers/issues/39285/events | https://github.com/huggingface/transformers/issues/39285 | 3,213,782,871 | I_kwDOCUB6oc6_jm9X | 39,285 | Pytorch language_modelling example run_clm fails when streaming is enabled | {
"login": "HRezaei",
"id": 657216,
"node_id": "MDQ6VXNlcjY1NzIxNg==",
"avatar_url": "https://avatars.githubusercontent.com/u/657216?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HRezaei",
"html_url": "https://github.com/HRezaei",
"followers_url": "https://api.github.com/users/HRezaei/followers",
"following_url": "https://api.github.com/users/HRezaei/following{/other_user}",
"gists_url": "https://api.github.com/users/HRezaei/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HRezaei/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HRezaei/subscriptions",
"organizations_url": "https://api.github.com/users/HRezaei/orgs",
"repos_url": "https://api.github.com/users/HRezaei/repos",
"events_url": "https://api.github.com/users/HRezaei/events{/privacy}",
"received_events_url": "https://api.github.com/users/HRezaei/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-08T21:09:40 | 2025-07-15T13:37:29 | 2025-07-15T13:37:29 | CONTRIBUTOR | null | null | null | null | ### System Info
- `transformers` version: 4.53.0.dev0
- Platform: macOS-15.5-arm64-arm-64bit
- Python version: 3.10.16
- Huggingface_hub version: 0.30.0
- Safetensors version: 0.5.2
- Accelerate version: 1.8.1
- Accelerate config: not found
- DeepSpeed version: not installed
- PyTorch version (accelerator?): 2.5.1 (NA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: No
### Who can help?
@ArthurZucker
@sgugger
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [x] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Run the example file: `examples/pytorch/language-modeling/run_clm.py` with the argument `--streaming` enabled:
1. Clone the repository
2. Install transformers from source, `cd transformers && pip install .`
3. For instance, run the command below:
```shell
python examples/pytorch/language-modeling/run_clm.py --model_type gpt2 \
--tokenizer_name openai-community/gpt2 --dataset_name=HuggingFaceFW/fineweb \
--dataset_config_name=sample-350BT --streaming True --do_train --do_eval \
--overwrite_output_dir --max_steps=100 --max_train_samples=2000 --max_eval_samples=1000 \
--output_dir=/tmp/gpt2TestClm
```
3. You will see an error like `ValueError: Bad split: train[:5%]. Available splits: ['train']`
That's because in streaming mode, range splits are not available.
Also there are other bugs like running `len(train_dataset)` which raises an error if the dataset is in streaming mode.
[Here](https://colab.research.google.com/drive/10AhU2vT1QfTfHCRM2kdxEsu6ofpPICJq?usp=sharing) is a colab notebook to see the issue in action.
### Expected behavior
It is expected to pre-train a GPT2 model from scratch using the dataset in streaming mode
I'm working on a [PR](https://github.com/huggingface/transformers/pull/39286) that fixes this issue.
| {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39285/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39285/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39284 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39284/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39284/comments | https://api.github.com/repos/huggingface/transformers/issues/39284/events | https://github.com/huggingface/transformers/pull/39284 | 3,213,347,995 | PR_kwDOCUB6oc6d_Wn4 | 39,284 | Bump transformers from 4.48.0 to 4.52.1 in /examples/tensorflow/language-modeling-tpu | {
"login": "dependabot[bot]",
"id": 49699333,
"node_id": "MDM6Qm90NDk2OTkzMzM=",
"avatar_url": "https://avatars.githubusercontent.com/in/29110?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dependabot%5Bbot%5D",
"html_url": "https://github.com/apps/dependabot",
"followers_url": "https://api.github.com/users/dependabot%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/dependabot%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/dependabot%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dependabot%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dependabot%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/dependabot%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/dependabot%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/dependabot%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/dependabot%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 1905493434,
"node_id": "MDU6TGFiZWwxOTA1NDkzNDM0",
"url": "https://api.github.com/repos/huggingface/transformers/labels/dependencies",
"name": "dependencies",
"color": "0366d6",
"default": false,
"description": "Pull requests that update a dependency file"
},
{
"id": 6410654816,
"node_id": "LA_kwDOCUB6oc8AAAABfhrUYA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/python",
"name": "python",
"color": "2b67c6",
"default": false,
"description": "Pull requests that update Python code"
}
] | closed | false | null | [] | null | [] | 2025-07-08T18:00:36 | 2025-08-07T11:15:01 | 2025-08-07T11:14:59 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39284",
"html_url": "https://github.com/huggingface/transformers/pull/39284",
"diff_url": "https://github.com/huggingface/transformers/pull/39284.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39284.patch",
"merged_at": null
} | Bumps [transformers](https://github.com/huggingface/transformers) from 4.48.0 to 4.52.1.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a href="https://github.com/huggingface/transformers/releases">transformers's releases</a>.</em></p>
<blockquote>
<h2>Patch release v4.51.3</h2>
<p>A mix of bugs were fixed in this patch; very exceptionally, we diverge from semantic versioning to merge GLM-4 in this patch release.</p>
<ul>
<li>Handle torch ver in flexattn (<a href="https://redirect.github.com/huggingface/transformers/issues/37400">#37400</a>)</li>
<li>handle torch version edge cases (<a href="https://redirect.github.com/huggingface/transformers/issues/37399">#37399</a>)</li>
<li>Add glm4 (<a href="https://redirect.github.com/huggingface/transformers/issues/37388">#37388</a>)</li>
</ul>
<h1>Patch Release 4.51.2</h1>
<p>This is another round of bug fixes, but they are a lot more minor and outputs were not really affected!</p>
<ul>
<li>Fix Llama4 offset (<a href="https://redirect.github.com/huggingface/transformers/issues/37414">#37414</a>) by <a href="https://github.com/Cyrilvallez"><code>@βCyrilvallez</code></a></li>
<li>Attention Quantization with FBGemm & TP (<a href="https://redirect.github.com/huggingface/transformers/issues/37384">#37384</a>) by <a href="https://github.com/MekkCyber"><code>@βMekkCyber</code></a></li>
<li>use rms_norm_eps for the L2Norm for Llama4 (<a href="https://redirect.github.com/huggingface/transformers/issues/37418">#37418</a>) by <a href="https://github.com/danielhanchen"><code>@βdanielhanchen</code></a></li>
<li>mark llama4 as not supported with fa2 (<a href="https://redirect.github.com/huggingface/transformers/issues/37416">#37416</a>) by <a href="https://github.com/winglian"><code>@βwinglian</code></a></li>
</ul>
<h1>Patch release v4.51.1</h1>
<p>Since the release of Llama 4, we have fixed a few issues that we are now releasing in patch v4.51.1</p>
<ul>
<li>Fixing flex attention for torch=2.6.0 (<a href="https://redirect.github.com/huggingface/transformers/issues/37285">#37285</a>)</li>
<li>more fixes for post-training llama4 (<a href="https://redirect.github.com/huggingface/transformers/issues/37329">#37329</a>)</li>
<li>Remove HQQ from caching allocator warmup (<a href="https://redirect.github.com/huggingface/transformers/issues/37347">#37347</a>)</li>
<li>fix derived berts _init_weights (<a href="https://redirect.github.com/huggingface/transformers/issues/37341">#37341</a>)</li>
<li>Fix init empty weights without accelerate (<a href="https://redirect.github.com/huggingface/transformers/issues/37337">#37337</a>)</li>
<li>Fix deepspeed with quantization (<a href="https://redirect.github.com/huggingface/transformers/issues/37324">#37324</a>)</li>
<li>fix llama4 training (<a href="https://redirect.github.com/huggingface/transformers/issues/37319">#37319</a>)</li>
<li>fix flex attn when optional args aren't passed (<a href="https://redirect.github.com/huggingface/transformers/issues/37327">#37327</a>)</li>
<li>Multiple llama4 fixe (<a href="https://redirect.github.com/huggingface/transformers/issues/37353">#37353</a>)</li>
</ul>
<p>Thanks all for your patience</p>
<h2>v4.51.0: Llama 4, Phi4-Multimodal, DeepSeek-v3, Qwen3</h2>
<h2>New Model Additions</h2>
<h3>Llama 4</h3>
<p><img src="https://github.com/user-attachments/assets/d613b292-94b0-4902-9dc7-2d00693222e4" alt="image" /></p>
<p>Llama 4, developed by Meta, introduces a new auto-regressive Mixture-of-Experts (MoE) architecture.This generation includes two models:</p>
<ul>
<li>The highly capable Llama 4 Maverick with 17B active parameters out of ~400B total, with 128 experts.</li>
<li>The efficient Llama 4 Scout also has 17B active parameters out of ~109B total, using just 16 experts.</li>
</ul>
<p>Both models leverage early fusion for native multimodality, enabling them to process text and image inputs. Maverick and Scout are both trained on up to 40 trillion tokens on data encompassing 200 languages (with specific fine-tuning support for 12 languages including Arabic, Spanish, German, and Hindi).</p>
<p>For deployment, Llama 4 Scout is designed for accessibility, fitting on a single server-grade GPU via on-the-fly 4-bit or 8-bit quantization, while Maverick is available in BF16 and FP8 formats. These models are released under the custom Llama 4 Community License Agreement, available on the model repositories</p>
<p>Getting started with Llama 4 using transformers is straightforward. Make sure you have transformers v4.51.0 or later installed:</p>
<pre><code>pip install -U transformers[hf_xet]
</tr></table>
</code></pre>
</blockquote>
<p>... (truncated)</p>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a href="https://github.com/huggingface/transformers/commit/945727948c1143a10ac6f7d811aa58bb0d126b5b"><code>9457279</code></a> Release: v4.52.1</li>
<li><a href="https://github.com/huggingface/transformers/commit/eaa301673a0a7a1a8c5d3f11c046d1592a7ae16b"><code>eaa3016</code></a> Revert parallelism temporarily (<a href="https://redirect.github.com/huggingface/transformers/issues/38240">#38240</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/b5f494632c0fff2527dd3140423408644a9b0076"><code>b5f4946</code></a> Protect ParallelInterface</li>
<li><a href="https://github.com/huggingface/transformers/commit/113424bcd53b92600f77d82f48add0a60fb41556"><code>113424b</code></a> Release: v4.52.0</li>
<li><a href="https://github.com/huggingface/transformers/commit/f834d368f6a21ed54188d9c96fbb9013b1d2c75f"><code>f834d36</code></a> [gemma3] fix bidirectional attention mask (<a href="https://redirect.github.com/huggingface/transformers/issues/38080">#38080</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/2edb0e4b4dda8172d5628ca7497a4125f28bf6fc"><code>2edb0e4</code></a> [mllama] fix loading and inference (<a href="https://redirect.github.com/huggingface/transformers/issues/38223">#38223</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/390f153469dfdc793e7a9c7eb4822ea76f4f796a"><code>390f153</code></a> Add padding-free to bamba (<a href="https://redirect.github.com/huggingface/transformers/issues/35861">#35861</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/2a79471318a9b7b16706f3bb5cd833c7e81919a6"><code>2a79471</code></a> Fixing Bitnet after use_rms_norm introduction (<a href="https://redirect.github.com/huggingface/transformers/issues/38229">#38229</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/9661896083c9d983341afa45cc4b84af01706e72"><code>9661896</code></a> Enable Quantize KV Cache for Mistral Model (<a href="https://redirect.github.com/huggingface/transformers/issues/35042">#35042</a>)</li>
<li><a href="https://github.com/huggingface/transformers/commit/1c2f36b480e02c9027d2523746d34e27b39e01a4"><code>1c2f36b</code></a> parallelism goes brrr (<a href="https://redirect.github.com/huggingface/transformers/issues/37877">#37877</a>)</li>
<li>Additional commits viewable in <a href="https://github.com/huggingface/transformers/compare/v4.48.0...v4.52.1">compare view</a></li>
</ul>
</details>
<br />
[](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)
Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.
[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)
---
<details>
<summary>Dependabot commands and options</summary>
<br />
You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
You can disable automated security fix PRs for this repo from the [Security Alerts page](https://github.com/huggingface/transformers/network/alerts).
</details> | {
"login": "dependabot[bot]",
"id": 49699333,
"node_id": "MDM6Qm90NDk2OTkzMzM=",
"avatar_url": "https://avatars.githubusercontent.com/in/29110?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dependabot%5Bbot%5D",
"html_url": "https://github.com/apps/dependabot",
"followers_url": "https://api.github.com/users/dependabot%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/dependabot%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/dependabot%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dependabot%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dependabot%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/dependabot%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/dependabot%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/dependabot%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/dependabot%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39284/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39284/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39283 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39283/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39283/comments | https://api.github.com/repos/huggingface/transformers/issues/39283/events | https://github.com/huggingface/transformers/pull/39283 | 3,213,325,013 | PR_kwDOCUB6oc6d_Rpc | 39,283 | fix: clarify pooled output logic in SigLIP2 | {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T17:50:33 | 2025-07-14T09:24:08 | 2025-07-11T10:43:05 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39283",
"html_url": "https://github.com/huggingface/transformers/pull/39283",
"diff_url": "https://github.com/huggingface/transformers/pull/39283.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39283.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes # (issue)
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [x] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [x] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [x] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
Clarified misleading comment in `modeling_siglip2.py` regarding pooled output token position.
Updated the comment to reflect that the final token may be `<pad>` or `<eos>`, consistent with SigLIP2βs training (which does not use attention masks).
- Adjusted the `_no_split_modules` list to match Hugging Face's modular formatting guidelines.
- Verified consistency using `check_modular_conversion.py`.
#39269
| {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39283/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39283/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39282 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39282/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39282/comments | https://api.github.com/repos/huggingface/transformers/issues/39282/events | https://github.com/huggingface/transformers/pull/39282 | 3,213,319,981 | PR_kwDOCUB6oc6d_QiI | 39,282 | Fix: qwen2vl Position Id create_causal_mask dimension | {
"login": "jeffrey-dot-li",
"id": 46302202,
"node_id": "MDQ6VXNlcjQ2MzAyMjAy",
"avatar_url": "https://avatars.githubusercontent.com/u/46302202?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jeffrey-dot-li",
"html_url": "https://github.com/jeffrey-dot-li",
"followers_url": "https://api.github.com/users/jeffrey-dot-li/followers",
"following_url": "https://api.github.com/users/jeffrey-dot-li/following{/other_user}",
"gists_url": "https://api.github.com/users/jeffrey-dot-li/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jeffrey-dot-li/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jeffrey-dot-li/subscriptions",
"organizations_url": "https://api.github.com/users/jeffrey-dot-li/orgs",
"repos_url": "https://api.github.com/users/jeffrey-dot-li/repos",
"events_url": "https://api.github.com/users/jeffrey-dot-li/events{/privacy}",
"received_events_url": "https://api.github.com/users/jeffrey-dot-li/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T17:48:17 | 2025-07-24T18:34:29 | 2025-07-24T18:34:29 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39282",
"html_url": "https://github.com/huggingface/transformers/pull/39282",
"diff_url": "https://github.com/huggingface/transformers/pull/39282.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39282.patch",
"merged_at": null
} | # What does this PR do?
Fixes qwen2vl position id tensor for create_causal_mask. Qwen2VL uses a 3d embedding system for [time, x, y] dimensions, however `create_causal_mask` expects a 1d position id and will cause errors in this line https://github.com/huggingface/transformers/blob/0cf27916f09a1a99af55ef4f2f3e8675372f38b6/src/transformers/masking_utils.py#L709 if it is given a 3d position id.
This just uses position_id[0] for creating the causal mask.
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "jeffrey-dot-li",
"id": 46302202,
"node_id": "MDQ6VXNlcjQ2MzAyMjAy",
"avatar_url": "https://avatars.githubusercontent.com/u/46302202?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jeffrey-dot-li",
"html_url": "https://github.com/jeffrey-dot-li",
"followers_url": "https://api.github.com/users/jeffrey-dot-li/followers",
"following_url": "https://api.github.com/users/jeffrey-dot-li/following{/other_user}",
"gists_url": "https://api.github.com/users/jeffrey-dot-li/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jeffrey-dot-li/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jeffrey-dot-li/subscriptions",
"organizations_url": "https://api.github.com/users/jeffrey-dot-li/orgs",
"repos_url": "https://api.github.com/users/jeffrey-dot-li/repos",
"events_url": "https://api.github.com/users/jeffrey-dot-li/events{/privacy}",
"received_events_url": "https://api.github.com/users/jeffrey-dot-li/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39282/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39282/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39281 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39281/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39281/comments | https://api.github.com/repos/huggingface/transformers/issues/39281/events | https://github.com/huggingface/transformers/pull/39281 | 3,213,167,424 | PR_kwDOCUB6oc6d-vSs | 39,281 | [Tests] Update model_id in AIMv2 Tests | {
"login": "yaswanth19",
"id": 82788246,
"node_id": "MDQ6VXNlcjgyNzg4MjQ2",
"avatar_url": "https://avatars.githubusercontent.com/u/82788246?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yaswanth19",
"html_url": "https://github.com/yaswanth19",
"followers_url": "https://api.github.com/users/yaswanth19/followers",
"following_url": "https://api.github.com/users/yaswanth19/following{/other_user}",
"gists_url": "https://api.github.com/users/yaswanth19/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yaswanth19/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yaswanth19/subscriptions",
"organizations_url": "https://api.github.com/users/yaswanth19/orgs",
"repos_url": "https://api.github.com/users/yaswanth19/repos",
"events_url": "https://api.github.com/users/yaswanth19/events{/privacy}",
"received_events_url": "https://api.github.com/users/yaswanth19/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T16:45:37 | 2025-07-08T19:46:32 | 2025-07-08T19:46:32 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39281",
"html_url": "https://github.com/huggingface/transformers/pull/39281",
"diff_url": "https://github.com/huggingface/transformers/pull/39281.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39281.patch",
"merged_at": "2025-07-08T19:46:32"
} | As per the title, updates model_id in Aimv2 integration tests with that of official ones as the PRs for HF compatible weights are merged. | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39281/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39281/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39280 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39280/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39280/comments | https://api.github.com/repos/huggingface/transformers/issues/39280/events | https://github.com/huggingface/transformers/pull/39280 | 3,212,970,597 | PR_kwDOCUB6oc6d-E4- | 39,280 | [Core] [Offloading] Fix saving offloaded submodules | {
"login": "kylesayrs",
"id": 17103692,
"node_id": "MDQ6VXNlcjE3MTAzNjky",
"avatar_url": "https://avatars.githubusercontent.com/u/17103692?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kylesayrs",
"html_url": "https://github.com/kylesayrs",
"followers_url": "https://api.github.com/users/kylesayrs/followers",
"following_url": "https://api.github.com/users/kylesayrs/following{/other_user}",
"gists_url": "https://api.github.com/users/kylesayrs/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kylesayrs/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kylesayrs/subscriptions",
"organizations_url": "https://api.github.com/users/kylesayrs/orgs",
"repos_url": "https://api.github.com/users/kylesayrs/repos",
"events_url": "https://api.github.com/users/kylesayrs/events{/privacy}",
"received_events_url": "https://api.github.com/users/kylesayrs/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T15:29:23 | 2025-07-16T12:27:05 | 2025-07-16T08:44:41 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39280",
"html_url": "https://github.com/huggingface/transformers/pull/39280",
"diff_url": "https://github.com/huggingface/transformers/pull/39280.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39280.patch",
"merged_at": "2025-07-16T08:44:40"
} | ## Purpose ##
* Fixes issue with saving offloaded modules with offloaded submodules whose parent is also offloaded
* Sometimes `get_state_dict_from_offload` will update the state dict with meta tensors, since `align_module_device` only aligns direct parameters, not submodule parameters. If these submodule parameters are offloaded, they will be returned as meta tensors
* This fix is useful for models with finer-grained offloading as well as registering offloadable submodules such as transforms
## Changes ##
* This change attempts to load the shard state dict if either the entry is empty ("") or if the entry is a meta tensor (returned by a parent which does not place submodules)
## Testing ##
* Added test fails on main but succeeds with these changes | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39280/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39280/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39279 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39279/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39279/comments | https://api.github.com/repos/huggingface/transformers/issues/39279/events | https://github.com/huggingface/transformers/pull/39279 | 3,212,859,545 | PR_kwDOCUB6oc6d9s5p | 39,279 | docs: clarify pooled_output behavior in SigLIP2 | {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T14:54:31 | 2025-07-08T16:31:48 | 2025-07-08T15:29:00 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39279",
"html_url": "https://github.com/huggingface/transformers/pull/39279",
"diff_url": "https://github.com/huggingface/transformers/pull/39279.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39279.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes # (issue)
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [x] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [x] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [x] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39279/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39279/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39278 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39278/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39278/comments | https://api.github.com/repos/huggingface/transformers/issues/39278/events | https://github.com/huggingface/transformers/pull/39278 | 3,212,842,009 | PR_kwDOCUB6oc6d9pEv | 39,278 | [wip, experiment] parallel mdx doc builder | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T14:48:51 | 2025-08-19T09:24:24 | 2025-08-19T09:24:24 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39278",
"html_url": "https://github.com/huggingface/transformers/pull/39278",
"diff_url": "https://github.com/huggingface/transformers/pull/39278.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39278.patch",
"merged_at": null
} | # What does this PR do?
| {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39278/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39278/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39277 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39277/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39277/comments | https://api.github.com/repos/huggingface/transformers/issues/39277/events | https://github.com/huggingface/transformers/pull/39277 | 3,212,619,683 | PR_kwDOCUB6oc6d84bU | 39,277 | fix `aria` tests | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T13:42:35 | 2025-07-09T11:49:35 | 2025-07-09T11:49:33 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39277",
"html_url": "https://github.com/huggingface/transformers/pull/39277",
"diff_url": "https://github.com/huggingface/transformers/pull/39277.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39277.patch",
"merged_at": "2025-07-09T11:49:33"
} | # What does this PR do?
| {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39277/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39277/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39276 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39276/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39276/comments | https://api.github.com/repos/huggingface/transformers/issues/39276/events | https://github.com/huggingface/transformers/pull/39276 | 3,212,417,441 | PR_kwDOCUB6oc6d8MKf | 39,276 | fix flaky `test_generate_compile_model_forward` | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T12:46:03 | 2025-07-08T13:36:07 | 2025-07-08T13:36:05 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39276",
"html_url": "https://github.com/huggingface/transformers/pull/39276",
"diff_url": "https://github.com/huggingface/transformers/pull/39276.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39276.patch",
"merged_at": "2025-07-08T13:36:05"
} | # What does this PR do?
Found it flaky today in https://app.circleci.com/jobs/github/huggingface/transformers/1816845
(7 failed in a run of 5000 times)
After this PR, all 5000 runs pass | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39276/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39276/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39275 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39275/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39275/comments | https://api.github.com/repos/huggingface/transformers/issues/39275/events | https://github.com/huggingface/transformers/issues/39275 | 3,212,211,135 | I_kwDOCUB6oc6_dnO_ | 39,275 | hangs during training using deepspeed | {
"login": "mxjmtxrm",
"id": 167944756,
"node_id": "U_kgDOCgKiNA",
"avatar_url": "https://avatars.githubusercontent.com/u/167944756?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mxjmtxrm",
"html_url": "https://github.com/mxjmtxrm",
"followers_url": "https://api.github.com/users/mxjmtxrm/followers",
"following_url": "https://api.github.com/users/mxjmtxrm/following{/other_user}",
"gists_url": "https://api.github.com/users/mxjmtxrm/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mxjmtxrm/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mxjmtxrm/subscriptions",
"organizations_url": "https://api.github.com/users/mxjmtxrm/orgs",
"repos_url": "https://api.github.com/users/mxjmtxrm/repos",
"events_url": "https://api.github.com/users/mxjmtxrm/events{/privacy}",
"received_events_url": "https://api.github.com/users/mxjmtxrm/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-08T11:44:43 | 2025-08-17T08:02:45 | 2025-08-17T08:02:45 | NONE | null | null | null | null | ### System Info
- `transformers` version: 4.54.0.dev0
- Platform: Linux-5.15.0-105-generic-x86_64-with-glibc2.39
- Python version: 3.12.3
- Huggingface_hub version: 0.33.0
- Safetensors version: 0.5.3
- Accelerate version: 1.8.1
- Accelerate config: not found
- DeepSpeed version: 0.17.1
- PyTorch version (accelerator?): 2.8.0a0+5228986c39.nv25.05 (CUDA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: <fill in>
- Using GPU in script?: <fill in>
- GPU type: NVIDIA H100 80GB HBM3
### Who can help?
@SunMarc @zach-huggingface
### Information
- [ ] The official example scripts
- [ ] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
train moe model using run_clm.py with deepspeed zero3.
```
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 accelerate launch --config_file "deepspeed.yaml" run_clm.py \
--do_train \
--do_eval \
--model_name_or_path Qwen3-30B-A3B \
--trust_remote_code True \
--torch_dtype auto \
--train_file train.json \
--validation_file val.json \
--block_size 8192 \
--output_dir ./output \
--weight_decay 0.01 \
--learning_rate 5e-6 \
--lr_scheduler_type "cosine" \
--per_device_train_batch_size 2 \
--per_device_eval_batch_size 2 \
--gradient_accumulation_steps 1 \
--gradient_checkpointing True \
--gradient_checkpointing_kwargs '{"use_reentrant": true}' \
--max_grad_norm 0.5 \
--warmup_ratio 0.03 \
--bf16 \
--num_train_epochs 1 \
--aux_lambda 0 \
--eval_strategy "steps" \
--eval_steps 100 \
--save_strategy "steps" \
--save_steps 100 \
--report_to "tensorboard" \
--save_total_limit 2 \
--logging_steps 10 \
--log_level "info" \
--include_num_input_tokens_seen True \
--overwrite_output_dir \
--save_only_model False
```
deepspeed.yaml:
```
compute_environment: LOCAL_MACHINE
debug: false
deepspeed_config:
deepspeed_multinode_launcher: standard
offload_optimizer_device: none
offload_param_device: none
zero3_init_flag: true
zero3_save_16bit_model: true
zero_stage: 3
distributed_type: DEEPSPEED
downcast_bf16: 'no'
machine_rank: 0
main_training_function: main
mixed_precision: bf16
num_machines: 1
num_processes: 8
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false
```
The training process hangs here:
```
0%| | 1/4966 [00:57<78:36:48, 57.00s/it]
```
And the GPU utilization remains at 100% and does not change.
### Expected behavior
Where is the problem? How to solve it? | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39275/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39275/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39274 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39274/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39274/comments | https://api.github.com/repos/huggingface/transformers/issues/39274/events | https://github.com/huggingface/transformers/pull/39274 | 3,212,115,142 | PR_kwDOCUB6oc6d7I71 | 39,274 | Add support for logging number of image tokens | {
"login": "h-tonywu",
"id": 211311067,
"node_id": "U_kgDODJhZ2w",
"avatar_url": "https://avatars.githubusercontent.com/u/211311067?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/h-tonywu",
"html_url": "https://github.com/h-tonywu",
"followers_url": "https://api.github.com/users/h-tonywu/followers",
"following_url": "https://api.github.com/users/h-tonywu/following{/other_user}",
"gists_url": "https://api.github.com/users/h-tonywu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/h-tonywu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/h-tonywu/subscriptions",
"organizations_url": "https://api.github.com/users/h-tonywu/orgs",
"repos_url": "https://api.github.com/users/h-tonywu/repos",
"events_url": "https://api.github.com/users/h-tonywu/events{/privacy}",
"received_events_url": "https://api.github.com/users/h-tonywu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-08T11:17:09 | 2025-07-10T09:01:14 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39274",
"html_url": "https://github.com/huggingface/transformers/pull/39274",
"diff_url": "https://github.com/huggingface/transformers/pull/39274.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39274.patch",
"merged_at": null
} | # What does this PR do?
With a single `include_num_input_image_tokens_seen` flag in the `TrainingArguments` (here with `SFTConfig`), you can report the number of visual tokens seen by the model.
Example:
```python
from trl import SFTConfig
# Configure training arguments using SFTConfig
training_args = SFTConfig(
# insert other arguments for training
report_to="wandb",
include_num_input_image_tokens_seen=True, # new
)
```
**Result when training a SmolVL**
- Model seed: `HuggingFaceTB/SmolVLM-256M-Instruct`
- Dataset: `merve/vqav2-small`
**Why is it useful?**
When training VLMs, monitoring the number of processed visual tokens helps with knowing the proportion of image-related tokens that are seen by the model during training. Moreover, for VLMs with dynamic patching (e.g., Qwen2-VL) this can also provide a signal about the resolution of images ingested.
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation). β TODO
- [ ] Did you write any new necessary tests? β there are no existing tests for logging metrics in the repository, so I'm not sure writing my own test is relevant.
## Who can review?
@SunMarc, @yonigozlan, @qubvel
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39274/reactions",
"total_count": 6,
"+1": 3,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 2,
"rocket": 1,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39274/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39273 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39273/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39273/comments | https://api.github.com/repos/huggingface/transformers/issues/39273/events | https://github.com/huggingface/transformers/pull/39273 | 3,211,897,813 | PR_kwDOCUB6oc6d6Yy5 | 39,273 | [CI] fix docs | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T10:09:35 | 2025-07-08T10:31:06 | 2025-07-08T10:31:03 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39273",
"html_url": "https://github.com/huggingface/transformers/pull/39273",
"diff_url": "https://github.com/huggingface/transformers/pull/39273.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39273.patch",
"merged_at": "2025-07-08T10:31:03"
} | # What does this PR do?
Fixes docs, causing red CI
- aimv2 bad return docstring (from #36625)
- `glossary` missing from the Korean table of contents (from #38804) | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39273/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39273/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39272 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39272/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39272/comments | https://api.github.com/repos/huggingface/transformers/issues/39272/events | https://github.com/huggingface/transformers/pull/39272 | 3,211,872,919 | PR_kwDOCUB6oc6d6TNb | 39,272 | Update Model Card for Encoder Decoder Model | {
"login": "ParagEkbote",
"id": 69567729,
"node_id": "MDQ6VXNlcjY5NTY3NzI5",
"avatar_url": "https://avatars.githubusercontent.com/u/69567729?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ParagEkbote",
"html_url": "https://github.com/ParagEkbote",
"followers_url": "https://api.github.com/users/ParagEkbote/followers",
"following_url": "https://api.github.com/users/ParagEkbote/following{/other_user}",
"gists_url": "https://api.github.com/users/ParagEkbote/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ParagEkbote/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ParagEkbote/subscriptions",
"organizations_url": "https://api.github.com/users/ParagEkbote/orgs",
"repos_url": "https://api.github.com/users/ParagEkbote/repos",
"events_url": "https://api.github.com/users/ParagEkbote/events{/privacy}",
"received_events_url": "https://api.github.com/users/ParagEkbote/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T10:02:55 | 2025-07-11T18:24:41 | 2025-07-11T18:23:08 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39272",
"html_url": "https://github.com/huggingface/transformers/pull/39272",
"diff_url": "https://github.com/huggingface/transformers/pull/39272.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39272.patch",
"merged_at": "2025-07-11T18:23:08"
} | # What does this PR do?
As described in the issue, this PR updates the model card for the encoder decoder model with an additional translation example. I have also re-added the contributor names for the Mamba and Mamba-2 models which were previously removed by me. Please let me know if any modifications are required and I will make the necessary changes.
Fixes #8944
Refs #36979
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
## Who can review?
@stevhliu | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39272/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39272/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39271 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39271/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39271/comments | https://api.github.com/repos/huggingface/transformers/issues/39271/events | https://github.com/huggingface/transformers/pull/39271 | 3,211,844,081 | PR_kwDOCUB6oc6d6M08 | 39,271 | [smollm3] add tokenizer mapping for `smollm3` | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 8103865784,
"node_id": "LA_kwDOCUB6oc8AAAAB4wctuA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/for%20patch",
"name": "for patch",
"color": "D93F0B",
"default": false,
"description": "Tag issues / labels that should be included in the next patch"
}
] | closed | false | null | [] | null | [] | 2025-07-08T09:54:07 | 2025-07-08T12:28:32 | 2025-07-08T10:44:01 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39271",
"html_url": "https://github.com/huggingface/transformers/pull/39271",
"diff_url": "https://github.com/huggingface/transformers/pull/39271.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39271.patch",
"merged_at": "2025-07-08T10:44:01"
} | # What does this PR do?
Adds a tokenizer mapping for `smollm3`, which fixes the following script
```py
from transformers import pipeline
# tokenizer is infered from the model, through the tokenizer mapping
pipe = pipeline("text-generation", model="HuggingFaceTB/SmolLM3-3B-Base")
pipe("Once upon a time,")
``` | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39271/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39271/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39270 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39270/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39270/comments | https://api.github.com/repos/huggingface/transformers/issues/39270/events | https://github.com/huggingface/transformers/pull/39270 | 3,211,727,330 | PR_kwDOCUB6oc6d5zTL | 39,270 | fix recompiles due to instance key, and deepcopy issues | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T09:18:28 | 2025-07-08T09:57:24 | 2025-07-08T09:38:11 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39270",
"html_url": "https://github.com/huggingface/transformers/pull/39270",
"diff_url": "https://github.com/huggingface/transformers/pull/39270.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39270.patch",
"merged_at": "2025-07-08T09:38:11"
} | # What does this PR do?
Deep copy and recompiles should be fixed | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39270/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39270/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39269 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39269/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39269/comments | https://api.github.com/repos/huggingface/transformers/issues/39269/events | https://github.com/huggingface/transformers/issues/39269 | 3,211,495,758 | I_kwDOCUB6oc6_a4lO | 39,269 | SigLip2 text pooler output selection | {
"login": "Jason3900",
"id": 55134225,
"node_id": "MDQ6VXNlcjU1MTM0MjI1",
"avatar_url": "https://avatars.githubusercontent.com/u/55134225?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Jason3900",
"html_url": "https://github.com/Jason3900",
"followers_url": "https://api.github.com/users/Jason3900/followers",
"following_url": "https://api.github.com/users/Jason3900/following{/other_user}",
"gists_url": "https://api.github.com/users/Jason3900/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Jason3900/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Jason3900/subscriptions",
"organizations_url": "https://api.github.com/users/Jason3900/orgs",
"repos_url": "https://api.github.com/users/Jason3900/repos",
"events_url": "https://api.github.com/users/Jason3900/events{/privacy}",
"received_events_url": "https://api.github.com/users/Jason3900/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
},
{
"id": 5769473378,
"node_id": "LA_kwDOCUB6oc8AAAABV-MtYg",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Vision",
"name": "Vision",
"color": "C079EF",
"default": false,
"description": ""
}
] | closed | false | null | [] | null | [] | 2025-07-08T08:08:02 | 2025-07-16T12:57:14 | 2025-07-16T12:57:13 | NONE | null | null | null | null | ### System Info
transformers==4.51.3
python==3.10.16
torch==2.7.0+cu128
### Who can help?
@amyeroberts @qubvel
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [x] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Hey. I'm finetuning a siglip2 model and find some issues.
From the original code from [google](https://github.com/google-research/big_vision/blob/0127fb6b337ee2a27bf4e54dea79cff176527356/big_vision/models/proj/image_text/text_transformer.py#L83), the text ids' last position is assumed to be eos token. While if I use transformers preprocessor following the doc, text whose token length is less than max_length will pad to max_length. In that case, the last token is not eos but pad. Since the [code](https://github.com/huggingface/transformers/blob/17b3c96c00cd8421bff85282aec32422bdfebd31/src/transformers/models/siglip2/modeling_siglip2.py#L692) always takes the last position to feed into the pool layer, I think there might be some mismatches?
```
# Assuming "sticky" EOS tokenization, last token is always EOS.
pooled_output = last_hidden_state[:, -1, :]
pooled_output = self.head(pooled_output)
```
And another question is the preprocessor's output doesn't contain attention_mask, should I manually build it according to padding positions or the original model doesn't use attention_mask for padding at all?
### Expected behavior
I want some explanation. | {
"login": "qubvel",
"id": 31920396,
"node_id": "MDQ6VXNlcjMxOTIwMzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/31920396?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qubvel",
"html_url": "https://github.com/qubvel",
"followers_url": "https://api.github.com/users/qubvel/followers",
"following_url": "https://api.github.com/users/qubvel/following{/other_user}",
"gists_url": "https://api.github.com/users/qubvel/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qubvel/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qubvel/subscriptions",
"organizations_url": "https://api.github.com/users/qubvel/orgs",
"repos_url": "https://api.github.com/users/qubvel/repos",
"events_url": "https://api.github.com/users/qubvel/events{/privacy}",
"received_events_url": "https://api.github.com/users/qubvel/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39269/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39269/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39268 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39268/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39268/comments | https://api.github.com/repos/huggingface/transformers/issues/39268/events | https://github.com/huggingface/transformers/pull/39268 | 3,211,303,778 | PR_kwDOCUB6oc6d4YY5 | 39,268 | π [i18n-KO] Translated quark.md to Korean | {
"login": "maximizemaxwell",
"id": 138701551,
"node_id": "U_kgDOCERq7w",
"avatar_url": "https://avatars.githubusercontent.com/u/138701551?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/maximizemaxwell",
"html_url": "https://github.com/maximizemaxwell",
"followers_url": "https://api.github.com/users/maximizemaxwell/followers",
"following_url": "https://api.github.com/users/maximizemaxwell/following{/other_user}",
"gists_url": "https://api.github.com/users/maximizemaxwell/gists{/gist_id}",
"starred_url": "https://api.github.com/users/maximizemaxwell/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/maximizemaxwell/subscriptions",
"organizations_url": "https://api.github.com/users/maximizemaxwell/orgs",
"repos_url": "https://api.github.com/users/maximizemaxwell/repos",
"events_url": "https://api.github.com/users/maximizemaxwell/events{/privacy}",
"received_events_url": "https://api.github.com/users/maximizemaxwell/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T07:02:34 | 2025-07-09T16:29:51 | 2025-07-09T16:29:51 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39268",
"html_url": "https://github.com/huggingface/transformers/pull/39268",
"diff_url": "https://github.com/huggingface/transformers/pull/39268.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39268.patch",
"merged_at": "2025-07-09T16:29:51"
} | # What does this PR do?
Translated the quark.md file of the documentation to Korean.
Thank you in advance for your review.
Part of https://github.com/huggingface/transformers/issues/20179
## Before reviewing
- [x] Check for missing / redundant translations (λ²μ λλ½/μ€λ³΅ κ²μ¬)
- [x] Grammar Check (λ§μΆ€λ² κ²μ¬)
- [x] Review or Add new terms to glossary (μ©μ΄ νμΈ λ° μΆκ°)
- [x] Check Inline TOC (e.g. [[lowercased-header]])
- [ ] Check live-preview for gotchas (live-previewλ‘ μ μμλ νμΈ)
## Who can review? (Initial)
@4N3MONE, @harheem @jungnerd @yijun-lee
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [x] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review? (Final)
@stevhliu May you please review this PR? | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39268/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39268/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39267 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39267/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39267/comments | https://api.github.com/repos/huggingface/transformers/issues/39267/events | https://github.com/huggingface/transformers/pull/39267 | 3,211,190,365 | PR_kwDOCUB6oc6d4ADC | 39,267 | Fix/timm version check gemma3n | {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T06:18:43 | 2025-07-11T11:41:29 | 2025-07-08T17:38:29 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39267",
"html_url": "https://github.com/huggingface/transformers/pull/39267",
"diff_url": "https://github.com/huggingface/transformers/pull/39267.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39267.patch",
"merged_at": null
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes # (issue)
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
Adds a version check in `Gemma3nForConditionalGeneration` to ensure `timm >= 0.9.16`.
This prevents users from accidentally using outdated `timm` versions that donβt support models like `mobilenetv5_300m_enc`, and raises a clear `ImportError` if the requirement isnβt met.
Fixes #39208
| {
"login": "sameerajashyam",
"id": 79054143,
"node_id": "MDQ6VXNlcjc5MDU0MTQz",
"avatar_url": "https://avatars.githubusercontent.com/u/79054143?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sameerajashyam",
"html_url": "https://github.com/sameerajashyam",
"followers_url": "https://api.github.com/users/sameerajashyam/followers",
"following_url": "https://api.github.com/users/sameerajashyam/following{/other_user}",
"gists_url": "https://api.github.com/users/sameerajashyam/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sameerajashyam/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sameerajashyam/subscriptions",
"organizations_url": "https://api.github.com/users/sameerajashyam/orgs",
"repos_url": "https://api.github.com/users/sameerajashyam/repos",
"events_url": "https://api.github.com/users/sameerajashyam/events{/privacy}",
"received_events_url": "https://api.github.com/users/sameerajashyam/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39267/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39267/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39266 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39266/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39266/comments | https://api.github.com/repos/huggingface/transformers/issues/39266/events | https://github.com/huggingface/transformers/issues/39266 | 3,211,068,048 | I_kwDOCUB6oc6_ZQKQ | 39,266 | Unable to create tensor, you should probably activate truncation and/or padding with 'padding=True' 'truncation=True' to have batched tensors with the same length. | {
"login": "mumu029",
"id": 51202355,
"node_id": "MDQ6VXNlcjUxMjAyMzU1",
"avatar_url": "https://avatars.githubusercontent.com/u/51202355?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mumu029",
"html_url": "https://github.com/mumu029",
"followers_url": "https://api.github.com/users/mumu029/followers",
"following_url": "https://api.github.com/users/mumu029/following{/other_user}",
"gists_url": "https://api.github.com/users/mumu029/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mumu029/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mumu029/subscriptions",
"organizations_url": "https://api.github.com/users/mumu029/orgs",
"repos_url": "https://api.github.com/users/mumu029/repos",
"events_url": "https://api.github.com/users/mumu029/events{/privacy}",
"received_events_url": "https://api.github.com/users/mumu029/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-08T05:19:35 | 2025-07-08T06:50:47 | 2025-07-08T06:50:47 | NONE | null | null | null | null | ### System Info
```bash
Traceback (most recent call last):
File "/home/cx/miniconda3/envs/demo/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 767, in convert_to_tensors
tensor = as_tensor(value)
File "/home/cx/miniconda3/envs/demo/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 729, in as_tensor
return torch.tensor(value)
ValueError: expected sequence of length 15757 at dim 1 (got 16242)
```
*DataCollatorForLanguageModeling* seems to only padding input ids and ignore labels, resulting in different lengths of labels in a batch. Why is this?
### Who can help?
_No response_
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
```python
def _process_fn(samples, tokenizer : PreTrainedTokenizerFast, config):
samples = [[{"role" : "user", "content" : x[0]}, {"role" : "assistant", "content" : x[1]}]
for x in zip(samples["input"], samples["output"])]
# tokenized_data = tokenizer.apply_chat_template(samples,
# return_tensors="pt",
# return_dict=True,
# padding="max_length",
# truncation=True,
# max_length=8000)
tokenized_data = tokenizer.apply_chat_template(samples,
return_tensors="pt",
return_dict=True,
padding=True
)
samples_ids = tokenized_data["input_ids"]
attention_mask = tokenized_data["attention_mask"]
output_ids = []
for i, seq in enumerate(samples_ids):
output_index = torch.where(seq == SPECIAL_GENERATE_TOKEN_ID)[0]
mask = attention_mask[i]
if len(output_index) == 1:
output_index = output_index[0].item()
else:
continue
temp = torch.full_like(seq, -100)
temp[output_index:] = seq[output_index:]
temp[mask == 0] = -100
output_ids.append(temp)
labels = torch.stack(output_ids)
return {"input_ids" : samples_ids,
"labels" : labels,
"attention_mask" : attention_mask}
trainer = Trainer(
model=peft_model,
args=train_config,
train_dataset=train_data,
eval_dataset=eval_data,
data_collator=DataCollatorForLanguageModeling(
tokenizer=tokenizer,
mlm=False,
pad_to_multiple_of=8 if torch.cuda.is_available() else None,
return_tensors="pt"
)
)
```
### Expected behavior
run code | {
"login": "mumu029",
"id": 51202355,
"node_id": "MDQ6VXNlcjUxMjAyMzU1",
"avatar_url": "https://avatars.githubusercontent.com/u/51202355?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mumu029",
"html_url": "https://github.com/mumu029",
"followers_url": "https://api.github.com/users/mumu029/followers",
"following_url": "https://api.github.com/users/mumu029/following{/other_user}",
"gists_url": "https://api.github.com/users/mumu029/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mumu029/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mumu029/subscriptions",
"organizations_url": "https://api.github.com/users/mumu029/orgs",
"repos_url": "https://api.github.com/users/mumu029/repos",
"events_url": "https://api.github.com/users/mumu029/events{/privacy}",
"received_events_url": "https://api.github.com/users/mumu029/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39266/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39266/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39265 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39265/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39265/comments | https://api.github.com/repos/huggingface/transformers/issues/39265/events | https://github.com/huggingface/transformers/pull/39265 | 3,210,866,744 | PR_kwDOCUB6oc6d26YX | 39,265 | Refactor label name handling for PEFT models in Trainer class | {
"login": "qgallouedec",
"id": 45557362,
"node_id": "MDQ6VXNlcjQ1NTU3MzYy",
"avatar_url": "https://avatars.githubusercontent.com/u/45557362?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qgallouedec",
"html_url": "https://github.com/qgallouedec",
"followers_url": "https://api.github.com/users/qgallouedec/followers",
"following_url": "https://api.github.com/users/qgallouedec/following{/other_user}",
"gists_url": "https://api.github.com/users/qgallouedec/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qgallouedec/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qgallouedec/subscriptions",
"organizations_url": "https://api.github.com/users/qgallouedec/orgs",
"repos_url": "https://api.github.com/users/qgallouedec/repos",
"events_url": "https://api.github.com/users/qgallouedec/events{/privacy}",
"received_events_url": "https://api.github.com/users/qgallouedec/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-08T03:13:28 | 2025-08-04T06:29:58 | 2025-08-04T06:29:57 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39265",
"html_url": "https://github.com/huggingface/transformers/pull/39265",
"diff_url": "https://github.com/huggingface/transformers/pull/39265.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39265.patch",
"merged_at": "2025-08-04T06:29:57"
} | # What does this PR do?
Fix https://github.com/huggingface/trl/issues/3694
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
@SunMarc
| {
"login": "SunMarc",
"id": 57196510,
"node_id": "MDQ6VXNlcjU3MTk2NTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/57196510?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SunMarc",
"html_url": "https://github.com/SunMarc",
"followers_url": "https://api.github.com/users/SunMarc/followers",
"following_url": "https://api.github.com/users/SunMarc/following{/other_user}",
"gists_url": "https://api.github.com/users/SunMarc/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SunMarc/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SunMarc/subscriptions",
"organizations_url": "https://api.github.com/users/SunMarc/orgs",
"repos_url": "https://api.github.com/users/SunMarc/repos",
"events_url": "https://api.github.com/users/SunMarc/events{/privacy}",
"received_events_url": "https://api.github.com/users/SunMarc/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39265/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39265/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39264 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39264/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39264/comments | https://api.github.com/repos/huggingface/transformers/issues/39264/events | https://github.com/huggingface/transformers/pull/39264 | 3,210,482,985 | PR_kwDOCUB6oc6d1osQ | 39,264 | Fix: Add version check for timm to support mobilenetv5 models (fixes #39208) | {
"login": "VIGNESH15103",
"id": 121597724,
"node_id": "U_kgDOBz9vHA",
"avatar_url": "https://avatars.githubusercontent.com/u/121597724?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/VIGNESH15103",
"html_url": "https://github.com/VIGNESH15103",
"followers_url": "https://api.github.com/users/VIGNESH15103/followers",
"following_url": "https://api.github.com/users/VIGNESH15103/following{/other_user}",
"gists_url": "https://api.github.com/users/VIGNESH15103/gists{/gist_id}",
"starred_url": "https://api.github.com/users/VIGNESH15103/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/VIGNESH15103/subscriptions",
"organizations_url": "https://api.github.com/users/VIGNESH15103/orgs",
"repos_url": "https://api.github.com/users/VIGNESH15103/repos",
"events_url": "https://api.github.com/users/VIGNESH15103/events{/privacy}",
"received_events_url": "https://api.github.com/users/VIGNESH15103/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-07T23:11:02 | 2025-07-14T11:15:17 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39264",
"html_url": "https://github.com/huggingface/transformers/pull/39264",
"diff_url": "https://github.com/huggingface/transformers/pull/39264.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39264.patch",
"merged_at": null
} | ### What does this PR do?
Fixes issue #39208: Adds a version check for `timm` to ensure support for models like `mobilenetv5_300m_enc`, which require `timm >= 0.9.10`.
### Motivation
The issue occurred because the `mobilenetv5_300m_enc` model name was not recognized in older versions of `timm`. This PR adds a conditional check to raise an informative `ImportError` if an unsupported version is detected.
### Changes
- Added a version check using `packaging.version` inside `TimmWrapperImageProcessor.__init__`.
- Raises an error if a `mobilenetv5` model is used with `timm < 0.9.10`.
### Issue Link
Closes #39208
### Reviewer Suggestion
@amyeroberts @qubvel (vision models)
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39264/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39264/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39263 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39263/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39263/comments | https://api.github.com/repos/huggingface/transformers/issues/39263/events | https://github.com/huggingface/transformers/pull/39263 | 3,210,409,046 | PR_kwDOCUB6oc6d1YYB | 39,263 | [Core] [Offloading] Enable saving offloaded models with multiple shared tensor groups | {
"login": "kylesayrs",
"id": 17103692,
"node_id": "MDQ6VXNlcjE3MTAzNjky",
"avatar_url": "https://avatars.githubusercontent.com/u/17103692?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kylesayrs",
"html_url": "https://github.com/kylesayrs",
"followers_url": "https://api.github.com/users/kylesayrs/followers",
"following_url": "https://api.github.com/users/kylesayrs/following{/other_user}",
"gists_url": "https://api.github.com/users/kylesayrs/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kylesayrs/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kylesayrs/subscriptions",
"organizations_url": "https://api.github.com/users/kylesayrs/orgs",
"repos_url": "https://api.github.com/users/kylesayrs/repos",
"events_url": "https://api.github.com/users/kylesayrs/events{/privacy}",
"received_events_url": "https://api.github.com/users/kylesayrs/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T22:32:29 | 2025-07-16T12:26:49 | 2025-07-10T16:33:30 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39263",
"html_url": "https://github.com/huggingface/transformers/pull/39263",
"diff_url": "https://github.com/huggingface/transformers/pull/39263.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39263.patch",
"merged_at": "2025-07-10T16:33:30"
} | ## Purpose ##
* Fixes https://github.com/huggingface/transformers/issues/39097
* Enables models with multiple groups of shared tensors to be saved
* For example, the token_embeds and lm_head weights are one group of shared tensors
* Right now, only one group can be shared at a time, as indicated by `tied_names = tied_params[0]`
* This enablement is useful for registering additional modules with shared tensors, such as quip/spinquant-like transforms
## Changes ##
* Instead of building a separate pathway for calculating shared tensors when offloaded, instead update `ptrs` calculation logic to be capable of handling meta tensors
* This is equivalent to the `find_tied_parameters` method, since in the meta device case, both check for equality of the python object itself
* Misc, update tied tensors runtime error to suggest using `_dynamic_tied_weights_keys`, which is also a valid way of saving tied tensors
## Testing ##
```python3
import torch
from transformers import AutoModelForCausalLM
device_map = {
"model.embed_tokens": "cuda:0",
"model.layers": "cpu",
"model.norm": "cpu",
"model.rotary_emb": "cpu",
"lm_head": "cuda:0"
}
model = AutoModelForCausalLM.from_pretrained("nm-testing/llama2.c-stories15M", device_map=device_map)
transform_a = torch.nn.Linear(1, 1, bias=False)
transform_a._dynamic_tied_weights_keys = ["weight"]
transform_b = torch.nn.Linear(1, 1, bias=False)
transform_b._dynamic_tied_weights_keys = ["weight"]
model.model.layers[0].self_attn.q_proj.register_module("transform", transform_a)
model.model.layers[1].self_attn.q_proj.register_module("transform", transform_a)
model.model.layers[2].self_attn.q_proj.register_module("transform", transform_b)
model.model.layers[3].self_attn.q_proj.register_module("transform", transform_b)
model.save_pretrained("tmp")
```
* Added test fails on main but succeeds with these changes | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39263/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39263/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39262 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39262/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39262/comments | https://api.github.com/repos/huggingface/transformers/issues/39262/events | https://github.com/huggingface/transformers/pull/39262 | 3,210,236,695 | PR_kwDOCUB6oc6d0zcy | 39,262 | Improve ELECTRA documentation with explanations and helpful links | {
"login": "PrakyathMC",
"id": 92665624,
"node_id": "U_kgDOBYX3GA",
"avatar_url": "https://avatars.githubusercontent.com/u/92665624?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/PrakyathMC",
"html_url": "https://github.com/PrakyathMC",
"followers_url": "https://api.github.com/users/PrakyathMC/followers",
"following_url": "https://api.github.com/users/PrakyathMC/following{/other_user}",
"gists_url": "https://api.github.com/users/PrakyathMC/gists{/gist_id}",
"starred_url": "https://api.github.com/users/PrakyathMC/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/PrakyathMC/subscriptions",
"organizations_url": "https://api.github.com/users/PrakyathMC/orgs",
"repos_url": "https://api.github.com/users/PrakyathMC/repos",
"events_url": "https://api.github.com/users/PrakyathMC/events{/privacy}",
"received_events_url": "https://api.github.com/users/PrakyathMC/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T20:57:58 | 2025-07-08T15:22:52 | 2025-07-08T15:22:52 | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39262",
"html_url": "https://github.com/huggingface/transformers/pull/39262",
"diff_url": "https://github.com/huggingface/transformers/pull/39262.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39262.patch",
"merged_at": null
} | ## What does this PR do?
Improves the ELECTRA model documentation following the standardized format outlined in #36979.
## Changes Made
- β
Added explanatory comments to Pipeline and AutoModel code examples
- β
Added "Quick Links" section with popular ELECTRA models on HuggingFace Hub
- β
Included additional resources (paper links, model collection)
- β
Improved accessibility for beginners learning to use ELECTRA
## Before and After
**Before**: Basic code examples without explanations
**After**: Commented code that teaches users what each line does + easy access to models
Fixes #36979 | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39262/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39262/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39261 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39261/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39261/comments | https://api.github.com/repos/huggingface/transformers/issues/39261/events | https://github.com/huggingface/transformers/pull/39261 | 3,210,147,900 | PR_kwDOCUB6oc6d0f0w | 39,261 | fix: providing a tensor to `cache_position` in `model.generate` kwargs always crashes because of boolean test | {
"login": "guicho271828",
"id": 1321968,
"node_id": "MDQ6VXNlcjEzMjE5Njg=",
"avatar_url": "https://avatars.githubusercontent.com/u/1321968?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/guicho271828",
"html_url": "https://github.com/guicho271828",
"followers_url": "https://api.github.com/users/guicho271828/followers",
"following_url": "https://api.github.com/users/guicho271828/following{/other_user}",
"gists_url": "https://api.github.com/users/guicho271828/gists{/gist_id}",
"starred_url": "https://api.github.com/users/guicho271828/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/guicho271828/subscriptions",
"organizations_url": "https://api.github.com/users/guicho271828/orgs",
"repos_url": "https://api.github.com/users/guicho271828/repos",
"events_url": "https://api.github.com/users/guicho271828/events{/privacy}",
"received_events_url": "https://api.github.com/users/guicho271828/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T20:19:39 | 2025-07-09T09:45:41 | 2025-07-09T09:41:28 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39261",
"html_url": "https://github.com/huggingface/transformers/pull/39261",
"diff_url": "https://github.com/huggingface/transformers/pull/39261.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39261.patch",
"merged_at": null
} | Currently, giving `cache_position` to `model.generate` is broken because providing any tensor value to it results in performing a boolean test, which causes a runtime error. This PR fixes it.
This is basically an oversight in the change commited in #37986 --- the code path is untested.
<!--
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
```
Traceback (most recent call last):
File "/home/masataro/test-kv.py", line 77, in <module>
outputs = model.generate(**inputs_trimmed, do_sample=False, max_new_tokens=256, past_key_values=past_key_values, cache_position=cache_position)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/masataro/miniforge3/envs/test/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/home/masataro/miniforge3/envs/test/lib/python3.12/site-packages/transformers/generation/utils.py", line 2623, in generate
result = self._sample(
^^^^^^^^^^^^^
File "/home/masataro/miniforge3/envs/test/lib/python3.12/site-packages/transformers/generation/utils.py", line 3568, in _sample
model_kwargs = self._get_initial_cache_position(cur_len, input_ids.device, model_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/masataro/miniforge3/envs/test/lib/python3.12/site-packages/transformers/generation/utils.py", line 1799, in _get_initial_cache_position
if "cache_position" in model_kwargs and model_kwargs["cache_position"]:
~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^
RuntimeError: Boolean value of Tensor with more than one value is ambiguous
```
sample code (test-kv.py):
```python
import torch
from transformers import AutoTokenizer,AutoModelForCausalLM
from transformers.cache_utils import (
DynamicCache,
)
model_id = "ibm-granite/granite-3.2-8b-instruct"
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map='auto')
tokenizer = AutoTokenizer.from_pretrained(model_id)
user_prompts = ["Hello, what's your name?",
"Btw, yesterday I was on a rock concert.",
"The song was 'we are the world'",
]
# with kv cache for all past context
past_key_values = DynamicCache()
last_output_length=0
messages = []
for prompt in user_prompts:
# outputs[0].shape[1]
# /~~~~~~~~~~~~~~~~~~~~~~~~~~\ output contains this string
# IIIIIIIIOOOOOOOOOOIIIIIIIIIOOOOOOOOOOOOOOOOOO
# | +-- input length |
# +--last output length + next output length
# \-----------------/
# completion
print(prompt)
messages.append({"role": "user", "content": prompt})
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt", return_dict=True).to(model.device)
input_length = inputs["input_ids"].shape[1]
inputs_trimmed={
'input_ids':inputs["input_ids"][:, last_output_length:],
'attention_mask':inputs["attention_mask"],
}
cache_position = torch.arange(last_output_length, input_length, dtype=torch.int64, device=model.device)
assert inputs_trimmed["input_ids"].shape[1] == cache_position.shape[0]
outputs = model.generate(**inputs_trimmed,
do_sample=False,
max_new_tokens=256,
use_cache=True,
past_key_values=past_key_values,
cache_position=cache_position)
completion = tokenizer.decode(outputs[0, input_length - last_output_length: ], skip_special_tokens=True)
print(completion)
messages.append({"role": "assistant", "content": completion})
last_output_length += outputs.shape[1]
```
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). --- *does not apply*
- [X] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case. --- *could not find any similar issue*
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation). --- *does not apply*
- [ ] Did you write any new necessary tests? --- *help needed.*
## Who can review?
@FremyCompany @ArthurZucker @zucchini-nlp @gante | {
"login": "gante",
"id": 12240844,
"node_id": "MDQ6VXNlcjEyMjQwODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/12240844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gante",
"html_url": "https://github.com/gante",
"followers_url": "https://api.github.com/users/gante/followers",
"following_url": "https://api.github.com/users/gante/following{/other_user}",
"gists_url": "https://api.github.com/users/gante/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gante/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gante/subscriptions",
"organizations_url": "https://api.github.com/users/gante/orgs",
"repos_url": "https://api.github.com/users/gante/repos",
"events_url": "https://api.github.com/users/gante/events{/privacy}",
"received_events_url": "https://api.github.com/users/gante/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39261/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39261/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39260 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39260/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39260/comments | https://api.github.com/repos/huggingface/transformers/issues/39260/events | https://github.com/huggingface/transformers/issues/39260 | 3,209,799,793 | I_kwDOCUB6oc6_Uahx | 39,260 | Please help i am trying to run model but issue | {
"login": "vinaydanidhariya",
"id": 66329591,
"node_id": "MDQ6VXNlcjY2MzI5NTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/66329591?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vinaydanidhariya",
"html_url": "https://github.com/vinaydanidhariya",
"followers_url": "https://api.github.com/users/vinaydanidhariya/followers",
"following_url": "https://api.github.com/users/vinaydanidhariya/following{/other_user}",
"gists_url": "https://api.github.com/users/vinaydanidhariya/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vinaydanidhariya/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vinaydanidhariya/subscriptions",
"organizations_url": "https://api.github.com/users/vinaydanidhariya/orgs",
"repos_url": "https://api.github.com/users/vinaydanidhariya/repos",
"events_url": "https://api.github.com/users/vinaydanidhariya/events{/privacy}",
"received_events_url": "https://api.github.com/users/vinaydanidhariya/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T17:40:47 | 2025-08-21T17:43:37 | 2025-08-17T08:02:47 | NONE | null | null | null | null | ```
# install libraries
# Uninstall broken stable PyTorch
!pip uninstall -y torch torchvision torchaudio
# Install latest nightly PyTorch compatible with Python 3.12 (for CPU)
!pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu
!pip install transformers==4.41.2 peft==0.11.1 librosa==0.10.2
!pip install triton
!pip install --upgrade accelerate
import transformers
import librosa
# load the model pipeline on gpu:0
# pipe = transformers.pipeline(model='sarvamai/shuka_v1', trust_remote_code=True, device=0, torch_dtype='bfloat16')
from transformers import pipeline
pipe = pipeline(
model='sarvamai/shuka_v1',
trust_remote_code=True,
device=-1 # force CPU to prevent Triton / GPU-related crashes
)
# get a sample audio
# wget https://huggingface.co/sarvamai/shuka_v1/resolve/main/hi-question.webm
audio, sr = librosa.load("./hi-question.webm", sr=16000)
turns = [
{'role': 'system', 'content': 'Respond naturally and informatively.'},
{'role': 'user', 'content': '<|audio|>'}
]
pipe({'audio': audio, 'turns': turns, 'sampling_rate': sr}, max_new_tokens=512)
```
Found existing installation: torch 2.9.0.dev20250707+cpu
Uninstalling torch-2.9.0.dev20250707+cpu:
Successfully uninstalled torch-2.9.0.dev20250707+cpu
Found existing installation: torchvision 0.24.0.dev20250707+cpu
Uninstalling torchvision-0.24.0.dev20250707+cpu:
Successfully uninstalled torchvision-0.24.0.dev20250707+cpu
Found existing installation: torchaudio 2.8.0.dev20250707+cpu
Uninstalling torchaudio-2.8.0.dev20250707+cpu:
Successfully uninstalled torchaudio-2.8.0.dev20250707+cpu
Looking in indexes: https://download.pytorch.org/whl/nightly/cpu
Collecting torch
Using cached https://download.pytorch.org/whl/nightly/cpu/torch-2.9.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl.metadata (28 kB)
Collecting torchvision
Using cached https://download.pytorch.org/whl/nightly/cpu/torchvision-0.24.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl.metadata (6.2 kB)
Collecting torchaudio
Using cached https://download.pytorch.org/whl/nightly/cpu/torchaudio-2.8.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl.metadata (7.3 kB)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from torch) (3.18.0)
Requirement already satisfied: typing-extensions>=4.10.0 in /usr/local/lib/python3.11/dist-packages (from torch) (4.13.2)
Requirement already satisfied: sympy>=1.13.3 in /usr/local/lib/python3.11/dist-packages (from torch) (1.14.0)
Requirement already satisfied: networkx in /usr/local/lib/python3.11/dist-packages (from torch) (3.4.2)
Requirement already satisfied: jinja2 in /usr/local/lib/python3.11/dist-packages (from torch) (3.1.6)
Requirement already satisfied: fsspec in /usr/local/lib/python3.11/dist-packages (from torch) (2025.3.2)
Requirement already satisfied: numpy in /usr/local/lib/python3.11/dist-packages (from torchvision) (1.26.4)
Requirement already satisfied: pillow!=8.3.*,<11.3,>=5.3.0 in /usr/local/lib/python3.11/dist-packages (from torchvision) (11.1.0)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from sympy>=1.13.3->torch) (1.3.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.11/dist-packages (from jinja2->torch) (3.0.2)
Requirement already satisfied: mkl_fft in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (1.3.8)
Requirement already satisfied: mkl_random in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (1.2.4)
Requirement already satisfied: mkl_umath in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (0.1.1)
Requirement already satisfied: mkl in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (2025.1.0)
Requirement already satisfied: tbb4py in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (2022.1.0)
Requirement already satisfied: mkl-service in /usr/local/lib/python3.11/dist-packages (from numpy->torchvision) (2.4.1)
Requirement already satisfied: intel-openmp<2026,>=2024 in /usr/local/lib/python3.11/dist-packages (from mkl->numpy->torchvision) (2024.2.0)
Requirement already satisfied: tbb==2022.* in /usr/local/lib/python3.11/dist-packages (from mkl->numpy->torchvision) (2022.1.0)
Requirement already satisfied: tcmlib==1.* in /usr/local/lib/python3.11/dist-packages (from tbb==2022.*->mkl->numpy->torchvision) (1.3.0)
Requirement already satisfied: intel-cmplr-lib-rt in /usr/local/lib/python3.11/dist-packages (from mkl_umath->numpy->torchvision) (2024.2.0)
Requirement already satisfied: intel-cmplr-lib-ur==2024.2.0 in /usr/local/lib/python3.11/dist-packages (from intel-openmp<2026,>=2024->mkl->numpy->torchvision) (2024.2.0)
Using cached https://download.pytorch.org/whl/nightly/cpu/torch-2.9.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl (184.4 MB)
Using cached https://download.pytorch.org/whl/nightly/cpu/torchvision-0.24.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl (2.0 MB)
Using cached https://download.pytorch.org/whl/nightly/cpu/torchaudio-2.8.0.dev20250707%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl (1.8 MB)
Installing collected packages: torch, torchaudio, torchvision
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
fastai 2.7.19 requires torch<2.7,>=1.10, but you have torch 2.9.0.dev20250707+cpu which is incompatible.
Successfully installed torch-2.9.0.dev20250707+cpu torchaudio-2.8.0.dev20250707+cpu torchvision-0.24.0.dev20250707+cpu
Requirement already satisfied: transformers==4.41.2 in /usr/local/lib/python3.11/dist-packages (4.41.2)
Requirement already satisfied: peft==0.11.1 in /usr/local/lib/python3.11/dist-packages (0.11.1)
Requirement already satisfied: librosa==0.10.2 in /usr/local/lib/python3.11/dist-packages (0.10.2)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (3.18.0)
Requirement already satisfied: huggingface-hub<1.0,>=0.23.0 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (0.31.1)
Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (1.26.4)
Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (25.0)
Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (6.0.2)
Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (2024.11.6)
Requirement already satisfied: requests in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (2.32.3)
Requirement already satisfied: tokenizers<0.20,>=0.19 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (0.19.1)
Requirement already satisfied: safetensors>=0.4.1 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (0.5.3)
Requirement already satisfied: tqdm>=4.27 in /usr/local/lib/python3.11/dist-packages (from transformers==4.41.2) (4.67.1)
Requirement already satisfied: psutil in /usr/local/lib/python3.11/dist-packages (from peft==0.11.1) (7.0.0)
Requirement already satisfied: torch>=1.13.0 in /usr/local/lib/python3.11/dist-packages (from peft==0.11.1) (2.9.0.dev20250707+cpu)
Requirement already satisfied: accelerate>=0.21.0 in /usr/local/lib/python3.11/dist-packages (from peft==0.11.1) (1.5.2)
Requirement already satisfied: audioread>=2.1.9 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (3.0.1)
Requirement already satisfied: scipy>=1.2.0 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (1.15.2)
Requirement already satisfied: scikit-learn>=0.20.0 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (1.2.2)
Requirement already satisfied: joblib>=0.14 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (1.5.0)
Requirement already satisfied: decorator>=4.3.0 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (4.4.2)
Requirement already satisfied: numba>=0.51.0 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (0.60.0)
Requirement already satisfied: soundfile>=0.12.1 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (0.13.1)
Requirement already satisfied: pooch>=1.1 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (1.8.2)
Requirement already satisfied: soxr>=0.3.2 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (0.5.0.post1)
Requirement already satisfied: typing-extensions>=4.1.1 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (4.13.2)
Requirement already satisfied: lazy-loader>=0.1 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (0.4)
Requirement already satisfied: msgpack>=1.0 in /usr/local/lib/python3.11/dist-packages (from librosa==0.10.2) (1.1.0)
Requirement already satisfied: fsspec>=2023.5.0 in /usr/local/lib/python3.11/dist-packages (from huggingface-hub<1.0,>=0.23.0->transformers==4.41.2) (2025.3.2)
Requirement already satisfied: hf-xet<2.0.0,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from huggingface-hub<1.0,>=0.23.0->transformers==4.41.2) (1.1.0)
Requirement already satisfied: llvmlite<0.44,>=0.43.0dev0 in /usr/local/lib/python3.11/dist-packages (from numba>=0.51.0->librosa==0.10.2) (0.43.0)
Requirement already satisfied: mkl_fft in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (1.3.8)
Requirement already satisfied: mkl_random in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (1.2.4)
Requirement already satisfied: mkl_umath in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (0.1.1)
Requirement already satisfied: mkl in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (2025.1.0)
Requirement already satisfied: tbb4py in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (2022.1.0)
Requirement already satisfied: mkl-service in /usr/local/lib/python3.11/dist-packages (from numpy>=1.17->transformers==4.41.2) (2.4.1)
Requirement already satisfied: platformdirs>=2.5.0 in /usr/local/lib/python3.11/dist-packages (from pooch>=1.1->librosa==0.10.2) (4.3.8)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.11/dist-packages (from requests->transformers==4.41.2) (3.4.2)
Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.11/dist-packages (from requests->transformers==4.41.2) (3.10)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.11/dist-packages (from requests->transformers==4.41.2) (2.4.0)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.11/dist-packages (from requests->transformers==4.41.2) (2025.4.26)
Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.11/dist-packages (from scikit-learn>=0.20.0->librosa==0.10.2) (3.6.0)
Requirement already satisfied: cffi>=1.0 in /usr/local/lib/python3.11/dist-packages (from soundfile>=0.12.1->librosa==0.10.2) (1.17.1)
Requirement already satisfied: sympy>=1.13.3 in /usr/local/lib/python3.11/dist-packages (from torch>=1.13.0->peft==0.11.1) (1.14.0)
Requirement already satisfied: networkx in /usr/local/lib/python3.11/dist-packages (from torch>=1.13.0->peft==0.11.1) (3.4.2)
Requirement already satisfied: jinja2 in /usr/local/lib/python3.11/dist-packages (from torch>=1.13.0->peft==0.11.1) (3.1.6)
Requirement already satisfied: pycparser in /usr/local/lib/python3.11/dist-packages (from cffi>=1.0->soundfile>=0.12.1->librosa==0.10.2) (2.22)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from sympy>=1.13.3->torch>=1.13.0->peft==0.11.1) (1.3.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.11/dist-packages (from jinja2->torch>=1.13.0->peft==0.11.1) (3.0.2)
Requirement already satisfied: intel-openmp<2026,>=2024 in /usr/local/lib/python3.11/dist-packages (from mkl->numpy>=1.17->transformers==4.41.2) (2024.2.0)
Requirement already satisfied: tbb==2022.* in /usr/local/lib/python3.11/dist-packages (from mkl->numpy>=1.17->transformers==4.41.2) (2022.1.0)
Requirement already satisfied: tcmlib==1.* in /usr/local/lib/python3.11/dist-packages (from tbb==2022.*->mkl->numpy>=1.17->transformers==4.41.2) (1.3.0)
Requirement already satisfied: intel-cmplr-lib-rt in /usr/local/lib/python3.11/dist-packages (from mkl_umath->numpy>=1.17->transformers==4.41.2) (2024.2.0)
Requirement already satisfied: intel-cmplr-lib-ur==2024.2.0 in /usr/local/lib/python3.11/dist-packages (from intel-openmp<2026,>=2024->mkl->numpy>=1.17->transformers==4.41.2) (2024.2.0)
Requirement already satisfied: triton in /usr/local/lib/python3.11/dist-packages (2.2.0)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from triton) (3.18.0)
Requirement already satisfied: accelerate in /usr/local/lib/python3.11/dist-packages (1.5.2)
Collecting accelerate
Downloading accelerate-1.8.1-py3-none-any.whl.metadata (19 kB)
Requirement already satisfied: numpy<3.0.0,>=1.17 in /usr/local/lib/python3.11/dist-packages (from accelerate) (1.26.4)
Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.11/dist-packages (from accelerate) (25.0)
Requirement already satisfied: psutil in /usr/local/lib/python3.11/dist-packages (from accelerate) (7.0.0)
Requirement already satisfied: pyyaml in /usr/local/lib/python3.11/dist-packages (from accelerate) (6.0.2)
Requirement already satisfied: torch>=2.0.0 in /usr/local/lib/python3.11/dist-packages (from accelerate) (2.9.0.dev20250707+cpu)
Requirement already satisfied: huggingface_hub>=0.21.0 in /usr/local/lib/python3.11/dist-packages (from accelerate) (0.31.1)
Requirement already satisfied: safetensors>=0.4.3 in /usr/local/lib/python3.11/dist-packages (from accelerate) (0.5.3)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (3.18.0)
Requirement already satisfied: fsspec>=2023.5.0 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (2025.3.2)
Requirement already satisfied: requests in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (2.32.3)
Requirement already satisfied: tqdm>=4.42.1 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (4.67.1)
Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (4.13.2)
Requirement already satisfied: hf-xet<2.0.0,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.21.0->accelerate) (1.1.0)
Requirement already satisfied: mkl_fft in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (1.3.8)
Requirement already satisfied: mkl_random in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (1.2.4)
Requirement already satisfied: mkl_umath in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (0.1.1)
Requirement already satisfied: mkl in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (2025.1.0)
Requirement already satisfied: tbb4py in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (2022.1.0)
Requirement already satisfied: mkl-service in /usr/local/lib/python3.11/dist-packages (from numpy<3.0.0,>=1.17->accelerate) (2.4.1)
Requirement already satisfied: sympy>=1.13.3 in /usr/local/lib/python3.11/dist-packages (from torch>=2.0.0->accelerate) (1.14.0)
Requirement already satisfied: networkx in /usr/local/lib/python3.11/dist-packages (from torch>=2.0.0->accelerate) (3.4.2)
Requirement already satisfied: jinja2 in /usr/local/lib/python3.11/dist-packages (from torch>=2.0.0->accelerate) (3.1.6)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from sympy>=1.13.3->torch>=2.0.0->accelerate) (1.3.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.11/dist-packages (from jinja2->torch>=2.0.0->accelerate) (3.0.2)
Requirement already satisfied: intel-openmp<2026,>=2024 in /usr/local/lib/python3.11/dist-packages (from mkl->numpy<3.0.0,>=1.17->accelerate) (2024.2.0)
Requirement already satisfied: tbb==2022.* in /usr/local/lib/python3.11/dist-packages (from mkl->numpy<3.0.0,>=1.17->accelerate) (2022.1.0)
Requirement already satisfied: tcmlib==1.* in /usr/local/lib/python3.11/dist-packages (from tbb==2022.*->mkl->numpy<3.0.0,>=1.17->accelerate) (1.3.0)
Requirement already satisfied: intel-cmplr-lib-rt in /usr/local/lib/python3.11/dist-packages (from mkl_umath->numpy<3.0.0,>=1.17->accelerate) (2024.2.0)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface_hub>=0.21.0->accelerate) (3.4.2)
Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface_hub>=0.21.0->accelerate) (3.10)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface_hub>=0.21.0->accelerate) (2.4.0)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface_hub>=0.21.0->accelerate) (2025.4.26)
Requirement already satisfied: intel-cmplr-lib-ur==2024.2.0 in /usr/local/lib/python3.11/dist-packages (from intel-openmp<2026,>=2024->mkl->numpy<3.0.0,>=1.17->accelerate) (2024.2.0)
Downloading accelerate-1.8.1-py3-none-any.whl (365 kB)
ββββββββββββββββββββββββββββββββββββββββ 365.3/365.3 kB 6.4 MB/s eta 0:00:00a 0:00:01
Installing collected packages: accelerate
Attempting uninstall: accelerate
Found existing installation: accelerate 1.5.2
Uninstalling accelerate-1.5.2:
Successfully uninstalled accelerate-1.5.2
Successfully installed accelerate-1.8.1
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/transformers/utils/import_utils.py in _get_module(self, module_name)
1534 try:
-> 1535 return importlib.import_module("." + module_name, self.__name__)
1536 except Exception as e:
/usr/lib/python3.11/importlib/__init__.py in import_module(name, package)
125 level += 1
--> 126 return _bootstrap._gcd_import(name[level:], package, level)
127
/usr/lib/python3.11/importlib/_bootstrap.py in _gcd_import(name, package, level)
/usr/lib/python3.11/importlib/_bootstrap.py in _find_and_load(name, import_)
/usr/lib/python3.11/importlib/_bootstrap.py in _find_and_load_unlocked(name, import_)
/usr/lib/python3.11/importlib/_bootstrap.py in _load_unlocked(spec)
/usr/lib/python3.11/importlib/_bootstrap_external.py in exec_module(self, module)
/usr/lib/python3.11/importlib/_bootstrap.py in _call_with_frames_removed(f, *args, **kwds)
/usr/local/lib/python3.11/dist-packages/transformers/pipelines/__init__.py in <module>
46 )
---> 47 from .audio_classification import AudioClassificationPipeline
48 from .automatic_speech_recognition import AutomaticSpeechRecognitionPipeline
/usr/local/lib/python3.11/dist-packages/transformers/pipelines/audio_classification.py in <module>
20 from ..utils import add_end_docstrings, is_torch_available, is_torchaudio_available, logging
---> 21 from .base import Pipeline, build_pipeline_init_args
22
/usr/local/lib/python3.11/dist-packages/transformers/pipelines/base.py in <module>
33 from ..image_processing_utils import BaseImageProcessor
---> 34 from ..modelcard import ModelCard
35 from ..models.auto.configuration_auto import AutoConfig
/usr/local/lib/python3.11/dist-packages/transformers/modelcard.py in <module>
47 )
---> 48 from .training_args import ParallelMode
49 from .utils import (
/usr/local/lib/python3.11/dist-packages/transformers/training_args.py in <module>
72 if is_accelerate_available():
---> 73 from accelerate.state import AcceleratorState, PartialState
74 from accelerate.utils import DistributedType
/usr/local/lib/python3.11/dist-packages/accelerate/__init__.py in <module>
15
---> 16 from .accelerator import Accelerator
17 from .big_modeling import (
/usr/local/lib/python3.11/dist-packages/accelerate/accelerator.py in <module>
35
---> 36 from .checkpointing import load_accelerator_state, load_custom_state, save_accelerator_state, save_custom_state
37 from .data_loader import DataLoaderDispatcher, prepare_data_loader, skip_first_batches
/usr/local/lib/python3.11/dist-packages/accelerate/checkpointing.py in <module>
21
---> 22 from .utils import (
23 MODEL_NAME,
/usr/local/lib/python3.11/dist-packages/accelerate/utils/__init__.py in <module>
34 )
---> 35 from .dataclasses import (
36 AORecipeKwargs,
/usr/local/lib/python3.11/dist-packages/accelerate/utils/dataclasses.py in <module>
33
---> 34 from .constants import (
35 BETA_TP_AVAILABLE_PYTORCH_VERSION,
ImportError: cannot import name 'FSDP2_PYTORCH_VERSION' from 'accelerate.utils.constants' (/usr/local/lib/python3.11/dist-packages/accelerate/utils/constants.py)
The above exception was the direct cause of the following exception:
RuntimeError Traceback (most recent call last)
/tmp/ipykernel_129/2705754419.py in <cell line: 0>()
15 # load the model pipeline on gpu:0
16 # pipe = transformers.pipeline(model='sarvamai/shuka_v1', trust_remote_code=True, device=0, torch_dtype='bfloat16')
---> 17 from transformers import pipeline
18
19 pipe = pipeline(
/usr/lib/python3.11/importlib/_bootstrap.py in _handle_fromlist(module, fromlist, import_, recursive)
/usr/local/lib/python3.11/dist-packages/transformers/utils/import_utils.py in __getattr__(self, name)
1523 value = self._get_module(name)
1524 elif name in self._class_to_module.keys():
-> 1525 module = self._get_module(self._class_to_module[name])
1526 value = getattr(module, name)
1527 else:
/usr/local/lib/python3.11/dist-packages/transformers/utils/import_utils.py in _get_module(self, module_name)
1535 return importlib.import_module("." + module_name, self.__name__)
1536 except Exception as e:
-> 1537 raise RuntimeError(
1538 f"Failed to import {self.__name__}.{module_name} because of the following error (look up to see its"
1539 f" traceback):\n{e}"
RuntimeError: Failed to import transformers.pipelines because of the following error (look up to see its traceback):
cannot import name 'FSDP2_PYTORCH_VERSION' from 'accelerate.utils.constants' (/usr/local/lib/python3.11/dist-packages/accelerate/utils/constants.py) | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39260/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39260/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39259 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39259/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39259/comments | https://api.github.com/repos/huggingface/transformers/issues/39259/events | https://github.com/huggingface/transformers/pull/39259 | 3,209,640,901 | PR_kwDOCUB6oc6dyzOY | 39,259 | fix failing `test_sdpa_can_dispatch_on_flash` | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T16:34:01 | 2025-07-13T11:25:06 | 2025-07-11T14:30:56 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39259",
"html_url": "https://github.com/huggingface/transformers/pull/39259",
"diff_url": "https://github.com/huggingface/transformers/pull/39259.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39259.patch",
"merged_at": "2025-07-11T14:30:56"
} | # What does this PR do?
| {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39259/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39259/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39258 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39258/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39258/comments | https://api.github.com/repos/huggingface/transformers/issues/39258/events | https://github.com/huggingface/transformers/pull/39258 | 3,209,588,658 | PR_kwDOCUB6oc6dynhS | 39,258 | [pagged-attention] fix off-by-1 error in pagged attention generation | {
"login": "kashif",
"id": 8100,
"node_id": "MDQ6VXNlcjgxMDA=",
"avatar_url": "https://avatars.githubusercontent.com/u/8100?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kashif",
"html_url": "https://github.com/kashif",
"followers_url": "https://api.github.com/users/kashif/followers",
"following_url": "https://api.github.com/users/kashif/following{/other_user}",
"gists_url": "https://api.github.com/users/kashif/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kashif/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kashif/subscriptions",
"organizations_url": "https://api.github.com/users/kashif/orgs",
"repos_url": "https://api.github.com/users/kashif/repos",
"events_url": "https://api.github.com/users/kashif/events{/privacy}",
"received_events_url": "https://api.github.com/users/kashif/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 8103865784,
"node_id": "LA_kwDOCUB6oc8AAAAB4wctuA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/for%20patch",
"name": "for patch",
"color": "D93F0B",
"default": false,
"description": "Tag issues / labels that should be included in the next patch"
}
] | closed | false | null | [] | null | [] | 2025-07-07T16:18:35 | 2025-07-11T10:49:56 | 2025-07-08T10:34:23 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39258",
"html_url": "https://github.com/huggingface/transformers/pull/39258",
"diff_url": "https://github.com/huggingface/transformers/pull/39258.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39258.patch",
"merged_at": "2025-07-08T10:34:23"
} | # What does this PR do?
we were adding the token first and then doing a check, leading to the number of tokens being 1 more than the`max_new_tokens`
This fix reorders the logic to:
1. Check completion conditions (EOS token or max length) before adding the token
1. Only add the token if we haven't exceeded the limit (with special handling for EOS tokens)
1. Terminate the request appropriately
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "kashif",
"id": 8100,
"node_id": "MDQ6VXNlcjgxMDA=",
"avatar_url": "https://avatars.githubusercontent.com/u/8100?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kashif",
"html_url": "https://github.com/kashif",
"followers_url": "https://api.github.com/users/kashif/followers",
"following_url": "https://api.github.com/users/kashif/following{/other_user}",
"gists_url": "https://api.github.com/users/kashif/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kashif/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kashif/subscriptions",
"organizations_url": "https://api.github.com/users/kashif/orgs",
"repos_url": "https://api.github.com/users/kashif/repos",
"events_url": "https://api.github.com/users/kashif/events{/privacy}",
"received_events_url": "https://api.github.com/users/kashif/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39258/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39258/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39257 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39257/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39257/comments | https://api.github.com/repos/huggingface/transformers/issues/39257/events | https://github.com/huggingface/transformers/pull/39257 | 3,209,565,516 | PR_kwDOCUB6oc6dyimt | 39,257 | Fix to tuple conversion with config | {
"login": "qubvel",
"id": 31920396,
"node_id": "MDQ6VXNlcjMxOTIwMzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/31920396?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qubvel",
"html_url": "https://github.com/qubvel",
"followers_url": "https://api.github.com/users/qubvel/followers",
"following_url": "https://api.github.com/users/qubvel/following{/other_user}",
"gists_url": "https://api.github.com/users/qubvel/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qubvel/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qubvel/subscriptions",
"organizations_url": "https://api.github.com/users/qubvel/orgs",
"repos_url": "https://api.github.com/users/qubvel/repos",
"events_url": "https://api.github.com/users/qubvel/events{/privacy}",
"received_events_url": "https://api.github.com/users/qubvel/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-07T16:10:43 | 2025-07-07T16:23:42 | null | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39257",
"html_url": "https://github.com/huggingface/transformers/pull/39257",
"diff_url": "https://github.com/huggingface/transformers/pull/39257.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39257.patch",
"merged_at": null
} | # What does this PR do?
setting `return_dict=False` with config fails for models with sub-models wrapped with `can_return_tuple` or `check_model_inputs`
```py
import torch
from transformers import LlamaConfig, LlamaForCausalLM
config = LlamaConfig(vocab_size=256, hidden_size=128, num_hidden_layers=2, num_attention_heads=4, intermediate_size=256)
model = LlamaForCausalLM(config)
# default: ModelOutput
input_ids = torch.tensor([[0, 1, 2, 3]])
with torch.no_grad():
output = model(input_ids)
print(output)
# passing return_dict=False as a kwarg
input_ids = torch.tensor([[0, 1, 2, 3]])
with torch.no_grad():
output = model(input_ids, return_dict=False)
print(output)
# ERROR: setting return_dict=False in the config
model.config.return_dict = False
with torch.no_grad():
output = model(input_ids)
print(output)
# Traceback (most recent call last):
# File "/home/ubuntu/projects/transformers/test_llama_small.py", line 17, in <module>
# output = model(input_ids)
# File "/home/ubuntu/projects/transformers/.venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
# return self._call_impl(*args, **kwargs)
# File "/home/ubuntu/projects/transformers/.venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1762, in _call_impl
# return forward_call(*args, **kwargs)
# File "/home/ubuntu/projects/transformers/src/transformers/utils/generic.py", line 962, in wrapper
# output = func(self, *args, **kwargs)
# File "/home/ubuntu/projects/transformers/src/transformers/models/llama/modeling_llama.py", line 506, in forward
# hidden_states = outputs.last_hidden_state
# AttributeError: 'tuple' object has no attribute 'last_hidden_state'
```
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39257/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39257/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39256 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39256/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39256/comments | https://api.github.com/repos/huggingface/transformers/issues/39256/events | https://github.com/huggingface/transformers/issues/39256 | 3,209,435,987 | I_kwDOCUB6oc6_TBtT | 39,256 | ModuleNotFoundError: No module named 'habana_frameworks.torch' | {
"login": "orkungedik",
"id": 32071110,
"node_id": "MDQ6VXNlcjMyMDcxMTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/32071110?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/orkungedik",
"html_url": "https://github.com/orkungedik",
"followers_url": "https://api.github.com/users/orkungedik/followers",
"following_url": "https://api.github.com/users/orkungedik/following{/other_user}",
"gists_url": "https://api.github.com/users/orkungedik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/orkungedik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/orkungedik/subscriptions",
"organizations_url": "https://api.github.com/users/orkungedik/orgs",
"repos_url": "https://api.github.com/users/orkungedik/repos",
"events_url": "https://api.github.com/users/orkungedik/events{/privacy}",
"received_events_url": "https://api.github.com/users/orkungedik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-07T15:25:41 | 2025-07-21T17:11:43 | 2025-07-07T19:09:48 | NONE | null | null | null | null | ### System Info
Hi,
I finetuned unsloth/Meta-Llama-3.1-8B-Instruct model and tried to run code below on CPU;
```
model_name = "orkungedik/ege-8b-1.0"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name,torch_dtype="auto", device_map="auto")
```
And getting error below;
> File /usr/local/lib/python3.10/dist-packages/neural_compressor/torch/utils/environ.py:67
65 if is_package_available("habana_frameworks"):
66 _hpex_available = True
---> 67 import habana_frameworks.torch.hpex # pylint: disable=E0401
68 else:
69 _hpex_available = False
ModuleNotFoundError: No module named 'habana_frameworks.torch'
To solve this problem executed commands below;
pip install --upgrade neural-compressor
pip install habana-tensorflow
But nothing changed.
What may cause this problem?
Thank you
### Who can help?
_No response_
### Information
- [ ] The official example scripts
- [ ] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Execute code on CPU based system and run provided source code
### Expected behavior
Loading model without a problem | {
"login": "orkungedik",
"id": 32071110,
"node_id": "MDQ6VXNlcjMyMDcxMTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/32071110?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/orkungedik",
"html_url": "https://github.com/orkungedik",
"followers_url": "https://api.github.com/users/orkungedik/followers",
"following_url": "https://api.github.com/users/orkungedik/following{/other_user}",
"gists_url": "https://api.github.com/users/orkungedik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/orkungedik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/orkungedik/subscriptions",
"organizations_url": "https://api.github.com/users/orkungedik/orgs",
"repos_url": "https://api.github.com/users/orkungedik/repos",
"events_url": "https://api.github.com/users/orkungedik/events{/privacy}",
"received_events_url": "https://api.github.com/users/orkungedik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39256/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39256/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39255 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39255/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39255/comments | https://api.github.com/repos/huggingface/transformers/issues/39255/events | https://github.com/huggingface/transformers/pull/39255 | 3,209,387,449 | PR_kwDOCUB6oc6dx8bD | 39,255 | remove broken block | {
"login": "molbap",
"id": 39954772,
"node_id": "MDQ6VXNlcjM5OTU0Nzcy",
"avatar_url": "https://avatars.githubusercontent.com/u/39954772?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/molbap",
"html_url": "https://github.com/molbap",
"followers_url": "https://api.github.com/users/molbap/followers",
"following_url": "https://api.github.com/users/molbap/following{/other_user}",
"gists_url": "https://api.github.com/users/molbap/gists{/gist_id}",
"starred_url": "https://api.github.com/users/molbap/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/molbap/subscriptions",
"organizations_url": "https://api.github.com/users/molbap/orgs",
"repos_url": "https://api.github.com/users/molbap/repos",
"events_url": "https://api.github.com/users/molbap/events{/privacy}",
"received_events_url": "https://api.github.com/users/molbap/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T15:09:24 | 2025-07-08T08:41:46 | 2025-07-08T08:41:44 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39255",
"html_url": "https://github.com/huggingface/transformers/pull/39255",
"diff_url": "https://github.com/huggingface/transformers/pull/39255.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39255.patch",
"merged_at": "2025-07-08T08:41:44"
} | # What does this PR do?
Removes an unused (and broken) block in modular model converter. Tested on a few modular conversions, as long as we don't use `return super().<...method...>` it will not cause issues.
| {
"login": "molbap",
"id": 39954772,
"node_id": "MDQ6VXNlcjM5OTU0Nzcy",
"avatar_url": "https://avatars.githubusercontent.com/u/39954772?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/molbap",
"html_url": "https://github.com/molbap",
"followers_url": "https://api.github.com/users/molbap/followers",
"following_url": "https://api.github.com/users/molbap/following{/other_user}",
"gists_url": "https://api.github.com/users/molbap/gists{/gist_id}",
"starred_url": "https://api.github.com/users/molbap/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/molbap/subscriptions",
"organizations_url": "https://api.github.com/users/molbap/orgs",
"repos_url": "https://api.github.com/users/molbap/repos",
"events_url": "https://api.github.com/users/molbap/events{/privacy}",
"received_events_url": "https://api.github.com/users/molbap/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39255/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39255/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39254 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39254/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39254/comments | https://api.github.com/repos/huggingface/transformers/issues/39254/events | https://github.com/huggingface/transformers/pull/39254 | 3,209,375,893 | PR_kwDOCUB6oc6dx5-L | 39,254 | fix some flaky tests in `tests/generation/test_utils.py` | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T15:05:24 | 2025-07-07T17:49:44 | 2025-07-07T17:49:42 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39254",
"html_url": "https://github.com/huggingface/transformers/pull/39254",
"diff_url": "https://github.com/huggingface/transformers/pull/39254.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39254.patch",
"merged_at": "2025-07-07T17:49:42"
} | # What does this PR do?
Use the helper functions to set the norm layers to have eps=1.0 to avoid numerical issues.
Now the following pass
> RUN_SLOW=1 python3 -m pytest -v -n 6 --flake-finder --flake-runs=500 tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_eager_matches_sdpa_generate
> RUN_SLOW=1 python3 -m pytest -v -n 6 --flake-finder --flake-runs=500 tests/models/glm4v/test_modeling_glm4v.py::Glm4vModelTest -k "test_generate_from_inputs_embeds"
The following
> RUN_SLOW=1 python3 -m pytest -v -n 6 --flake-finder --flake-runs=100 tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_eager_matches_sdpa_generate
pass, while on `main` it has 22 failures.
> RUN_SLOW=1 python3 -m pytest -v -n 6 --flake-finder --flake-runs=500 tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_eager_matches_sdpa_generate
still has 1 failure. | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39254/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39254/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39253 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39253/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39253/comments | https://api.github.com/repos/huggingface/transformers/issues/39253/events | https://github.com/huggingface/transformers/pull/39253 | 3,209,324,321 | PR_kwDOCUB6oc6dxu9K | 39,253 | Add trust_remote_code in LightGlueConfig | {
"login": "sbucaille",
"id": 24275548,
"node_id": "MDQ6VXNlcjI0Mjc1NTQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/24275548?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sbucaille",
"html_url": "https://github.com/sbucaille",
"followers_url": "https://api.github.com/users/sbucaille/followers",
"following_url": "https://api.github.com/users/sbucaille/following{/other_user}",
"gists_url": "https://api.github.com/users/sbucaille/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sbucaille/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sbucaille/subscriptions",
"organizations_url": "https://api.github.com/users/sbucaille/orgs",
"repos_url": "https://api.github.com/users/sbucaille/repos",
"events_url": "https://api.github.com/users/sbucaille/events{/privacy}",
"received_events_url": "https://api.github.com/users/sbucaille/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 5769473378,
"node_id": "LA_kwDOCUB6oc8AAAABV-MtYg",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Vision",
"name": "Vision",
"color": "C079EF",
"default": false,
"description": ""
}
] | closed | false | null | [] | null | [] | 2025-07-07T14:48:58 | 2025-07-08T15:12:11 | 2025-07-08T15:03:05 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39253",
"html_url": "https://github.com/huggingface/transformers/pull/39253",
"diff_url": "https://github.com/huggingface/transformers/pull/39253.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39253.patch",
"merged_at": "2025-07-08T15:03:04"
} | # What does this PR do?
Add trust_remote_code parameter to LightGlue config so that LightGlue can be used with other not implemented keypoint detectors than SuperPoint.
With `from_pretrained`, `trust_remote_code` is not provided to the config, so I had to include it into the config itself as an attribute.
Discussed on Slack.
## Who can review?
@qubvel
| {
"login": "qubvel",
"id": 31920396,
"node_id": "MDQ6VXNlcjMxOTIwMzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/31920396?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/qubvel",
"html_url": "https://github.com/qubvel",
"followers_url": "https://api.github.com/users/qubvel/followers",
"following_url": "https://api.github.com/users/qubvel/following{/other_user}",
"gists_url": "https://api.github.com/users/qubvel/gists{/gist_id}",
"starred_url": "https://api.github.com/users/qubvel/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/qubvel/subscriptions",
"organizations_url": "https://api.github.com/users/qubvel/orgs",
"repos_url": "https://api.github.com/users/qubvel/repos",
"events_url": "https://api.github.com/users/qubvel/events{/privacy}",
"received_events_url": "https://api.github.com/users/qubvel/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39253/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39253/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39252 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39252/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39252/comments | https://api.github.com/repos/huggingface/transformers/issues/39252/events | https://github.com/huggingface/transformers/pull/39252 | 3,209,284,560 | PR_kwDOCUB6oc6dxmaI | 39,252 | Simplify Mixtral and its modular children | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T14:36:13 | 2025-07-07T17:40:42 | 2025-07-07T17:40:41 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39252",
"html_url": "https://github.com/huggingface/transformers/pull/39252",
"diff_url": "https://github.com/huggingface/transformers/pull/39252.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39252.patch",
"merged_at": "2025-07-07T17:40:41"
} | # What does this PR do?
As per the title. This is useful for modular with models to be released | {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39252/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39252/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39251 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39251/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39251/comments | https://api.github.com/repos/huggingface/transformers/issues/39251/events | https://github.com/huggingface/transformers/pull/39251 | 3,209,144,080 | PR_kwDOCUB6oc6dxIFn | 39,251 | Fix slow test_moshika_greedy_unconditional_fp16 | {
"login": "manueldeprada",
"id": 6536835,
"node_id": "MDQ6VXNlcjY1MzY4MzU=",
"avatar_url": "https://avatars.githubusercontent.com/u/6536835?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/manueldeprada",
"html_url": "https://github.com/manueldeprada",
"followers_url": "https://api.github.com/users/manueldeprada/followers",
"following_url": "https://api.github.com/users/manueldeprada/following{/other_user}",
"gists_url": "https://api.github.com/users/manueldeprada/gists{/gist_id}",
"starred_url": "https://api.github.com/users/manueldeprada/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/manueldeprada/subscriptions",
"organizations_url": "https://api.github.com/users/manueldeprada/orgs",
"repos_url": "https://api.github.com/users/manueldeprada/repos",
"events_url": "https://api.github.com/users/manueldeprada/events{/privacy}",
"received_events_url": "https://api.github.com/users/manueldeprada/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-07T13:55:15 | 2025-09-02T14:34:38 | null | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39251",
"html_url": "https://github.com/huggingface/transformers/pull/39251",
"diff_url": "https://github.com/huggingface/transformers/pull/39251.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39251.patch",
"merged_at": null
} | Fix #38725
Coming from #38725, previously, e18f233f6c8cba029324e2868fb68abdaf6badf3 attempted to fix the default attention mask issue that appeated with #34464, but it was still failing the slow test `tests/models/moshi/test_modeling_moshi.py::MoshiIntegrationTests::test_moshika_greedy_unconditional_fp16`
History from git bisect:
- https://github.com/huggingface/transformers/blob/84a6789145c3d728f2e405d31e9a35df5d74f05c/src/transformers/modeling_utils.py#L1413 from #34873 broke the types for some commits, then it was fixed in a later commit.
- auto-compilation on generate also made it fail for some commits, it was also corrected after some commits.
- Then #35993 broke here the test (didnt have the -1):
https://github.com/huggingface/transformers/blob/36bf1d2324ee8a971763eabc2950a044d5eff3fa/src/transformers/generation/utils.py#L2090
- And #37873 modified sliding window behaviour breaking it as well (>= to >)
https://github.com/huggingface/transformers/blob/1b222903c3e1cfd9492d75e4b2548aa8bd458674/src/transformers/cache_utils.py#L1740
Setting cache_implementation="dynamic" makes the test pass, but the sliding window cache should not behave different. I believe this is due to the depth decoder being window 8 by default, but audio is confusing to me.
This PR is not a fix: the modeling code should be changed to accomodate what I highlight in the diff
cc @eustlb @ydshieh | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39251/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 1,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39251/timeline | null | null | null | null | true | false |
https://api.github.com/repos/huggingface/transformers/issues/39250 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39250/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39250/comments | https://api.github.com/repos/huggingface/transformers/issues/39250/events | https://github.com/huggingface/transformers/pull/39250 | 3,209,108,210 | PR_kwDOCUB6oc6dxAQs | 39,250 | Fix license text, duplicate assignment, and typo in constant names | {
"login": "gudwls215",
"id": 96422648,
"node_id": "U_kgDOBb9K-A",
"avatar_url": "https://avatars.githubusercontent.com/u/96422648?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gudwls215",
"html_url": "https://github.com/gudwls215",
"followers_url": "https://api.github.com/users/gudwls215/followers",
"following_url": "https://api.github.com/users/gudwls215/following{/other_user}",
"gists_url": "https://api.github.com/users/gudwls215/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gudwls215/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gudwls215/subscriptions",
"organizations_url": "https://api.github.com/users/gudwls215/orgs",
"repos_url": "https://api.github.com/users/gudwls215/repos",
"events_url": "https://api.github.com/users/gudwls215/events{/privacy}",
"received_events_url": "https://api.github.com/users/gudwls215/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T13:44:57 | 2025-07-08T08:20:52 | 2025-07-08T08:20:52 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39250",
"html_url": "https://github.com/huggingface/transformers/pull/39250",
"diff_url": "https://github.com/huggingface/transformers/pull/39250.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39250.patch",
"merged_at": "2025-07-08T08:20:52"
} | # What does this PR do?
This PR fixes several minor but important issues across the codebase:
1. **License text completion** in Italian documentation
2. **Code syntax error** in Perceiver model conversion script
3. **Typo in metadata constant** name
## Changes Made
### π Documentation Fix
- **File**: `docs/source/it/perf_train_special.md`
- **Issue**: Incomplete Apache License 2.0 text (missing ending)
- **Fix**: Added the missing license text ending
### π Code Syntax Fix
- **File**: `src/transformers/models/perceiver/convert_perceiver_haiku_to_pytorch.py`
- **Issue**: Duplicate assignment `name = name = name.replace(...)`
- **Fix**: Removed duplicate assignment to `name = name.replace(...)`
### π§ Metadata Constant Fix
- **File**: `utils/update_metadata.py`
- **Issue**: Typo in constant name `"MODEL_FOR_FOR_VISION_2_SEQ_MAPPING_NAMES"` (duplicate "FOR")
- **Fix**: Corrected to `"MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES"`
## Testing
- All changes are minor text/syntax fixes
- No functional changes to model behavior
- Verified syntax correctness in Python files
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request), Pull Request section?
- [x] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link to it if that's the case. (N/A - minor fixes)
- [x] Did you make sure to update the documentation with your changes? (Documentation fix included)
- [x] Did you write any new necessary tests? (N/A - no functional changes)
## Who can review?
@stevhliu (Documentation fixes) or any community member - these are minor text/syntax corrections. | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39250/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39250/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39249 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39249/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39249/comments | https://api.github.com/repos/huggingface/transformers/issues/39249/events | https://github.com/huggingface/transformers/issues/39249 | 3,208,384,440 | I_kwDOCUB6oc6_PA-4 | 39,249 | OSError: Tensor parallel is only supported for `torch>=2.5` | {
"login": "mxjmtxrm",
"id": 167944756,
"node_id": "U_kgDOCgKiNA",
"avatar_url": "https://avatars.githubusercontent.com/u/167944756?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mxjmtxrm",
"html_url": "https://github.com/mxjmtxrm",
"followers_url": "https://api.github.com/users/mxjmtxrm/followers",
"following_url": "https://api.github.com/users/mxjmtxrm/following{/other_user}",
"gists_url": "https://api.github.com/users/mxjmtxrm/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mxjmtxrm/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mxjmtxrm/subscriptions",
"organizations_url": "https://api.github.com/users/mxjmtxrm/orgs",
"repos_url": "https://api.github.com/users/mxjmtxrm/repos",
"events_url": "https://api.github.com/users/mxjmtxrm/events{/privacy}",
"received_events_url": "https://api.github.com/users/mxjmtxrm/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-07T10:16:00 | 2025-07-08T11:37:36 | 2025-07-08T11:37:36 | NONE | null | null | null | null | ### System Info
- `transformers` version: 4.53.1
- Platform: Linux-5.15.0-105-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- Huggingface_hub version: 0.33.2
- Safetensors version: 0.4.5
- Accelerate version: 1.2.1
- Accelerate config: not found
- DeepSpeed version: 0.17.1
- PyTorch version (accelerator?): 2.5.0a0+e000cf0ad9.nv24.10 (CUDA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: <fill in>
- Using GPU in script?: <fill in>
- GPU type: NVIDIA H100 80GB HBM3
### Who can help?
@SunMarc @zach-huggingface
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
Hi, I met the following error when I tried to finetune a model using run_clm.py.
```
Traceback (most recent call last):
File "/workspace/run_clm.py", line 679, in <module>
main()
File "/workspace/run_clm.py", line 235, in main
model = AutoModelForCausalLM.from_pretrained(
File "/usr/local/lib/python3.10/dist-packages/transformers/models/auto/auto_factory.py", line 593, in from_pretrained
return model_class.from_pretrained(
File "/usr/local/lib/python3.10/dist-packages/transformers/modeling_utils.py", line 311, in _wrapper
return func(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/transformers/modeling_utils.py", line 4443, in from_pretrained
tp_plan, device_map, device_mesh = initialize_tensor_parallelism(tp_plan, tp_size=None)
File "/usr/local/lib/python3.10/dist-packages/transformers/integrations/tensor_parallel.py", line 49, in initialize_tensor_parallelism
raise OSError("Tensor parallel is only supported for `torch>=2.5`.")
OSError: Tensor parallel is only supported for `torch>=2.5`.
```
The torch version is `PyTorch version (accelerator?): 2.5.0a0+e000cf0ad9.nv24.10 (CUDA)`.
The startup command is as follows:
```
accelerate launch --config_file "deepspeeds.yaml" run_clm.py \
--do_train \
--do_eval \
--model_name_or_path model_to_ft \
--trust_remote_code True \
--torch_dtype auto \
--train_file train.json \
--validation_file val.json \
--streaming True \
--block_size 8192 \
--output_dir ./output \
--weight_decay 0.01 \
--learning_rate 5e-6 \
--lr_scheduler_type "cosine" \
--per_device_train_batch_size 4 \
--per_device_eval_batch_size 4 \
--gradient_accumulation_steps 16 \
--gradient_checkpointing True \
--gradient_checkpointing_kwargs '{"use_reentrant": true}' \
--max_grad_norm 0.5 \
--warmup_ratio 0.03 \
--bf16 \
--num_train_epochs 1 \
--aux_lambda 1e-5 \
--eval_strategy "steps" \
--eval_steps 5000 \
--save_strategy "steps" \
--save_steps 1000 \
--report_to "tensorboard" \
--save_total_limit 2 \
--logging_steps 10 \
--log_level "info" \
--include_num_input_tokens_seen True \
--save_only_model False
```
### Expected behavior
I didn't set any parameters related to TP, so why is TP being used by default? How can I resolve this issue? | {
"login": "mxjmtxrm",
"id": 167944756,
"node_id": "U_kgDOCgKiNA",
"avatar_url": "https://avatars.githubusercontent.com/u/167944756?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mxjmtxrm",
"html_url": "https://github.com/mxjmtxrm",
"followers_url": "https://api.github.com/users/mxjmtxrm/followers",
"following_url": "https://api.github.com/users/mxjmtxrm/following{/other_user}",
"gists_url": "https://api.github.com/users/mxjmtxrm/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mxjmtxrm/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mxjmtxrm/subscriptions",
"organizations_url": "https://api.github.com/users/mxjmtxrm/orgs",
"repos_url": "https://api.github.com/users/mxjmtxrm/repos",
"events_url": "https://api.github.com/users/mxjmtxrm/events{/privacy}",
"received_events_url": "https://api.github.com/users/mxjmtxrm/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39249/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39249/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39248 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39248/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39248/comments | https://api.github.com/repos/huggingface/transformers/issues/39248/events | https://github.com/huggingface/transformers/pull/39248 | 3,208,184,949 | PR_kwDOCUB6oc6dt1Us | 39,248 | Skip `test_eager_matches sdpa generate` and update an integration test for blip-like models | {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T09:13:34 | 2025-07-08T08:38:29 | 2025-07-08T08:38:25 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39248",
"html_url": "https://github.com/huggingface/transformers/pull/39248",
"diff_url": "https://github.com/huggingface/transformers/pull/39248.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39248.patch",
"merged_at": "2025-07-08T08:38:25"
} | # What does this PR do?
For skip, see [this comment](https://github.com/huggingface/transformers/pull/38510#discussion_r2121138005). Let's make the CI report clean though for now.
| {
"login": "ydshieh",
"id": 2521628,
"node_id": "MDQ6VXNlcjI1MjE2Mjg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2521628?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ydshieh",
"html_url": "https://github.com/ydshieh",
"followers_url": "https://api.github.com/users/ydshieh/followers",
"following_url": "https://api.github.com/users/ydshieh/following{/other_user}",
"gists_url": "https://api.github.com/users/ydshieh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ydshieh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ydshieh/subscriptions",
"organizations_url": "https://api.github.com/users/ydshieh/orgs",
"repos_url": "https://api.github.com/users/ydshieh/repos",
"events_url": "https://api.github.com/users/ydshieh/events{/privacy}",
"received_events_url": "https://api.github.com/users/ydshieh/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39248/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39248/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39247 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39247/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39247/comments | https://api.github.com/repos/huggingface/transformers/issues/39247/events | https://github.com/huggingface/transformers/pull/39247 | 3,208,083,286 | PR_kwDOCUB6oc6dtfpW | 39,247 | Glm 4 doc | {
"login": "zRzRzRzRzRzRzR",
"id": 93239683,
"node_id": "U_kgDOBY65gw",
"avatar_url": "https://avatars.githubusercontent.com/u/93239683?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zRzRzRzRzRzRzR",
"html_url": "https://github.com/zRzRzRzRzRzRzR",
"followers_url": "https://api.github.com/users/zRzRzRzRzRzRzR/followers",
"following_url": "https://api.github.com/users/zRzRzRzRzRzRzR/following{/other_user}",
"gists_url": "https://api.github.com/users/zRzRzRzRzRzRzR/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zRzRzRzRzRzRzR/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zRzRzRzRzRzRzR/subscriptions",
"organizations_url": "https://api.github.com/users/zRzRzRzRzRzRzR/orgs",
"repos_url": "https://api.github.com/users/zRzRzRzRzRzRzR/repos",
"events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/events{/privacy}",
"received_events_url": "https://api.github.com/users/zRzRzRzRzRzRzR/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T08:44:28 | 2025-07-14T04:39:23 | 2025-07-08T06:22:04 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39247",
"html_url": "https://github.com/huggingface/transformers/pull/39247",
"diff_url": "https://github.com/huggingface/transformers/pull/39247.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39247.patch",
"merged_at": "2025-07-08T06:22:04"
} | Improve the introduction documentation for GLM-4-0414 and GLM-4.1V.
Improve the pytest testing for GLM-4.1V. | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39247/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39247/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39246 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39246/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39246/comments | https://api.github.com/repos/huggingface/transformers/issues/39246/events | https://github.com/huggingface/transformers/pull/39246 | 3,207,996,206 | PR_kwDOCUB6oc6dtNOK | 39,246 | enable xpu on kv-cache and hqq doc | {
"login": "jiqing-feng",
"id": 107918818,
"node_id": "U_kgDOBm614g",
"avatar_url": "https://avatars.githubusercontent.com/u/107918818?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jiqing-feng",
"html_url": "https://github.com/jiqing-feng",
"followers_url": "https://api.github.com/users/jiqing-feng/followers",
"following_url": "https://api.github.com/users/jiqing-feng/following{/other_user}",
"gists_url": "https://api.github.com/users/jiqing-feng/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jiqing-feng/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jiqing-feng/subscriptions",
"organizations_url": "https://api.github.com/users/jiqing-feng/orgs",
"repos_url": "https://api.github.com/users/jiqing-feng/repos",
"events_url": "https://api.github.com/users/jiqing-feng/events{/privacy}",
"received_events_url": "https://api.github.com/users/jiqing-feng/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T08:13:41 | 2025-07-07T13:12:33 | 2025-07-07T13:12:02 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39246",
"html_url": "https://github.com/huggingface/transformers/pull/39246",
"diff_url": "https://github.com/huggingface/transformers/pull/39246.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39246.patch",
"merged_at": "2025-07-07T13:12:02"
} | Enable xpu on kv-cache and hqq doc. The previous quantized kv-cache usage is out-of-date, I have updated it and verified it on both A100 and XPU. | {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39246/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39246/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39245 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39245/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39245/comments | https://api.github.com/repos/huggingface/transformers/issues/39245/events | https://github.com/huggingface/transformers/pull/39245 | 3,207,925,464 | PR_kwDOCUB6oc6ds9z0 | 39,245 | Update tiny-agents example | {
"login": "Wauplin",
"id": 11801849,
"node_id": "MDQ6VXNlcjExODAxODQ5",
"avatar_url": "https://avatars.githubusercontent.com/u/11801849?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Wauplin",
"html_url": "https://github.com/Wauplin",
"followers_url": "https://api.github.com/users/Wauplin/followers",
"following_url": "https://api.github.com/users/Wauplin/following{/other_user}",
"gists_url": "https://api.github.com/users/Wauplin/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Wauplin/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Wauplin/subscriptions",
"organizations_url": "https://api.github.com/users/Wauplin/orgs",
"repos_url": "https://api.github.com/users/Wauplin/repos",
"events_url": "https://api.github.com/users/Wauplin/events{/privacy}",
"received_events_url": "https://api.github.com/users/Wauplin/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T07:50:34 | 2025-07-07T13:58:38 | 2025-07-07T13:58:36 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39245",
"html_url": "https://github.com/huggingface/transformers/pull/39245",
"diff_url": "https://github.com/huggingface/transformers/pull/39245.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39245.patch",
"merged_at": "2025-07-07T13:58:36"
} | Fix docs example after https://github.com/huggingface/huggingface_hub/pull/3166 / https://github.com/huggingface/huggingface.js/pull/1556. Since release [0.33.2](https://github.com/huggingface/huggingface_hub/releases/tag/v0.33.2) `tiny-agents` config follow VSCode format. We made the change without a proper deprecation warning as it's still experimental and we wanted to harmonize with VSCode as quickly as possible (to avoid future conflicts).
Related PRs:
- https://github.com/huggingface/hub-docs/pull/1816
- https://github.com/huggingface/transformers/pull/39245
- https://github.com/huggingface/huggingface_hub/pull/3205
- https://github.com/huggingface/huggingface.js/pull/1599 | {
"login": "Wauplin",
"id": 11801849,
"node_id": "MDQ6VXNlcjExODAxODQ5",
"avatar_url": "https://avatars.githubusercontent.com/u/11801849?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Wauplin",
"html_url": "https://github.com/Wauplin",
"followers_url": "https://api.github.com/users/Wauplin/followers",
"following_url": "https://api.github.com/users/Wauplin/following{/other_user}",
"gists_url": "https://api.github.com/users/Wauplin/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Wauplin/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Wauplin/subscriptions",
"organizations_url": "https://api.github.com/users/Wauplin/orgs",
"repos_url": "https://api.github.com/users/Wauplin/repos",
"events_url": "https://api.github.com/users/Wauplin/events{/privacy}",
"received_events_url": "https://api.github.com/users/Wauplin/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39245/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39245/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39244 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39244/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39244/comments | https://api.github.com/repos/huggingface/transformers/issues/39244/events | https://github.com/huggingface/transformers/pull/39244 | 3,207,733,609 | PR_kwDOCUB6oc6dsUg5 | 39,244 | Fix missing fast tokenizer/image_processor in whisper/qwen2.5-omni processor | {
"login": "Isotr0py",
"id": 41363108,
"node_id": "MDQ6VXNlcjQxMzYzMTA4",
"avatar_url": "https://avatars.githubusercontent.com/u/41363108?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Isotr0py",
"html_url": "https://github.com/Isotr0py",
"followers_url": "https://api.github.com/users/Isotr0py/followers",
"following_url": "https://api.github.com/users/Isotr0py/following{/other_user}",
"gists_url": "https://api.github.com/users/Isotr0py/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Isotr0py/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Isotr0py/subscriptions",
"organizations_url": "https://api.github.com/users/Isotr0py/orgs",
"repos_url": "https://api.github.com/users/Isotr0py/repos",
"events_url": "https://api.github.com/users/Isotr0py/events{/privacy}",
"received_events_url": "https://api.github.com/users/Isotr0py/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T06:38:53 | 2025-07-22T04:11:26 | 2025-07-07T11:54:18 | COLLABORATOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39244",
"html_url": "https://github.com/huggingface/transformers/pull/39244",
"diff_url": "https://github.com/huggingface/transformers/pull/39244.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39244.patch",
"merged_at": "2025-07-07T11:54:18"
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes https://github.com/vllm-project/vllm/issues/20224
- Fix missing `WhisperTokenizerFast` in whisper processor
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "ArthurZucker",
"id": 48595927,
"node_id": "MDQ6VXNlcjQ4NTk1OTI3",
"avatar_url": "https://avatars.githubusercontent.com/u/48595927?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurZucker",
"html_url": "https://github.com/ArthurZucker",
"followers_url": "https://api.github.com/users/ArthurZucker/followers",
"following_url": "https://api.github.com/users/ArthurZucker/following{/other_user}",
"gists_url": "https://api.github.com/users/ArthurZucker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ArthurZucker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ArthurZucker/subscriptions",
"organizations_url": "https://api.github.com/users/ArthurZucker/orgs",
"repos_url": "https://api.github.com/users/ArthurZucker/repos",
"events_url": "https://api.github.com/users/ArthurZucker/events{/privacy}",
"received_events_url": "https://api.github.com/users/ArthurZucker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39244/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39244/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39243 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39243/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39243/comments | https://api.github.com/repos/huggingface/transformers/issues/39243/events | https://github.com/huggingface/transformers/pull/39243 | 3,207,623,582 | PR_kwDOCUB6oc6dr9Xo | 39,243 | depreated kernels of ms_deform_attn support fp16 | {
"login": "chagmgang",
"id": 37325825,
"node_id": "MDQ6VXNlcjM3MzI1ODI1",
"avatar_url": "https://avatars.githubusercontent.com/u/37325825?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/chagmgang",
"html_url": "https://github.com/chagmgang",
"followers_url": "https://api.github.com/users/chagmgang/followers",
"following_url": "https://api.github.com/users/chagmgang/following{/other_user}",
"gists_url": "https://api.github.com/users/chagmgang/gists{/gist_id}",
"starred_url": "https://api.github.com/users/chagmgang/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/chagmgang/subscriptions",
"organizations_url": "https://api.github.com/users/chagmgang/orgs",
"repos_url": "https://api.github.com/users/chagmgang/repos",
"events_url": "https://api.github.com/users/chagmgang/events{/privacy}",
"received_events_url": "https://api.github.com/users/chagmgang/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T05:43:20 | 2025-07-08T06:48:54 | 2025-07-08T06:48:54 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39243",
"html_url": "https://github.com/huggingface/transformers/pull/39243",
"diff_url": "https://github.com/huggingface/transformers/pull/39243.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39243.patch",
"merged_at": null
} | # What does this PR do?
* This PR for deprecated kernels of ms_deform_attn support fp16.
<!-- Remove if not applicable -->
## Who can review?
@amyeroberts | {
"login": "chagmgang",
"id": 37325825,
"node_id": "MDQ6VXNlcjM3MzI1ODI1",
"avatar_url": "https://avatars.githubusercontent.com/u/37325825?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/chagmgang",
"html_url": "https://github.com/chagmgang",
"followers_url": "https://api.github.com/users/chagmgang/followers",
"following_url": "https://api.github.com/users/chagmgang/following{/other_user}",
"gists_url": "https://api.github.com/users/chagmgang/gists{/gist_id}",
"starred_url": "https://api.github.com/users/chagmgang/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/chagmgang/subscriptions",
"organizations_url": "https://api.github.com/users/chagmgang/orgs",
"repos_url": "https://api.github.com/users/chagmgang/repos",
"events_url": "https://api.github.com/users/chagmgang/events{/privacy}",
"received_events_url": "https://api.github.com/users/chagmgang/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39243/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39243/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39242 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39242/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39242/comments | https://api.github.com/repos/huggingface/transformers/issues/39242/events | https://github.com/huggingface/transformers/pull/39242 | 3,207,581,875 | PR_kwDOCUB6oc6dr0hZ | 39,242 | [vlm] fix loading of retrieval VLMs | {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-07T05:20:44 | 2025-07-15T15:23:54 | 2025-07-15T15:23:54 | MEMBER | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39242",
"html_url": "https://github.com/huggingface/transformers/pull/39242",
"diff_url": "https://github.com/huggingface/transformers/pull/39242.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39242.patch",
"merged_at": "2025-07-15T15:23:54"
} | # What does this PR do?
As per title, reported internally that slow tests are failing. We need to apply same changes as in VLMs to the models that use VLMs in their architecture
| {
"login": "zucchini-nlp",
"id": 100715397,
"node_id": "U_kgDOBgDLhQ",
"avatar_url": "https://avatars.githubusercontent.com/u/100715397?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zucchini-nlp",
"html_url": "https://github.com/zucchini-nlp",
"followers_url": "https://api.github.com/users/zucchini-nlp/followers",
"following_url": "https://api.github.com/users/zucchini-nlp/following{/other_user}",
"gists_url": "https://api.github.com/users/zucchini-nlp/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zucchini-nlp/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zucchini-nlp/subscriptions",
"organizations_url": "https://api.github.com/users/zucchini-nlp/orgs",
"repos_url": "https://api.github.com/users/zucchini-nlp/repos",
"events_url": "https://api.github.com/users/zucchini-nlp/events{/privacy}",
"received_events_url": "https://api.github.com/users/zucchini-nlp/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39242/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39242/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39241 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39241/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39241/comments | https://api.github.com/repos/huggingface/transformers/issues/39241/events | https://github.com/huggingface/transformers/issues/39241 | 3,207,216,335 | I_kwDOCUB6oc6_KjzP | 39,241 | [Trainer] Eval loss depends on batch size (with solution) | {
"login": "ba144220",
"id": 62748666,
"node_id": "MDQ6VXNlcjYyNzQ4NjY2",
"avatar_url": "https://avatars.githubusercontent.com/u/62748666?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ba144220",
"html_url": "https://github.com/ba144220",
"followers_url": "https://api.github.com/users/ba144220/followers",
"following_url": "https://api.github.com/users/ba144220/following{/other_user}",
"gists_url": "https://api.github.com/users/ba144220/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ba144220/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ba144220/subscriptions",
"organizations_url": "https://api.github.com/users/ba144220/orgs",
"repos_url": "https://api.github.com/users/ba144220/repos",
"events_url": "https://api.github.com/users/ba144220/events{/privacy}",
"received_events_url": "https://api.github.com/users/ba144220/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-07T00:42:36 | 2025-08-18T08:03:20 | 2025-08-18T08:03:20 | NONE | null | null | null | null | ### System Info
- `transformers` version: 4.54.0.dev0
- Platform: Linux-5.15.0-1047-oracle-x86_64-with-glibc2.35
- Python version: 3.12.11
- Huggingface_hub version: 0.33.1
- Safetensors version: 0.5.3
- Accelerate version: 1.8.1
- Accelerate config: not found
- DeepSpeed version: 0.17.1
- PyTorch version (accelerator?): 2.7.1+cu126 (CUDA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: No
- Using GPU in script?: Yes
- GPU type: NVIDIA A100-SXM4-80GB
### Who can help?
@SunMarc @ArthurZucker
### Information
- [x] The official example scripts
- [ ] My own modified scripts
### Tasks
- [x] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
> This is actually a known issue for several years, see: https://discuss.huggingface.co/t/batch-size-during-training-vs-batch-size-during-evaluation/20827 and https://discuss.huggingface.co/t/evaluation-loss-depends-on-batch-size/112046
Iβve been evaluating a few causal LMs (e.g. `Qwen/Qwen2.5-3B`) on 512 samples from the `togethercomputer/RedPajama-Data-1T-Sample` pre-train dataset, and I noticed that **eval loss consistently decreases as I increase the batch size**:
Batch size | Eval loss
-- | --
1 | 2.414
2 | 2.340
4 | 2.299
8 | 2.298
16 | 2.296
I saw the same trend across other models as well.
This is the code Iβm using:
```python
import argparse
import os
import torch
from dotenv import load_dotenv
load_dotenv()
from transformers import AutoModelForCausalLM, AutoTokenizer
from trl.trl import SFTConfig, SFTTrainer
from datasets import load_dataset
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--model_name", type=str, default="Qwen/Qwen2.5-3B")
parser.add_argument("--dataset_name", type=str, default="togethercomputer/RedPajama-Data-1T-Sample")
parser.add_argument("--batch_size", type=int, default=1)
parser.add_argument("--max_seq_length", type=int, default=2048)
parser.add_argument("--max_eval_samples", type=int, default=512)
parser.add_argument("--seed", type=int, default=42)
args = parser.parse_args()
model = AutoModelForCausalLM.from_pretrained(args.model_name, trust_remote_code=True, device_map="auto", token=os.getenv("HF_TOKEN"), torch_dtype=torch.bfloat16)
tokenizer = AutoTokenizer.from_pretrained(args.model_name, trust_remote_code=True, token=os.getenv("HF_TOKEN"))
# Load dataset
dataset = load_dataset(args.dataset_name, split="train")
dataset = dataset.shuffle(args.seed).select(range(args.max_eval_samples))
sft_config = SFTConfig(
output_dir="./results",
per_device_train_batch_size=args.batch_size,
per_device_eval_batch_size=args.batch_size,
dataset_text_field="text",
max_seq_length=args.max_seq_length,
)
trainer = SFTTrainer(
model=model,
args=sft_config,
train_dataset=dataset,
eval_dataset=dataset,
processing_class=tokenizer,
)
eval_result = trainer.evaluate()
print(eval_result)
if __name__ == "__main__":
main()
```
### Solution
Digging in, I found that `fixed_cross_entropy` (in `transformers/src/transformers/loss/loss_utils.py`) does a token-level sum then divides by the total non-padding token count (micro-averaging). To fix the issue, I implemented a **sample-wise average** (macro-averaging):
```python
# Hugging Face: token-sum / total_tokens
loss = F.cross_entropy(..., reduction="sum") / num_items_in_batch
# My version: per-sequence average then mean across sequences
loss = F.cross_entropy(..., reduction="none")
loss = loss.view(B, -1).sum(dim=1) / token_counts_per_seq
loss = loss.mean()
```
With macro-averaging, **eval loss is identical across batch sizes and input orderings**, enabling a few nice benefits:
1. We can choose optimal batch size to speed up evaluation, especially when comparing models of different sizes.
2. Sorting samples by length before batching reduces padding, reducing evaluation time by over **50%**.
So I'm wondering:
1. Is the Trainerβs default (micro-averaging) behavior on purposeβto tie loss scale strictly to total token count?
2. Does this have any documented effect on training stability or convergence when you vary batch size?
3. Are there recommended best practices for loss normalization in large-batch LLM training (e.g. should I always override this to macro-average)?
Iβd love to hear from anyone whoβs dug into this or has empirical experience with different loss-averaging schemes in the π€Trainer. Thanks in advance! | {
"login": "github-actions[bot]",
"id": 41898282,
"node_id": "MDM6Qm90NDE4OTgyODI=",
"avatar_url": "https://avatars.githubusercontent.com/in/15368?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/github-actions%5Bbot%5D",
"html_url": "https://github.com/apps/github-actions",
"followers_url": "https://api.github.com/users/github-actions%5Bbot%5D/followers",
"following_url": "https://api.github.com/users/github-actions%5Bbot%5D/following{/other_user}",
"gists_url": "https://api.github.com/users/github-actions%5Bbot%5D/gists{/gist_id}",
"starred_url": "https://api.github.com/users/github-actions%5Bbot%5D/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/github-actions%5Bbot%5D/subscriptions",
"organizations_url": "https://api.github.com/users/github-actions%5Bbot%5D/orgs",
"repos_url": "https://api.github.com/users/github-actions%5Bbot%5D/repos",
"events_url": "https://api.github.com/users/github-actions%5Bbot%5D/events{/privacy}",
"received_events_url": "https://api.github.com/users/github-actions%5Bbot%5D/received_events",
"type": "Bot",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39241/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39241/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39240 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39240/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39240/comments | https://api.github.com/repos/huggingface/transformers/issues/39240/events | https://github.com/huggingface/transformers/pull/39240 | 3,206,912,446 | PR_kwDOCUB6oc6dpn46 | 39,240 | Improve RoBERTa documentation with explanations and helpful links | {
"login": "PrakyathMC",
"id": 92665624,
"node_id": "U_kgDOBYX3GA",
"avatar_url": "https://avatars.githubusercontent.com/u/92665624?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/PrakyathMC",
"html_url": "https://github.com/PrakyathMC",
"followers_url": "https://api.github.com/users/PrakyathMC/followers",
"following_url": "https://api.github.com/users/PrakyathMC/following{/other_user}",
"gists_url": "https://api.github.com/users/PrakyathMC/gists{/gist_id}",
"starred_url": "https://api.github.com/users/PrakyathMC/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/PrakyathMC/subscriptions",
"organizations_url": "https://api.github.com/users/PrakyathMC/orgs",
"repos_url": "https://api.github.com/users/PrakyathMC/repos",
"events_url": "https://api.github.com/users/PrakyathMC/events{/privacy}",
"received_events_url": "https://api.github.com/users/PrakyathMC/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-06T19:37:52 | 2025-07-07T16:29:05 | 2025-07-07T16:29:05 | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39240",
"html_url": "https://github.com/huggingface/transformers/pull/39240",
"diff_url": "https://github.com/huggingface/transformers/pull/39240.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39240.patch",
"merged_at": null
} | ## What does this PR do?
Improves the RoBERTa model documentation following the standardized format outlined in #36979.
## Changes Made
- β
Added explanatory comments to Pipeline and AutoModel code examples
- β
Added "Quick Links" section with popular RoBERTa models on HuggingFace Hub
- β
Included additional resources (paper links, model collection)
- β
Improved accessibility for beginners learning to use RoBERTa
## Before and After
**Before**: Basic code examples without explanations
**After**: Commented code that teaches users what each line does + easy access to models
#36979 | {
"login": "stevhliu",
"id": 59462357,
"node_id": "MDQ6VXNlcjU5NDYyMzU3",
"avatar_url": "https://avatars.githubusercontent.com/u/59462357?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stevhliu",
"html_url": "https://github.com/stevhliu",
"followers_url": "https://api.github.com/users/stevhliu/followers",
"following_url": "https://api.github.com/users/stevhliu/following{/other_user}",
"gists_url": "https://api.github.com/users/stevhliu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stevhliu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stevhliu/subscriptions",
"organizations_url": "https://api.github.com/users/stevhliu/orgs",
"repos_url": "https://api.github.com/users/stevhliu/repos",
"events_url": "https://api.github.com/users/stevhliu/events{/privacy}",
"received_events_url": "https://api.github.com/users/stevhliu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39240/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39240/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39239 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39239/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39239/comments | https://api.github.com/repos/huggingface/transformers/issues/39239/events | https://github.com/huggingface/transformers/pull/39239 | 3,206,662,595 | PR_kwDOCUB6oc6do02F | 39,239 | Fix missing initializations for models created in 2023 | {
"login": "bvantuan",
"id": 37981884,
"node_id": "MDQ6VXNlcjM3OTgxODg0",
"avatar_url": "https://avatars.githubusercontent.com/u/37981884?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bvantuan",
"html_url": "https://github.com/bvantuan",
"followers_url": "https://api.github.com/users/bvantuan/followers",
"following_url": "https://api.github.com/users/bvantuan/following{/other_user}",
"gists_url": "https://api.github.com/users/bvantuan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/bvantuan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/bvantuan/subscriptions",
"organizations_url": "https://api.github.com/users/bvantuan/orgs",
"repos_url": "https://api.github.com/users/bvantuan/repos",
"events_url": "https://api.github.com/users/bvantuan/events{/privacy}",
"received_events_url": "https://api.github.com/users/bvantuan/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-06T15:14:13 | 2025-07-28T08:43:59 | 2025-07-21T12:43:52 | CONTRIBUTOR | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39239",
"html_url": "https://github.com/huggingface/transformers/pull/39239",
"diff_url": "https://github.com/huggingface/transformers/pull/39239.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39239.patch",
"merged_at": "2025-07-21T12:43:52"
} | # What does this PR do?
<!--
Congratulations! You've made it this far! You're not quite done yet though.
Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution.
Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change.
Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost.
-->
<!-- Remove if not applicable -->
Fixes missing weight initializations for models created in 2023.
## Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@Cyrilvallez
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| {
"login": "Cyrilvallez",
"id": 71554963,
"node_id": "MDQ6VXNlcjcxNTU0OTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/71554963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Cyrilvallez",
"html_url": "https://github.com/Cyrilvallez",
"followers_url": "https://api.github.com/users/Cyrilvallez/followers",
"following_url": "https://api.github.com/users/Cyrilvallez/following{/other_user}",
"gists_url": "https://api.github.com/users/Cyrilvallez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Cyrilvallez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Cyrilvallez/subscriptions",
"organizations_url": "https://api.github.com/users/Cyrilvallez/orgs",
"repos_url": "https://api.github.com/users/Cyrilvallez/repos",
"events_url": "https://api.github.com/users/Cyrilvallez/events{/privacy}",
"received_events_url": "https://api.github.com/users/Cyrilvallez/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39239/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39239/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39238 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39238/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39238/comments | https://api.github.com/repos/huggingface/transformers/issues/39238/events | https://github.com/huggingface/transformers/pull/39238 | 3,206,279,866 | PR_kwDOCUB6oc6dnnxP | 39,238 | [dump] support npu fusion patch | {
"login": "zheliuyu",
"id": 190869220,
"node_id": "U_kgDOC2Bu5A",
"avatar_url": "https://avatars.githubusercontent.com/u/190869220?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zheliuyu",
"html_url": "https://github.com/zheliuyu",
"followers_url": "https://api.github.com/users/zheliuyu/followers",
"following_url": "https://api.github.com/users/zheliuyu/following{/other_user}",
"gists_url": "https://api.github.com/users/zheliuyu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zheliuyu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zheliuyu/subscriptions",
"organizations_url": "https://api.github.com/users/zheliuyu/orgs",
"repos_url": "https://api.github.com/users/zheliuyu/repos",
"events_url": "https://api.github.com/users/zheliuyu/events{/privacy}",
"received_events_url": "https://api.github.com/users/zheliuyu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | closed | false | null | [] | null | [] | 2025-07-06T08:11:13 | 2025-09-17T06:32:04 | 2025-08-03T01:28:25 | CONTRIBUTOR | null | null | true | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39238",
"html_url": "https://github.com/huggingface/transformers/pull/39238",
"diff_url": "https://github.com/huggingface/transformers/pull/39238.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39238.patch",
"merged_at": null
} | # What does this PR do?
An attempt for https://github.com/huggingface/transformers/issues/39105
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [x] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?
## Who can review?
WIP
| {
"login": "zheliuyu",
"id": 190869220,
"node_id": "U_kgDOC2Bu5A",
"avatar_url": "https://avatars.githubusercontent.com/u/190869220?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zheliuyu",
"html_url": "https://github.com/zheliuyu",
"followers_url": "https://api.github.com/users/zheliuyu/followers",
"following_url": "https://api.github.com/users/zheliuyu/following{/other_user}",
"gists_url": "https://api.github.com/users/zheliuyu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zheliuyu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zheliuyu/subscriptions",
"organizations_url": "https://api.github.com/users/zheliuyu/orgs",
"repos_url": "https://api.github.com/users/zheliuyu/repos",
"events_url": "https://api.github.com/users/zheliuyu/events{/privacy}",
"received_events_url": "https://api.github.com/users/zheliuyu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39238/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39238/timeline | null | null | null | null | true | true |
https://api.github.com/repos/huggingface/transformers/issues/39237 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39237/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39237/comments | https://api.github.com/repos/huggingface/transformers/issues/39237/events | https://github.com/huggingface/transformers/issues/39237 | 3,205,640,608 | I_kwDOCUB6oc6_EjGg | 39,237 | disable_grouping parameter missed in image_processing_glm4v_fast.py | {
"login": "r0mar0ma",
"id": 76665418,
"node_id": "MDQ6VXNlcjc2NjY1NDE4",
"avatar_url": "https://avatars.githubusercontent.com/u/76665418?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/r0mar0ma",
"html_url": "https://github.com/r0mar0ma",
"followers_url": "https://api.github.com/users/r0mar0ma/followers",
"following_url": "https://api.github.com/users/r0mar0ma/following{/other_user}",
"gists_url": "https://api.github.com/users/r0mar0ma/gists{/gist_id}",
"starred_url": "https://api.github.com/users/r0mar0ma/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/r0mar0ma/subscriptions",
"organizations_url": "https://api.github.com/users/r0mar0ma/orgs",
"repos_url": "https://api.github.com/users/r0mar0ma/repos",
"events_url": "https://api.github.com/users/r0mar0ma/events{/privacy}",
"received_events_url": "https://api.github.com/users/r0mar0ma/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [
{
"id": 3817266200,
"node_id": "MDU6TGFiZWwzODE3MjY2MjAw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": null
}
] | closed | false | null | [] | null | [] | 2025-07-05T21:21:59 | 2025-07-07T09:42:53 | 2025-07-07T09:42:53 | NONE | null | null | null | null | ### System Info
transformers 4.53.1
### Who can help?
@amyeroberts, @qubvel
### Information
- [ ] The official example scripts
- [ ] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
AutoProcessor.from_pretrained(path, **use_fast = True**)
in models/glm4v/image_processing_glm4v_fast.py
```
# Group images by size for batched resizing
grouped_images, grouped_images_index = group_images_by_shape(images)
```
while group_images_by_shape has required positional argument 'disable_grouping'
```
def group_images_by_shape(
images: Union[list["torch.Tensor"], "torch.Tensor"],
disable_grouping: bool,
is_nested: bool = False,
) -> tuple[
```
### Expected behavior
disable_grouping parameter should be added | {
"login": "r0mar0ma",
"id": 76665418,
"node_id": "MDQ6VXNlcjc2NjY1NDE4",
"avatar_url": "https://avatars.githubusercontent.com/u/76665418?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/r0mar0ma",
"html_url": "https://github.com/r0mar0ma",
"followers_url": "https://api.github.com/users/r0mar0ma/followers",
"following_url": "https://api.github.com/users/r0mar0ma/following{/other_user}",
"gists_url": "https://api.github.com/users/r0mar0ma/gists{/gist_id}",
"starred_url": "https://api.github.com/users/r0mar0ma/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/r0mar0ma/subscriptions",
"organizations_url": "https://api.github.com/users/r0mar0ma/orgs",
"repos_url": "https://api.github.com/users/r0mar0ma/repos",
"events_url": "https://api.github.com/users/r0mar0ma/events{/privacy}",
"received_events_url": "https://api.github.com/users/r0mar0ma/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39237/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39237/timeline | null | completed | {
"total": 0,
"completed": 0,
"percent_completed": 0
} | {
"blocked_by": 0,
"total_blocked_by": 0,
"blocking": 0,
"total_blocking": 0
} | false | true |
https://api.github.com/repos/huggingface/transformers/issues/39236 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/39236/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/39236/comments | https://api.github.com/repos/huggingface/transformers/issues/39236/events | https://github.com/huggingface/transformers/pull/39236 | 3,205,328,822 | PR_kwDOCUB6oc6dkndH | 39,236 | added moment_p sampling | {
"login": "Hellisotherpeople",
"id": 12686966,
"node_id": "MDQ6VXNlcjEyNjg2OTY2",
"avatar_url": "https://avatars.githubusercontent.com/u/12686966?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Hellisotherpeople",
"html_url": "https://github.com/Hellisotherpeople",
"followers_url": "https://api.github.com/users/Hellisotherpeople/followers",
"following_url": "https://api.github.com/users/Hellisotherpeople/following{/other_user}",
"gists_url": "https://api.github.com/users/Hellisotherpeople/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Hellisotherpeople/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Hellisotherpeople/subscriptions",
"organizations_url": "https://api.github.com/users/Hellisotherpeople/orgs",
"repos_url": "https://api.github.com/users/Hellisotherpeople/repos",
"events_url": "https://api.github.com/users/Hellisotherpeople/events{/privacy}",
"received_events_url": "https://api.github.com/users/Hellisotherpeople/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
} | [] | open | false | null | [] | null | [] | 2025-07-05T17:45:07 | 2025-10-07T14:24:21 | null | NONE | null | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/39236",
"html_url": "https://github.com/huggingface/transformers/pull/39236",
"diff_url": "https://github.com/huggingface/transformers/pull/39236.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/39236.patch",
"merged_at": null
} | # Moment-P Sampling Implementation for Hugging Face Transformers
## Overview
This implementation adds a new sampling method called "Moment-P" to the Hugging Face Transformers library. Moment-P is a novel sampling technique that filters tokens based on the moment (sum of probabilities raised to a power) of the probability distribution. It uses the whole probability distribution and enables very high temperature generations with coherence. A paper is incoming.
## What is Moment-P Sampling?
Moment-P sampling filters tokens by calculating a threshold based on the moment of the probability distribution. The moment is calculated as:
```
moment = Ξ£(p_i^exponent)
```
Where:
- `p_i` is the probability of token i
- `exponent` is a configurable parameter (default: 2.0)
Tokens are kept until their cumulative moment exceeds `total_moment * alpha`.
### Key Parameters
- **moment_p_exponent** (float, default=2.0): The exponent to which probabilities are raised
- **moment_p_alpha** (float, default=1.0): Scaling factor for the threshold
### Behavior with Different Exponents
- **exponent=1**: Linear moment, truncate all tokens
- **exponent=2** (default): Second moment, emphasizes variance-like behavior
- **exponent>2**: Higher emphasis on the tail of the distribution, lower truncation values
## Implementation Details
The implementation consists of:
1. **MomentPLogitsWarper** class in `logits_process.py`
2. Configuration parameters in `configuration_utils.py`
3. Integration in `utils.py` for the generation pipeline
4. Export in `__init__.py`
## Usage Examples
### Basic Usage
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("distilbert/distilgpt2")
tokenizer = AutoTokenizer.from_pretrained("distilbert/distilgpt2")
inputs = tokenizer("The future of AI is", return_tensors="pt")
# Generate with moment_p sampling
output = model.generate(
**inputs,
do_sample=True,
moment_p_exponent=2.0,
moment_p_alpha=1.0,
max_new_tokens=50
)
print(tokenizer.decode(output[0], skip_special_tokens=True))
```
### Advanced Usage with Different Parameters
```python
# More selective filtering with lower alpha
output = model.generate(
**inputs,
do_sample=True,
moment_p_exponent=2.0,
moment_p_alpha=0.5, # More selective
max_new_tokens=50
)
# Higher exponent for more tail emphasis
output = model.generate(
**inputs,
do_sample=True,
moment_p_exponent=3.0, # Higher moment
moment_p_alpha=1.0,
max_new_tokens=50
)
```
## Mathematical Foundation
The moment-based approach allows for flexible control over the token selection process:
1. **Calculate probabilities**: `probs = softmax(logits / temperature)`
2. **Calculate total moment**: `total_moment = Ξ£(probs^exponent)`
3. **Set threshold**: `threshold = total_moment * alpha`
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [X] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link
to it if that's the case.
- [X] Did you make sure to update the documentation with your changes? Here are the
[documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and
[here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [X] Did you write any new necessary tests?
## Who can review?
@gante
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
<!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @
If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
Please tag fewer than 3 people.
Models:
- text models: @ArthurZucker
- vision models: @amyeroberts, @qubvel
- speech models: @eustlb
- graph models: @clefourrier
Library:
- flax: @gante and @Rocketknight1
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- pipelines: @Rocketknight1
- tensorflow: @gante and @Rocketknight1
- tokenizers: @ArthurZucker
- trainer: @zach-huggingface, @SunMarc and @qgallouedec
- chat templates: @Rocketknight1
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
Documentation: @stevhliu
HF projects:
- accelerate: [different repo](https://github.com/huggingface/accelerate)
- datasets: [different repo](https://github.com/huggingface/datasets)
- diffusers: [different repo](https://github.com/huggingface/diffusers)
- rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
Maintained examples (not research project or legacy):
- Flax: @Rocketknight1
- PyTorch: See Models above and tag the person corresponding to the modality of the example.
- TensorFlow: @Rocketknight1
-->
| null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/39236/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/39236/timeline | null | null | null | null | true | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.