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/41542
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41542/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41542/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41542/events
|
https://github.com/huggingface/transformers/pull/41542
| 3,509,763,626
|
PR_kwDOCUB6oc6tcrYw
| 41,542
|
Add conditional checks to _check_and_adjust_attn_implementation()
|
{
"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 |
[] | 2025-10-13T11:56:11
| 2025-10-14T15:53:04
| 2025-10-14T13:00:08
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41542",
"html_url": "https://github.com/huggingface/transformers/pull/41542",
"diff_url": "https://github.com/huggingface/transformers/pull/41542.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41542.patch",
"merged_at": "2025-10-14T13:00:07"
}
|
# 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.
-->
To prevent unnecessary downloads by kernels, avoid installing kernels-community/flash-attn and kernels-community/vllm-flash-attn3 when attn_implementation=flash_attention_2 is specified for NPU.
## test
### script
```
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(
"Qwen/Qwen3-0.6B",
device_map="auto",
torch_dtype="auto",
attn_implementation="flash_attention_2",
).eval()
print("Operation successful")
```
### Before
```
`torch_dtype` is deprecated! Use `dtype` instead!
Fetching 0 files: 0it [00:00, ?it/s]
Traceback (most recent call last):
File "/root/kernels-main/src/kernels/utils.py", line 144, in install_kernel
return _load_kernel_from_path(repo_path, package_name, variant_locks)
File "/root/kernels-main/src/kernels/utils.py", line 177, in _load_kernel_from_path
raise FileNotFoundError(
FileNotFoundError: Kernel at path `/root/.cache/huggingface/hub/models--kernels-community--flash-attn/snapshots/90b3e941627659b28ff001c08b218315e1b7183b` does not have build: torch27-cxx11-cann81-aarch64-linux
```
### After
```
`torch_dtype` is deprecated! Use `dtype` instead!
Operation successful
```
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41542/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/41542/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41541
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41541/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41541/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41541/events
|
https://github.com/huggingface/transformers/pull/41541
| 3,509,294,372
|
PR_kwDOCUB6oc6tbDID
| 41,541
|
Untangle config inheritance
|
{
"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
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-13T09:57:13
| 2025-10-20T09:30:31
| null |
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41541",
"html_url": "https://github.com/huggingface/transformers/pull/41541",
"diff_url": "https://github.com/huggingface/transformers/pull/41541.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41541.patch",
"merged_at": null
}
|
# What does this PR do?
As per title, deletes base config attributes that are not actually universal for all models (e.g. token ids used only for text models and tying can be done only if model has embedding layers)
After this PR, we need to clean up generation-related params from config classes (https://github.com/huggingface/transformers/pull/41695), and then we can easily turn config classes to a `dataclass`. Using `dataclass` is one of the requirements for hf hub type validation which I am working on currently
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41541/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/41541/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41540
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41540/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41540/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41540/events
|
https://github.com/huggingface/transformers/pull/41540
| 3,509,101,112
|
PR_kwDOCUB6oc6taZVQ
| 41,540
|
Added kernels from kernel hub for Bamba model
|
{
"login": "romitjain",
"id": 11757603,
"node_id": "MDQ6VXNlcjExNzU3NjAz",
"avatar_url": "https://avatars.githubusercontent.com/u/11757603?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/romitjain",
"html_url": "https://github.com/romitjain",
"followers_url": "https://api.github.com/users/romitjain/followers",
"following_url": "https://api.github.com/users/romitjain/following{/other_user}",
"gists_url": "https://api.github.com/users/romitjain/gists{/gist_id}",
"starred_url": "https://api.github.com/users/romitjain/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/romitjain/subscriptions",
"organizations_url": "https://api.github.com/users/romitjain/orgs",
"repos_url": "https://api.github.com/users/romitjain/repos",
"events_url": "https://api.github.com/users/romitjain/events{/privacy}",
"received_events_url": "https://api.github.com/users/romitjain/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-13T08:59:35
| 2025-10-17T08:27:51
| null |
CONTRIBUTOR
| null | null | true
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41540",
"html_url": "https://github.com/huggingface/transformers/pull/41540",
"diff_url": "https://github.com/huggingface/transformers/pull/41540.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41540.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 -->
Adds support for `mamba_ssm` and `causal_conv1d` kernels from the kernel-hub in bamba models.
Fixes # (issue)
https://github.com/huggingface/transformers/issues/41208
## 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?
- [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?
@vasqu @MekkCyber @drbh
<!-- 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41540/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/41540/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41539
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41539/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41539/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41539/events
|
https://github.com/huggingface/transformers/issues/41539
| 3,509,035,329
|
I_kwDOCUB6oc7RJ6FB
| 41,539
|
All POETRY operations fail on latest version 4.57.0
|
{
"login": "bfuia",
"id": 159527810,
"node_id": "U_kgDOCYIzgg",
"avatar_url": "https://avatars.githubusercontent.com/u/159527810?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bfuia",
"html_url": "https://github.com/bfuia",
"followers_url": "https://api.github.com/users/bfuia/followers",
"following_url": "https://api.github.com/users/bfuia/following{/other_user}",
"gists_url": "https://api.github.com/users/bfuia/gists{/gist_id}",
"starred_url": "https://api.github.com/users/bfuia/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/bfuia/subscriptions",
"organizations_url": "https://api.github.com/users/bfuia/orgs",
"repos_url": "https://api.github.com/users/bfuia/repos",
"events_url": "https://api.github.com/users/bfuia/events{/privacy}",
"received_events_url": "https://api.github.com/users/bfuia/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 |
[] | 2025-10-13T08:40:49
| 2025-10-13T14:18:02
| 2025-10-13T14:18:02
|
NONE
| null | null | null | null |
### System Info
I import transformers (always latest) in my poetry project.
I use poetry 2.1.2
After this transformers release (4.57.0) I regenerated the poetry lock with command: `poetry lock`
Then when retrying to generate the lock again after other updates - it fails with message:
`Could not parse constrains version: <emtpy>`
### 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
Doing a simple search in the poetry.lock file I found out that transformers latest package needs `optax (<empty>)`
which produces this failure because poetry does not know how to parse this type of version.
Note I am sure that this is the problem because commenting out the transformers the lock works fine, and also by using 4.56.2 from September it also works fine and that `optax (<empty>)` cannot be found in the lock in this case.
### Expected behavior
A developer should be able to use the latest transformers package version with poetry.
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41539/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/transformers/issues/41539/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/41538
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41538/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41538/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41538/events
|
https://github.com/huggingface/transformers/pull/41538
| 3,508,811,198
|
PR_kwDOCUB6oc6tZa6G
| 41,538
|
examples/rag from original paper issue
|
{
"login": "pratap834",
"id": 117664342,
"node_id": "U_kgDOBwNqVg",
"avatar_url": "https://avatars.githubusercontent.com/u/117664342?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/pratap834",
"html_url": "https://github.com/pratap834",
"followers_url": "https://api.github.com/users/pratap834/followers",
"following_url": "https://api.github.com/users/pratap834/following{/other_user}",
"gists_url": "https://api.github.com/users/pratap834/gists{/gist_id}",
"starred_url": "https://api.github.com/users/pratap834/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/pratap834/subscriptions",
"organizations_url": "https://api.github.com/users/pratap834/orgs",
"repos_url": "https://api.github.com/users/pratap834/repos",
"events_url": "https://api.github.com/users/pratap834/events{/privacy}",
"received_events_url": "https://api.github.com/users/pratap834/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[
{
"id": 9258341780,
"node_id": "LA_kwDOCUB6oc8AAAACJ9cVlA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Code%20agent%20slop",
"name": "Code agent slop",
"color": "C59579",
"default": false,
"description": ""
}
] |
closed
| false
| null |
[] | null |
[] | 2025-10-13T07:27:41
| 2025-10-13T14:16:21
| 2025-10-13T14:16:21
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41538",
"html_url": "https://github.com/huggingface/transformers/pull/41538",
"diff_url": "https://github.com/huggingface/transformers/pull/41538.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41538.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
- [ ] 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41538/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/41538/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41537
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41537/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41537/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41537/events
|
https://github.com/huggingface/transformers/pull/41537
| 3,508,788,137
|
PR_kwDOCUB6oc6tZV0L
| 41,537
|
examples/rag from original paper issue
|
{
"login": "pratap834",
"id": 117664342,
"node_id": "U_kgDOBwNqVg",
"avatar_url": "https://avatars.githubusercontent.com/u/117664342?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/pratap834",
"html_url": "https://github.com/pratap834",
"followers_url": "https://api.github.com/users/pratap834/followers",
"following_url": "https://api.github.com/users/pratap834/following{/other_user}",
"gists_url": "https://api.github.com/users/pratap834/gists{/gist_id}",
"starred_url": "https://api.github.com/users/pratap834/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/pratap834/subscriptions",
"organizations_url": "https://api.github.com/users/pratap834/orgs",
"repos_url": "https://api.github.com/users/pratap834/repos",
"events_url": "https://api.github.com/users/pratap834/events{/privacy}",
"received_events_url": "https://api.github.com/users/pratap834/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-13T07:21:11
| 2025-10-13T07:24:45
| 2025-10-13T07:24:45
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41537",
"html_url": "https://github.com/huggingface/transformers/pull/41537",
"diff_url": "https://github.com/huggingface/transformers/pull/41537.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41537.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
- [ ] 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
|
{
"login": "pratap834",
"id": 117664342,
"node_id": "U_kgDOBwNqVg",
"avatar_url": "https://avatars.githubusercontent.com/u/117664342?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/pratap834",
"html_url": "https://github.com/pratap834",
"followers_url": "https://api.github.com/users/pratap834/followers",
"following_url": "https://api.github.com/users/pratap834/following{/other_user}",
"gists_url": "https://api.github.com/users/pratap834/gists{/gist_id}",
"starred_url": "https://api.github.com/users/pratap834/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/pratap834/subscriptions",
"organizations_url": "https://api.github.com/users/pratap834/orgs",
"repos_url": "https://api.github.com/users/pratap834/repos",
"events_url": "https://api.github.com/users/pratap834/events{/privacy}",
"received_events_url": "https://api.github.com/users/pratap834/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41537/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/41537/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41536
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41536/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41536/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41536/events
|
https://github.com/huggingface/transformers/pull/41536
| 3,507,694,564
|
PR_kwDOCUB6oc6tVttx
| 41,536
|
[Qwen3VL] fix device mismatch error for FSDP2 training
|
{
"login": "HollowMan6",
"id": 43995067,
"node_id": "MDQ6VXNlcjQzOTk1MDY3",
"avatar_url": "https://avatars.githubusercontent.com/u/43995067?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HollowMan6",
"html_url": "https://github.com/HollowMan6",
"followers_url": "https://api.github.com/users/HollowMan6/followers",
"following_url": "https://api.github.com/users/HollowMan6/following{/other_user}",
"gists_url": "https://api.github.com/users/HollowMan6/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HollowMan6/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HollowMan6/subscriptions",
"organizations_url": "https://api.github.com/users/HollowMan6/orgs",
"repos_url": "https://api.github.com/users/HollowMan6/repos",
"events_url": "https://api.github.com/users/HollowMan6/events{/privacy}",
"received_events_url": "https://api.github.com/users/HollowMan6/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-12T18:46:38
| 2025-10-14T10:29:06
| 2025-10-14T10:28:25
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41536",
"html_url": "https://github.com/huggingface/transformers/pull/41536",
"diff_url": "https://github.com/huggingface/transformers/pull/41536.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41536.patch",
"merged_at": "2025-10-14T10:28:25"
}
|
# 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 -->
For FSDP2, parameters might be on a meta device, and the weight.device attribute may not accurately reflect where the actual computation will happen during forward passes.
```log
File "transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", line 776, in forward
pos_embeds = self.fast_pos_embed_interpolate(grid_thw)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", line 745, in fast_pos_embed_interpolate
pos_embeds = self.pos_embed(idx_tensor) * weight_tensor[:, :, None]
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "torch/nn/modules/module.py", line 1879, in _call_impl
return inner()
^^^^^^^
File "torch/nn/modules/module.py", line 1827, in inner
result = forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "torch/nn/modules/sparse.py", line 192, in forward
return F.embedding(
^^^^^^^^^^^^
File "torch/nn/functional.py", line 2546, in embedding
return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Expected all tensors to be on the same device, but got index is on cpu, different from other tensors on cuda:0 (when checking argument in method wrapper_CUDA__index_select)
```
https://github.com/volcengine/verl/pull/3686#issuecomment-3380981817
Since the device for grid_thw is pretty much dependent on the user-side implementation (passed as a parameter for the forward method), I think it's better to take the device of grid_thw for unifying the device of idx_tensor and weight_tensor, so that user-side implementation can have more control over this and to guarantee nothing can go wrong here. User-side code can ensure the input grid is on the same device as the positional embedding weight, as the size of grid_thw is small, so there shouldn't be too much overhead. This is tested on [verl](https://github.com/volcengine/verl) and worked fine.
## 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
@yonigozlan @molbap @ArthurZucker @Cyrilvallez @zucchini-nlp
|
{
"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/41536/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/41536/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41535
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41535/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41535/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41535/events
|
https://github.com/huggingface/transformers/pull/41535
| 3,507,692,154
|
PR_kwDOCUB6oc6tVtX1
| 41,535
|
[Qwen3VL] fix: hidden_states in place modification error
|
{
"login": "HollowMan6",
"id": 43995067,
"node_id": "MDQ6VXNlcjQzOTk1MDY3",
"avatar_url": "https://avatars.githubusercontent.com/u/43995067?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HollowMan6",
"html_url": "https://github.com/HollowMan6",
"followers_url": "https://api.github.com/users/HollowMan6/followers",
"following_url": "https://api.github.com/users/HollowMan6/following{/other_user}",
"gists_url": "https://api.github.com/users/HollowMan6/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HollowMan6/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HollowMan6/subscriptions",
"organizations_url": "https://api.github.com/users/HollowMan6/orgs",
"repos_url": "https://api.github.com/users/HollowMan6/repos",
"events_url": "https://api.github.com/users/HollowMan6/events{/privacy}",
"received_events_url": "https://api.github.com/users/HollowMan6/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-12T18:45:09
| 2025-10-13T08:53:58
| 2025-10-13T08:50:14
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41535",
"html_url": "https://github.com/huggingface/transformers/pull/41535",
"diff_url": "https://github.com/huggingface/transformers/pull/41535.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41535.patch",
"merged_at": "2025-10-13T08:50:14"
}
|
# 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.
-->
```
File "transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", line 941, in forward
hidden_states = self._deepstack_process(
^^^^^^^^^^^^^^^^^^^^^^^^
File "transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", line 960, in _deepstack_process
hidden_states[visual_pos_masks, :] = local_this
~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Output 0 of SliceBackward0 is a view and is being modified inplace. This view was created inside a custom Function (or because an input was returned as-is) and the autograd logic to handle view+inplace would override the custom backward associated with the custom Function, leading to incorrect gradients. This behavior is forbidden. You can fix this by cloning the output of the custom Function.
```
## 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
@yonigozlan @molbap @ArthurZucker @Cyrilvallez @zucchini-nlp
|
{
"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/41535/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/41535/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41534
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41534/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41534/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41534/events
|
https://github.com/huggingface/transformers/pull/41534
| 3,507,251,407
|
PR_kwDOCUB6oc6tUNtM
| 41,534
|
Add VideoMAE video processor
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-12T14:05:59
| 2025-10-13T13:42:28
| 2025-10-13T13:42:28
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41534",
"html_url": "https://github.com/huggingface/transformers/pull/41534",
"diff_url": "https://github.com/huggingface/transformers/pull/41534.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41534.patch",
"merged_at": "2025-10-13T13:42:27"
}
|
## What does this PR do?
- add a dedicated `VideoMAEVideoProcessor` that decodes/samples videos via TorchCodec and emits `pixel_values` ready for VideoMAE models
- document the new processor alongside the existing image processor so users can discover the GPU-friendly path
- cover the processor with torchvision-gated regression tests to ensure serialization, sampling, and output naming stay stable
Fixes #41520
## 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?
- [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. (follow-up to the discussion with @zucchini-nlp on video processors)
- [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?
|
{
"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/41534/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/41534/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41533
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41533/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41533/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41533/events
|
https://github.com/huggingface/transformers/issues/41533
| 3,507,227,594
|
I_kwDOCUB6oc7RDAvK
| 41,533
|
Add_Specifical_tokens and resize_toked_embeddings result in an error
|
{
"login": "jialiangZ",
"id": 54654343,
"node_id": "MDQ6VXNlcjU0NjU0MzQz",
"avatar_url": "https://avatars.githubusercontent.com/u/54654343?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jialiangZ",
"html_url": "https://github.com/jialiangZ",
"followers_url": "https://api.github.com/users/jialiangZ/followers",
"following_url": "https://api.github.com/users/jialiangZ/following{/other_user}",
"gists_url": "https://api.github.com/users/jialiangZ/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jialiangZ/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jialiangZ/subscriptions",
"organizations_url": "https://api.github.com/users/jialiangZ/orgs",
"repos_url": "https://api.github.com/users/jialiangZ/repos",
"events_url": "https://api.github.com/users/jialiangZ/events{/privacy}",
"received_events_url": "https://api.github.com/users/jialiangZ/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 |
[] | 2025-10-12T13:50:40
| 2025-10-13T14:09:29
| 2025-10-13T14:08:57
|
NONE
| null | null | null | null |
### System Info
I want to add a few special tokens to my Qwen2.5VL model as separators, and after executing the following code, he received the following error message. I don't know how to solve this problem.
``` bash
[rank1]: Traceback (most recent call last):
[rank1]: RuntimeError: shape '[-1, 151936]' is invalid for input of size 329273399
[rank0]: Traceback (most recent call last):
[rank0]: RuntimeError: shape '[-1, 151936]' is invalid for input of size 217038339
[rank3]: Traceback (most recent call last):
[rank3]: RuntimeError: shape '[-1, 151936]' is invalid for input of size 116936799
[rank2]: Traceback (most recent call last):
[rank2]: RuntimeError: shape '[-1, 151936]' is invalid for input of size 215673318
Traceback (most recent call last):
File "/home/hk-project-p0022189/tum_yvc3016/miniconda3/envs/qwen2_5-VL/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 355, in wrapper
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
qwenvl/train/train_livecc.py FAILED
Failures:
<NO_OTHER_FAILURES>
Root Cause (first observed failure):
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
```
### 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 os
import logging
import pathlib
import torch
import transformers
import json
from typing import Dict
import shutil
import sys
from pathlib import Path
project_root = Path(__file__).parent.parent.parent
sys.path.append(str(project_root))
import qwenvl.train.trainer
from trainer import replace_qwen2_vl_attention_class
from transformers import (
Qwen2VLForConditionalGeneration,
)
from model_code.modeling_qwen2_5_vl import Qwen2_5_VLForConditionalGeneration
# from qwenvl.data.data_qwen import make_supervised_data_module
from qwenvl.data.lmm_dataset_for_batch import make_supervised_data_module
from qwenvl.train.argument import (
ModelArguments,
DataArguments,
TrainingArguments,
)
from transformers import AutoTokenizer, AutoProcessor, Qwen2VLImageProcessor, Trainer
local_rank = None
os.environ["TOKENIZERS_PARALLELISM"] = "false"
def rank0_print(*args):
if local_rank == 0:
print(*args)
def add_special_tokens_safely(tokenizer, new_tokens):
"""
安全地向 tokenizer 添加新的 special tokens，保留原有的 additional_special_tokens。
Args:
tokenizer: Hugging Face tokenizer
model: 对应的语言模型
new_tokens: list of str, 要添加的新 token
Returns:
bool: 是否有新 token 被添加
"""
# 获取当前词表中的所有 token
current_vocab = set(tokenizer.get_vocab().keys())
# 过滤出真正需要添加的 token
tokens_to_add = [t for t in new_tokens if t not in current_vocab]
if not tokens_to_add:
rank0_print("🟢 所有指定的 token 已存在于词表中，无需添加。")
return False
# 获取原有 additional_special_tokens（如 <image>, <ref> 等）
orig_special_tokens = tokenizer.special_tokens_map.get(
"additional_special_tokens", []
)
# 合并：保留原有 + 新增
updated_special_tokens = orig_special_tokens + [
t for t in tokens_to_add if t not in orig_special_tokens
]
rank0_print(f"📌 正在添加新 token: {tokens_to_add}")
rank0_print(f"🔧 更新后的 additional_special_tokens 总数: {len(updated_special_tokens)}")
# 使用 add_special_tokens API（会自动去重）
num_added = tokenizer.add_special_tokens(
{"additional_special_tokens": updated_special_tokens}
)
if num_added > 0:
rank0_print(f"✅ 成功添加 {num_added} 个新 token 到词表")
return num_added > 0
def safe_save_model_for_hf_trainer(trainer: transformers.Trainer, output_dir: str):
"""Collects the state dict and dump to disk."""
if trainer.deepspeed:
torch.cuda.synchronize()
trainer.save_model(output_dir)
return
state_dict = trainer.model.state_dict()
if trainer.args.should_save:
cpu_state_dict = {key: value.cpu() for key, value in state_dict.items()}
del state_dict
trainer._save(output_dir, state_dict=cpu_state_dict) # noqa
def set_model(model_args, model):
if model_args.tune_mm_vision:
for n, p in model.visual.named_parameters():
p.requires_grad = True
else:
for n, p in model.visual.named_parameters():
p.requires_grad = False
if model_args.tune_mm_mlp:
for n, p in model.visual.merger.named_parameters():
p.requires_grad = True
else:
for n, p in model.visual.merger.named_parameters():
p.requires_grad = False
if model_args.tune_mm_llm:
for n, p in model.model.named_parameters():
p.requires_grad = True
model.lm_head.requires_grad = True
else:
for n, p in model.model.named_parameters():
p.requires_grad = False
model.lm_head.requires_grad = False
def train(attn_implementation="flash_attention_2"):
global local_rank
parser = transformers.HfArgumentParser(
(ModelArguments, DataArguments, TrainingArguments)
)
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
local_rank = training_args.local_rank
os.makedirs(training_args.output_dir, exist_ok=True)
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
model_args.model_name_or_path,
cache_dir=training_args.cache_dir,
attn_implementation=attn_implementation,
torch_dtype=(torch.bfloat16 if training_args.bf16 else None),
)
processor = AutoProcessor.from_pretrained(
model_args.model_name_or_path,
)
tokenizer = AutoTokenizer.from_pretrained(
model_args.model_name_or_path,
cache_dir=training_args.cache_dir,
model_max_length=training_args.model_max_length,
padding_side="right",
use_fast=False,
)
data_args.image_processor = processor.image_processor
data_args.model_type = "qwen2.5vl"
if data_args.data_flatten:
replace_qwen2_vl_attention_class()
model.config.use_cache = False
if training_args.gradient_checkpointing:
if hasattr(model, "enable_input_require_grads"):
model.enable_input_require_grads()
else:
def make_inputs_require_grad(module, input, output):
output.requires_grad_(True)
model.get_input_embeddings().register_forward_hook(make_inputs_require_grad)
####################新加的
# 要添加的自定义 special tokens
# Token ID
# -----------------------
# <think> 151665
# </think> 151666
# <answer> 151667
# </answer> 151668
NEW_SPECIAL_TOKENS = ["<think>", "</think>", "<answer>", "</answer>"]
was_updated = add_special_tokens_safely(tokenizer, NEW_SPECIAL_TOKENS)
if was_updated:
model.resize_token_embeddings(len(tokenizer))
else:
rank0_print("ℹ️ 未检测到新 token 添加，继续执行...")
model.tokenizer = tokenizer # 👈 将 tokenizer 挂载到模型上
model.randK = data_args.randK # 👈 将 randK 挂载到模型上
model.randF = data_args.randF # 👈 将 randF 挂载到模型上
model.dataset_use = data_args.dataset_use # 👈 将 dataset_use 挂载到模型上
set_model(model_args, model)
# if torch.distributed.get_rank() == 0:
# model.visual.print_trainable_parameters()
# model.model.print_trainable_parameters()
processor.tokenizer = tokenizer
data_module = make_supervised_data_module(processor=processor, data_args=data_args)
trainer = Trainer(
model=model, processing_class=tokenizer, args=training_args, **data_module
)
if list(pathlib.Path(training_args.output_dir).glob("checkpoint-*")):
logging.info("checkpoint found, resume training")
trainer.train(resume_from_checkpoint=True)
else:
trainer.train()
trainer.save_state()
data_args.image_processor.save_pretrained(training_args.output_dir)
source_path = "chat_template.json"
template_path = os.path.join(training_args.output_dir, "chat_template.json")
shutil.copy2(source_path, template_path)
model.config.use_cache = True
safe_save_model_for_hf_trainer(trainer=trainer, output_dir=training_args.output_dir)
if __name__ == "__main__":
# train(attn_implementation="flash_attention_2")
train()
```
``` bash
#!/bin/bash
# Distributed training configuration
NPROC_PER_NODE=4
MASTER_ADDR=${MASTER_ADDR:-"127.0.0.1"}
MASTER_PORT=${MASTER_PORT:-$(shuf -i 20001-29999 -n 1)}
NNODES=${WORLD_SIZE:-1}
# 获取当前脚本所在的目录（例如 pe_version）
SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
# 向上两级：Qwen2_5_vl_group/
PY_ROOT="$(dirname "$(dirname "$SCRIPT_DIR")")"
# 设置 PYTHONPATH，指向你想添加的子目录
export PYTHONPATH="$PY_ROOT:$PYTHONPATH"
export VIDEO_MIN_PIXELS=78400 # 100*28*28. the minimum visual frame tokens sent to llm is 100
export FPS_MAX_FRAMES=60 # maximum number of frames for each video (480/60/2 = 4min)
export VIDEO_MAX_PIXELS=2408448 # 24576*28*28. the maximum overall video tokens sent to llm is 24k (leave 8k for language)
# DeepSpeed configuration
deepspeed=./scripts/zero3.json
# Model configuration
if [ -d "../../../../model/Qwen2.5-VL-3B-Instruct" ]; then
llm=../../../../model/Qwen2.5-VL-3B-Instruct # Using local model
else
llm=Qwen/Qwen2.5-VL-3B-Instruct # Using HuggingFace model ID
fi
# Training hyperparameters
lr=2e-5
batch_size=1
grad_accum_steps=16
# Training entry point
entry_file=qwenvl/train/train_livecc.py
# Dataset configuration (replace with public dataset names)
datasets=../../../../data/VideoEspresso/videoespresso_train_video_duration_le_40s_rewrited_completed.jsonl
# Output configuration
run_name="qwen2_5vl-batch-CoT-3B"
output_dir=./output/qwen2_5vl-batch-CoT-3B-$(date +"%Y%m%d_%H%M")
# Training arguments
args="
--deepspeed ${deepspeed} \
--model_name_or_path "${llm}" \
--dataset_use ${datasets} \
--data_flatten False \
--tune_mm_vision False \
--tune_mm_mlp True \
--tune_mm_llm True \
--bf16 True \
--output_dir ${output_dir} \
--num_train_epochs 1 \
--per_device_train_batch_size ${batch_size} \
--per_device_eval_batch_size $((batch_size*2)) \
--gradient_accumulation_steps ${grad_accum_steps} \
--max_pixels 50176 \
--min_pixels 784 \
--eval_strategy "no" \
--save_strategy "steps" \
--save_steps 100 \
--save_total_limit 1 \
--learning_rate ${lr} \
--weight_decay 0 \
--warmup_ratio 0.03 \
--max_grad_norm 1 \
--lr_scheduler_type "cosine" \
--logging_steps 1 \
--model_max_length 8192 \
--gradient_checkpointing True \
--dataloader_num_workers 16 \
--use_liger_kernel True \
--run_name ${run_name} \
--report_to none \
--randK False \
--randF False \
"
# Launch training
torchrun --nproc_per_node=${NPROC_PER_NODE} \
--master_addr=${MASTER_ADDR} \
--master_port=${MASTER_PORT} \
${entry_file} ${args}
```
### Expected behavior
How to add tokens correctly
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41533/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/41533/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/41532
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41532/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41532/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41532/events
|
https://github.com/huggingface/transformers/issues/41532
| 3,507,181,077
|
I_kwDOCUB6oc7RC1YV
| 41,532
|
where is examples/rag from original paper?
|
{
"login": "IgorKasianenko",
"id": 17688220,
"node_id": "MDQ6VXNlcjE3Njg4MjIw",
"avatar_url": "https://avatars.githubusercontent.com/u/17688220?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/IgorKasianenko",
"html_url": "https://github.com/IgorKasianenko",
"followers_url": "https://api.github.com/users/IgorKasianenko/followers",
"following_url": "https://api.github.com/users/IgorKasianenko/following{/other_user}",
"gists_url": "https://api.github.com/users/IgorKasianenko/gists{/gist_id}",
"starred_url": "https://api.github.com/users/IgorKasianenko/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/IgorKasianenko/subscriptions",
"organizations_url": "https://api.github.com/users/IgorKasianenko/orgs",
"repos_url": "https://api.github.com/users/IgorKasianenko/repos",
"events_url": "https://api.github.com/users/IgorKasianenko/events{/privacy}",
"received_events_url": "https://api.github.com/users/IgorKasianenko/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 |
[] | 2025-10-12T13:17:53
| 2025-10-17T09:34:15
| 2025-10-17T09:34:15
|
NONE
| null | null | null | null |
### System Info
https://arxiv.org/pdf/2005.11401 mentions https://github.com/huggingface/transformers/blob/main/examples/rag but it is not there. Add redirect if possible
### 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
Go to https://github.com/huggingface/transformers/blob/main/examples/rag
### Expected behavior
some example instead of 404
|
{
"login": "IgorKasianenko",
"id": 17688220,
"node_id": "MDQ6VXNlcjE3Njg4MjIw",
"avatar_url": "https://avatars.githubusercontent.com/u/17688220?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/IgorKasianenko",
"html_url": "https://github.com/IgorKasianenko",
"followers_url": "https://api.github.com/users/IgorKasianenko/followers",
"following_url": "https://api.github.com/users/IgorKasianenko/following{/other_user}",
"gists_url": "https://api.github.com/users/IgorKasianenko/gists{/gist_id}",
"starred_url": "https://api.github.com/users/IgorKasianenko/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/IgorKasianenko/subscriptions",
"organizations_url": "https://api.github.com/users/IgorKasianenko/orgs",
"repos_url": "https://api.github.com/users/IgorKasianenko/repos",
"events_url": "https://api.github.com/users/IgorKasianenko/events{/privacy}",
"received_events_url": "https://api.github.com/users/IgorKasianenko/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41532/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/41532/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/41531
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41531/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41531/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41531/events
|
https://github.com/huggingface/transformers/pull/41531
| 3,506,767,284
|
PR_kwDOCUB6oc6tSnbr
| 41,531
|
🌐 [i18n-KO] Translated `video_processor.md` to Korean
|
{
"login": "chelsseeey",
"id": 152389483,
"node_id": "U_kgDOCRVHaw",
"avatar_url": "https://avatars.githubusercontent.com/u/152389483?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/chelsseeey",
"html_url": "https://github.com/chelsseeey",
"followers_url": "https://api.github.com/users/chelsseeey/followers",
"following_url": "https://api.github.com/users/chelsseeey/following{/other_user}",
"gists_url": "https://api.github.com/users/chelsseeey/gists{/gist_id}",
"starred_url": "https://api.github.com/users/chelsseeey/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/chelsseeey/subscriptions",
"organizations_url": "https://api.github.com/users/chelsseeey/orgs",
"repos_url": "https://api.github.com/users/chelsseeey/repos",
"events_url": "https://api.github.com/users/chelsseeey/events{/privacy}",
"received_events_url": "https://api.github.com/users/chelsseeey/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-12T07:18:04
| 2025-10-29T09:17:55
| null |
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41531",
"html_url": "https://github.com/huggingface/transformers/pull/41531",
"diff_url": "https://github.com/huggingface/transformers/pull/41531.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41531.patch",
"merged_at": null
}
|
# What does this PR do?
Translated the `video_processor.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 (번역 누락/중복 검사)
- [X] Grammar Check (맞춤법 검사)
- [X] Review or Add new terms to glossary (용어 확인 및 추가)
- [X] Check Inline TOC (e.g. `[[lowercased-header]]`)
- [X] Check live-preview for gotchas (live-preview로 정상작동 확인)
## Who can review? (Initial)
<!-- 1. 위 체크가 모두 완료된 뒤에만 KREW 팀원들에게 리뷰를 요청하는 아래 주석을 노출해주세요!-->
May you please review this PR?
@jungnerd, @luckyvickyricky, @chelsseeey, @skwh54, @maximizemaxwell, @D15M4S
## 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?
- [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? (Final)
<!-- 2. KREW 팀원들의 리뷰가 끝난 후에 아래 주석을 노출해주세요! -->
<!---transformers, course는 @stevhliu, agent-course는 @sergiopanieg smol-agents는 @albertvillanova입니다!--->
@stevhliu May you please review this PR?
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41531/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/41531/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41530
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41530/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41530/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41530/events
|
https://github.com/huggingface/transformers/pull/41530
| 3,506,748,189
|
PR_kwDOCUB6oc6tSjmw
| 41,530
|
Fix: Correct loss normalization in training_step for multi-GPU training
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-12T06:51:55
| 2025-10-13T14:02:31
| 2025-10-13T14:02:30
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41530",
"html_url": "https://github.com/huggingface/transformers/pull/41530",
"diff_url": "https://github.com/huggingface/transformers/pull/41530.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41530.patch",
"merged_at": null
}
|
# Fix: Correct loss normalization in training_step for multi-GPU training
## Description
Fixes #37474
This PR corrects the loss aggregation logic in `Trainer.training_step` when training with multiple GPUs.
## Problem
When `num_items_in_batch` is provided (e.g., for token-level loss normalization), each device computes loss as:
```python
per_device_loss = sum_of_losses / total_items_across_all_devices
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41530/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/41530/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41529
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41529/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41529/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41529/events
|
https://github.com/huggingface/transformers/pull/41529
| 3,506,734,244
|
PR_kwDOCUB6oc6tShCa
| 41,529
|
Add num_logits_to_keep to GPT-2 and improve Llama implementation
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-12T06:31:47
| 2025-10-13T13:47:44
| 2025-10-13T13:47:44
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41529",
"html_url": "https://github.com/huggingface/transformers/pull/41529",
"diff_url": "https://github.com/huggingface/transformers/pull/41529.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41529.patch",
"merged_at": null
}
|
# Add num_logits_to_keep to GPT-2 and improve Llama implementation
## Description
Adds `num_logits_to_keep` parameter to GPT-2 for memory-efficient generation and simplifies the implementation in Llama.
## Changes
### GPT-2
- ✅ Added `num_logits_to_keep` parameter to `GPT2LMHeadModel.forward()`
- ✅ Only computes logits for last N tokens when specified
- ✅ Default behavior unchanged (computes all logits when `num_logits_to_keep=0`)
- ✅ Added parameter documentation
### Llama
- ✅ Simplified `num_logits_to_keep` implementation
- ✅ Removed redundant slicing operation
- ✅ Fixed edge case where `logits_to_keep=0` would incorrectly slice all tokens
- ✅ Cleaner, more maintainable code
## Motivation
During generation, models typically compute logits for ALL tokens in the sequence. However, for next-token prediction, we only need the last token's logits. This parameter allows:
- 🚀 **Memory efficiency**: Reduce memory usage during generation
- ⚡ **Speed**: Less computation needed
- 🎯 **Flexibility**: Users can choose how many logits to keep
## Usage Example
```python
from transformers import GPT2LMHeadModel, GPT2Tokenizer
model = GPT2LMHeadModel.from_pretrained("gpt2")
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
inputs = tokenizer("Hello, I'm a language model", return_tensors="pt")
# Default: compute all logits
outputs = model(**inputs)
print(outputs.logits.shape) # [1, sequence_length, vocab_size]
# Memory efficient: only compute last token's logits
outputs = model(**inputs, num_logits_to_keep=1)
print(outputs.logits.shape) # [1, 1, vocab_size]
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41529/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/41529/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41528
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41528/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41528/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41528/events
|
https://github.com/huggingface/transformers/pull/41528
| 3,506,719,764
|
PR_kwDOCUB6oc6tSeX6
| 41,528
|
Add position encoding interpolation to DeiT
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-12T06:08:48
| 2025-10-13T13:54:12
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41528",
"html_url": "https://github.com/huggingface/transformers/pull/41528",
"diff_url": "https://github.com/huggingface/transformers/pull/41528.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41528.patch",
"merged_at": null
}
|
# Add position encoding interpolation to DeiT
## Description
Adds position encoding interpolation to DeiT models, enabling the use of pretrained checkpoints on images with different resolutions than those used during training.
Contributes to #30579
## Changes
- ✅ Added `interpolate_pos_encoding` parameter to `DeiTModel`, `DeiTForImageClassification`, `DeiTForImageClassificationWithTeacher`, and `DeiTForMaskedImageModeling`
- ✅ Implemented `interpolate_pos_encoding()` method in `DeiTEmbeddings` class
- ✅ Properly handles both CLS and distillation tokens (unique to DeiT)
- ✅ Added comprehensive test `test_model_with_different_image_size`
- ✅ All tests passing
## Implementation Details
DeiT has two special tokens (CLS + distillation), unlike ViT which only has one (CLS). The interpolation method:
1. Separates the CLS/distillation token embeddings from patch embeddings
2. Interpolates patch embeddings to match new image size
3. Recombines CLS/distillation tokens with interpolated patches
## Testing
```python
# Example usage
model = DeiTModel.from_pretrained("facebook/deit-base-distilled-patch16-224")
# Use with 480x480 images instead of 224x224
outputs = model(large_images, interpolate_pos_encoding=True)
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41528/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/41528/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41527
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41527/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41527/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41527/events
|
https://github.com/huggingface/transformers/pull/41527
| 3,506,624,178
|
PR_kwDOCUB6oc6tSMu4
| 41,527
|
🌐 [i18n-KO] Translated selecting.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
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-12T03:33:30
| 2025-10-22T23:52:25
| null |
CONTRIBUTOR
| null | null | true
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41527",
"html_url": "https://github.com/huggingface/transformers/pull/41527",
"diff_url": "https://github.com/huggingface/transformers/pull/41527.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41527.patch",
"merged_at": null
}
|
# What does this PR do?
Translated the `selecting.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 (번역 누락/중복 검사)
- [X] Grammar Check (맞춤법 검사)
- [X] Review or Add new terms to glossary (용어 확인 및 추가)
- [X] Check Inline TOC (e.g. `[[lowercased-header]]`)
- [X] Check live-preview for gotchas (live-preview로 정상작동 확인)
## Who can review? (Initial)
<!-- 1. 위 체크가 모두 완료된 뒤에만 KREW 팀원들에게 리뷰를 요청하는 아래 주석을 노출해주세요!-->
May you please review this PR?
@4N3MONE, @yijun-lee, @jungnerd , @harheem
## 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#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.
- [ ] 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)
<!-- 2. KREW 팀원들의 리뷰가 끝난 후에 아래 주석을 노출해주세요! -->
<!-- @stevhliu May you please review this PR? -->
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41527/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/41527/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41526
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41526/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41526/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41526/events
|
https://github.com/huggingface/transformers/issues/41526
| 3,506,175,434
|
I_kwDOCUB6oc7Q-_3K
| 41,526
|
Eager Attention Works with OneVision 0.5B but Not with 7B
|
{
"login": "varungupta31",
"id": 51288316,
"node_id": "MDQ6VXNlcjUxMjg4MzE2",
"avatar_url": "https://avatars.githubusercontent.com/u/51288316?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/varungupta31",
"html_url": "https://github.com/varungupta31",
"followers_url": "https://api.github.com/users/varungupta31/followers",
"following_url": "https://api.github.com/users/varungupta31/following{/other_user}",
"gists_url": "https://api.github.com/users/varungupta31/gists{/gist_id}",
"starred_url": "https://api.github.com/users/varungupta31/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/varungupta31/subscriptions",
"organizations_url": "https://api.github.com/users/varungupta31/orgs",
"repos_url": "https://api.github.com/users/varungupta31/repos",
"events_url": "https://api.github.com/users/varungupta31/events{/privacy}",
"received_events_url": "https://api.github.com/users/varungupta31/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 |
[] | 2025-10-11T17:45:23
| 2025-10-13T07:13:16
| 2025-10-13T07:13:16
|
NONE
| null | null | null | null |
### System Info
I am trying to extract attention weights from the model and thus need to use `eager` implementation.
7B fails gives garbage output,
```python
import requests
from PIL import Image
import torch
from transformers import AutoProcessor, LlavaOnevisionForConditionalGeneration
model_id = "llava-hf/llava-onevision-qwen2-7b-ov-hf"
model = LlavaOnevisionForConditionalGeneration.from_pretrained(
model_id,
torch_dtype=torch.float16,
low_cpu_mem_usage=True,
attn_implementation='eager',
device_map='auto'
)
processor = AutoProcessor.from_pretrained(model_id)
# Define a chat history and use `apply_chat_template` to get correctly formatted prompt
# Each value in "content" has to be a list of dicts with types ("text", "image")
conversation = [
{
"role": "user",
"content": [
{"type": "text", "text": "What are these?"},
{"type": "image"},
],
},
]
prompt = processor.apply_chat_template(conversation, add_generation_prompt=True)
image_file = "http://images.cocodataset.org/val2017/000000039769.jpg"
raw_image = Image.open(requests.get(image_file, stream=True).raw)
inputs = processor(images=raw_image, text=prompt, return_tensors='pt').to(model.device, torch.float16)
output = model.generate(**inputs, max_new_tokens=200, do_sample=False)
print(processor.decode(output[0][2:], skip_special_tokens=True))
```
Output
```bash
What are these?assistant
!
```
Changing,
model from,
```
model_id = "llava-hf/llava-onevision-qwen2-7b-ov-hf"
```
to
```
model_id = "llava-hf/llava-onevision-qwen2-0.5b-ov-hf"
```
Works,
```bash
What are these?assistant
The image shows two cats, one larger and one smaller, lying on a pink blanket. The larger cat is a tabby with a striped coat, while the smaller cat is a calico with a mix of white, black, and brown fur. They are both resting comfortably on a pink blanket, with the larger cat on the left and the smaller cat on the right.
```
I am actually interested in doing attention extraction - which seems tricky with `sdpa` and flash attention, right?
Can you please help me fix this? - why is this happening and if this is too problematic to fix, how do I get attention weights at each generation step?
Thanks!
@zucchini-nlp likely has an answer on this, I would really appreciate your time and inputs, 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
Run the code above - model _works_ but gives garbage outputs, as shown.
Changing model to 0.5B variant with everything else constant, works as expected.
### Expected behavior
Consisent outputs (non-garbage) when running Onevision 7B with eager attention.
|
{
"login": "varungupta31",
"id": 51288316,
"node_id": "MDQ6VXNlcjUxMjg4MzE2",
"avatar_url": "https://avatars.githubusercontent.com/u/51288316?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/varungupta31",
"html_url": "https://github.com/varungupta31",
"followers_url": "https://api.github.com/users/varungupta31/followers",
"following_url": "https://api.github.com/users/varungupta31/following{/other_user}",
"gists_url": "https://api.github.com/users/varungupta31/gists{/gist_id}",
"starred_url": "https://api.github.com/users/varungupta31/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/varungupta31/subscriptions",
"organizations_url": "https://api.github.com/users/varungupta31/orgs",
"repos_url": "https://api.github.com/users/varungupta31/repos",
"events_url": "https://api.github.com/users/varungupta31/events{/privacy}",
"received_events_url": "https://api.github.com/users/varungupta31/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41526/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/41526/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/41525
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41525/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41525/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41525/events
|
https://github.com/huggingface/transformers/pull/41525
| 3,506,137,270
|
PR_kwDOCUB6oc6tQvkf
| 41,525
|
Fixed Type-hints in function defintions
|
{
"login": "Sai-Suraj-27",
"id": 87087741,
"node_id": "MDQ6VXNlcjg3MDg3NzQx",
"avatar_url": "https://avatars.githubusercontent.com/u/87087741?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Sai-Suraj-27",
"html_url": "https://github.com/Sai-Suraj-27",
"followers_url": "https://api.github.com/users/Sai-Suraj-27/followers",
"following_url": "https://api.github.com/users/Sai-Suraj-27/following{/other_user}",
"gists_url": "https://api.github.com/users/Sai-Suraj-27/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Sai-Suraj-27/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Sai-Suraj-27/subscriptions",
"organizations_url": "https://api.github.com/users/Sai-Suraj-27/orgs",
"repos_url": "https://api.github.com/users/Sai-Suraj-27/repos",
"events_url": "https://api.github.com/users/Sai-Suraj-27/events{/privacy}",
"received_events_url": "https://api.github.com/users/Sai-Suraj-27/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-11T17:10:42
| 2025-10-13T09:48:38
| 2025-10-13T09:48:37
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41525",
"html_url": "https://github.com/huggingface/transformers/pull/41525",
"diff_url": "https://github.com/huggingface/transformers/pull/41525.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41525.patch",
"merged_at": "2025-10-13T09:48:37"
}
|
# What does this PR do?
Although [RUF013](https://docs.astral.sh/ruff/rules/implicit-optional/#implicit-optional-ruf013) rule is [enabled](https://github.com/huggingface/transformers/blob/3927ffed31e3c0d2929bf98bd05b7c61fcc48b62/pyproject.toml#L35), it doesn't catch when we are using custom types ([known issue](https://github.com/astral-sh/ruff/issues/14018?utm_source=chatgpt.com)). Used [Pyrefly](https://github.com/facebook/pyrefly) to find these.
## 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?
@zucchini-nlp @molbap
|
{
"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/41525/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/41525/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41524
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41524/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41524/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41524/events
|
https://github.com/huggingface/transformers/pull/41524
| 3,506,049,723
|
PR_kwDOCUB6oc6tQdOg
| 41,524
|
Add max_eval_batches argument to TrainingArguments
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-11T16:08:35
| 2025-10-13T13:56:14
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41524",
"html_url": "https://github.com/huggingface/transformers/pull/41524",
"diff_url": "https://github.com/huggingface/transformers/pull/41524.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41524.patch",
"merged_at": null
}
|
# Add max_eval_batches argument to TrainingArguments
## Description
Adds a `max_eval_batches` parameter to `TrainingArguments` that allows users to limit the number of batches used during evaluation.
Fixes #31561
## Motivation
When working with large evaluation datasets, running evaluation on the entire dataset can be very slow. During development, hyperparameter tuning, or quick iteration, it's often sufficient to evaluate on a subset of the data.
This is similar to PyTorch Lightning's `limit_val_batches` parameter.
## Changes
- ✅ Added `max_eval_batches` parameter to `TrainingArguments`
- ✅ Implemented batch limiting in `Trainer.evaluation_loop`
- ✅ Added test coverage
- ✅ Added documentation in parameter metadata
## Usage Example
```python
from transformers import Trainer, TrainingArguments
training_args = TrainingArguments(
output_dir="./output",
evaluation_strategy="steps",
eval_steps=100,
max_eval_batches=50, # Only evaluate on 50 batches
)
trainer = Trainer(
model=model,
args=training_args,
eval_dataset=eval_dataset,
)
# Evaluation will stop after 50 batches instead of going through entire dataset
trainer.evaluate()
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41524/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/41524/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41523
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41523/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41523/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41523/events
|
https://github.com/huggingface/transformers/pull/41523
| 3,505,936,204
|
PR_kwDOCUB6oc6tQFYu
| 41,523
|
Add test coverage for ConvNextImageProcessorFast
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-11T14:52:04
| 2025-10-11T14:53:14
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41523",
"html_url": "https://github.com/huggingface/transformers/pull/41523",
"diff_url": "https://github.com/huggingface/transformers/pull/41523.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41523.patch",
"merged_at": null
}
|
# Add test coverage for ConvNextImageProcessorFast
## Description
Completes test coverage for the existing `ConvNextImageProcessorFast` implementation by adding `image_processor_list` to enable testing both slow and fast processors.
## Changes
- ✅ Added `image_processor_list = [image_processing_class, fast_image_processing_class]` to test class
- ✅ Enables all existing tests to run on both slow and fast processors
- ✅ All 19 tests passing
## Testing Results
```bash
RUN_SLOW=1 python -m pytest tests/models/convnext/test_image_processing_convnext.py -v
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41523/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/41523/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41522
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41522/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41522/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41522/events
|
https://github.com/huggingface/transformers/pull/41522
| 3,505,830,231
|
PR_kwDOCUB6oc6tPvkv
| 41,522
|
Fix _init_weights to safely skip int8 quantized weights
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-11T13:28:13
| 2025-10-13T13:57:22
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41522",
"html_url": "https://github.com/huggingface/transformers/pull/41522",
"diff_url": "https://github.com/huggingface/transformers/pull/41522.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41522.patch",
"merged_at": null
}
|
This PR addresses a RuntimeError that occurs when loading W8A8 quantized Qwen2.5-VL models.
Changes:
- Modified `_init_weights` in `Qwen2_5_VLPreTrainedModel` to safely skip non-floating tensors (e.g., int8 quantized weights) during initialization.
- Ensures float tensors are initialized normally, while int8 or other non-float tensors are ignored, preventing errors from `normal_()` on integer dtypes.
- Improves compatibility for loading quantized models without vLLM and allows safe custom hooks for research purposes.
Note: This PR does not include the tester file `test_init_weights_safe.py`; that file exists only in the fork for private testing.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41522/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/41522/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41521
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41521/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41521/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41521/events
|
https://github.com/huggingface/transformers/pull/41521
| 3,505,718,903
|
PR_kwDOCUB6oc6tPXlB
| 41,521
|
Fix forced_bos_token_id not set in generation_config
|
{
"login": "Addyk-24",
"id": 174926659,
"node_id": "U_kgDOCm0rQw",
"avatar_url": "https://avatars.githubusercontent.com/u/174926659?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Addyk-24",
"html_url": "https://github.com/Addyk-24",
"followers_url": "https://api.github.com/users/Addyk-24/followers",
"following_url": "https://api.github.com/users/Addyk-24/following{/other_user}",
"gists_url": "https://api.github.com/users/Addyk-24/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Addyk-24/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Addyk-24/subscriptions",
"organizations_url": "https://api.github.com/users/Addyk-24/orgs",
"repos_url": "https://api.github.com/users/Addyk-24/repos",
"events_url": "https://api.github.com/users/Addyk-24/events{/privacy}",
"received_events_url": "https://api.github.com/users/Addyk-24/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-11T12:19:20
| 2025-10-15T10:03:08
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41521",
"html_url": "https://github.com/huggingface/transformers/pull/41521",
"diff_url": "https://github.com/huggingface/transformers/pull/41521.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41521.patch",
"merged_at": null
}
|
# What does this PR do?
Fixes #41492
Fixes incorrect target language generation during evaluation/validation in run_translation.py for multilingual translation models (mBART , M2M100).
### Problem
When fine-tuning multilingual models, forced_bos_token_id was only set in model.config but not in model.generation_config. During evaluation, model.generate() reads from generation_config, causing generation in wrong language and artificially low BLEU scores.Previously would be ~2-5 (wrong language)
### Solution
Set forced_bos_token_id in both model.config and model.generation_config.
Results:
<img width="275" height="112" alt="eval_metrics" src="https://github.com/user-attachments/assets/5b499354-bdbc-4eb0-b07a-fcb75cc2690f" />
- ✅ this generated correct language token id with correct target language.
- ✅ This warning appears if you modify model.config directly for generation. Using model.generation_config removes this warning and ensures Transformers v5+ uses the setting correctly.
- ✅ All evaluations complete without errors
- ✅ Using decoder_start_token_id only/both causes empty outputs.
- ✅ With this fix, the target language ID is automatically handled during generation.
## 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?
- [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?
@zach-huggingface @CyrilVallez
|
{
"login": "Addyk-24",
"id": 174926659,
"node_id": "U_kgDOCm0rQw",
"avatar_url": "https://avatars.githubusercontent.com/u/174926659?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Addyk-24",
"html_url": "https://github.com/Addyk-24",
"followers_url": "https://api.github.com/users/Addyk-24/followers",
"following_url": "https://api.github.com/users/Addyk-24/following{/other_user}",
"gists_url": "https://api.github.com/users/Addyk-24/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Addyk-24/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Addyk-24/subscriptions",
"organizations_url": "https://api.github.com/users/Addyk-24/orgs",
"repos_url": "https://api.github.com/users/Addyk-24/repos",
"events_url": "https://api.github.com/users/Addyk-24/events{/privacy}",
"received_events_url": "https://api.github.com/users/Addyk-24/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41521/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/41521/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41520
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41520/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41520/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41520/events
|
https://github.com/huggingface/transformers/issues/41520
| 3,505,415,635
|
I_kwDOCUB6oc7Q8GXT
| 41,520
|
Add GPU-Accelerated Video Decoding for VideoMAE, TimeSformer, and Similar Models
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/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 |
[] | 2025-10-11T08:15:22
| 2025-10-13T13:42:29
| 2025-10-13T13:42:29
|
CONTRIBUTOR
| null | null | null | null |
### Feature request
I’d like to propose adding a lightweight, GPU-friendly video decoding utility to Transformers.
Right now, models like VideoMAE and TimeSformer depend on PIL or OpenCV for frame extraction, which keeps preprocessing on the CPU and slows down training and inference. A unified VideoReader module that wraps libraries such as PyAV or Decord could decode and batch frames directly as CUDA tensors, eliminating a major bottleneck in current video workflows and making it easier for users to stream large clips efficiently into Transformer-based video and multimodal models.
### Motivation
Most video pipelines in Transformers still rely on CPU-bound decoding through PIL or OpenCV.
For large-batch or long-clip training, this quickly becomes the main bottleneck — GPU utilization stays low while the CPU decodes frames one by one.
Libraries like PyAV and Decord already provide fast, batched video loading that can hand off tensors directly to the GPU.
Integrating this capability natively would make training and inference with models such as VideoMAE, TimeSformer, or InternVideo far more efficient.
### Your contribution
Love working in this space, if given an green sign
Will start working on it
|
{
"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/41520/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/41520/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/41519
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41519/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41519/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41519/events
|
https://github.com/huggingface/transformers/issues/41519
| 3,505,413,795
|
I_kwDOCUB6oc7Q8F6j
| 41,519
|
Add Shared Vision–Language Negative Sampling Utility for BLIP / LLaVA / IDEFICS
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/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 |
[] | 2025-10-11T08:13:34
| 2025-10-13T13:58:31
| null |
CONTRIBUTOR
| null | null | null | null |
### Feature request
Add a unified vision-language negative sampling utility for contrastive learning. Support hard negative mining and in-batch negative sampling. Compatible with BLIP, LLaVA, and IDEFICS training pipelines. Modular design for easy integration into existing VLM frameworks
### Motivation
VLM training still relies on ad-hoc contrastive negative sampling code.
A common utility to batch image–text negatives (with optional hard-negative mining) would simplify research pipelines and encourage standardization.
### Your contribution
If given a approval, will start working on this
Contributor: [@Aki-07](https://github.com/Aki-07)
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41519/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/41519/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/41518
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41518/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41518/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41518/events
|
https://github.com/huggingface/transformers/issues/41518
| 3,505,411,105
|
I_kwDOCUB6oc7Q8FQh
| 41,518
|
Add Structured Prompt Templates Registry for LLM / VLM / Diffusion Tasks
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/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 |
[] | 2025-10-11T08:10:20
| 2025-10-13T15:06:20
| null |
CONTRIBUTOR
| null | null | null | null |
### Feature request
Introduce transformers.prompt_templates — a YAML-based registry and accessor API:
```
from transformers import PromptTemplates
PromptTemplates.get("summarization") # "Summarize the following text:"
PromptTemplates.list_tasks() # ["summarization","vqa","ocr",...]
```
- Templates stored as yaml/json under src/transformers/prompt_templates/templates/.
- Accessor + validation in registry.py.
- Optional CLI command transformers-cli list-prompts.
- Pipelines can import a template by task name instead of hard-coding.
### Motivation
Every pipeline and model today embeds its own prompt strings (e.g., summarization, OCR, VQA).
This duplication makes results inconsistent and hard to benchmark.
A central registry of task-specific prompt templates would unify defaults and enable easy community additions.
### Your contribution
I’ll implement the registry module, add unit tests and docs, and migrate 1–2 pipelines (summarization / captioning) to use it.
Contributor: [@Aki-07](https://github.com/Aki-07)
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41518/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/41518/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/41517
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41517/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41517/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41517/events
|
https://github.com/huggingface/transformers/issues/41517
| 3,505,388,270
|
I_kwDOCUB6oc7Q7_ru
| 41,517
|
Add MobileNetV2 Fast Image Processor
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/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 |
[] | 2025-10-11T07:55:14
| 2025-10-11T07:57:55
| 2025-10-11T07:57:55
|
CONTRIBUTOR
| null | null | null | null |
### Feature request
Add a fast image processor for **MobileNetV2** to bring it in line with other vision models (e.g., ViT, ConvNeXt) that already have fast variants.
This will improve performance for all MobileNetV2-based checkpoints and multimodal processors that depend on it.
### Scope
- Mirror current preprocessing steps (resize, crop, normalize).
- Base on existing `BaseImageProcessorFast` class.
- Register it in the model’s auto mapping for automatic detection.
### Motivation
MobileNetV2 is one of the most widely used lightweight vision models — both as a standalone backbone and inside multimodal processors.
However, it still lacks a Fast Image Processor implementation.
Adding MobileNetV2ImageProcessorFast will:
- Enable torch-native preprocessing, avoiding PIL bottlenecks.
- Significantly reduce preprocessing latency in training and inference pipelines.
- Improve GPU utilization, especially in batched or streaming inference.
- Ensure feature parity with other vision models (ViT, ConvNeXt, ResNet) that already provide fast variants.
This improvement benefits millions of downstream users and deployments that rely on MobileNetV2-based models, while keeping the implementation lightweight and consistent with Hugging Face’s design for fast processors.
### Your contribution
My planned steps:
- Create the new MobileNetV2ImageProcessorFast based on the existing BaseImageProcessorFast design pattern.
- Ensure full feature parity with the current (slow) MobileNetV2 image processor — including resizing, cropping, normalization, and label reduction for segmentation tasks.
- Register the processor in the MobileNetV2 module and auto mapping so it’s automatically used by any MobileNetV2 checkpoint.
- Add/update tests to validate consistency between the slow and fast versions.
Contributor: @Aki-07
Happy to take full ownership of this feature once approved.
|
{
"login": "Aki-07",
"id": 95642646,
"node_id": "U_kgDOBbNkFg",
"avatar_url": "https://avatars.githubusercontent.com/u/95642646?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Aki-07",
"html_url": "https://github.com/Aki-07",
"followers_url": "https://api.github.com/users/Aki-07/followers",
"following_url": "https://api.github.com/users/Aki-07/following{/other_user}",
"gists_url": "https://api.github.com/users/Aki-07/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Aki-07/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Aki-07/subscriptions",
"organizations_url": "https://api.github.com/users/Aki-07/orgs",
"repos_url": "https://api.github.com/users/Aki-07/repos",
"events_url": "https://api.github.com/users/Aki-07/events{/privacy}",
"received_events_url": "https://api.github.com/users/Aki-07/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41517/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/41517/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/41516
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41516/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41516/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41516/events
|
https://github.com/huggingface/transformers/issues/41516
| 3,505,305,779
|
I_kwDOCUB6oc7Q7riz
| 41,516
|
when finetune deepseek v3.1 with lora，an ValueError was reported: QuantizationMethod.FP8 do not support training
|
{
"login": "Francisapzii",
"id": 101245339,
"node_id": "U_kgDOBgjhmw",
"avatar_url": "https://avatars.githubusercontent.com/u/101245339?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Francisapzii",
"html_url": "https://github.com/Francisapzii",
"followers_url": "https://api.github.com/users/Francisapzii/followers",
"following_url": "https://api.github.com/users/Francisapzii/following{/other_user}",
"gists_url": "https://api.github.com/users/Francisapzii/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Francisapzii/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Francisapzii/subscriptions",
"organizations_url": "https://api.github.com/users/Francisapzii/orgs",
"repos_url": "https://api.github.com/users/Francisapzii/repos",
"events_url": "https://api.github.com/users/Francisapzii/events{/privacy}",
"received_events_url": "https://api.github.com/users/Francisapzii/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
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-11T06:43:50
| 2025-10-16T18:38:56
| null |
NONE
| null | null | null | null |
### System Info
The error was as below. Please help.
Traceback (most recent call last):
File "/hail/cjs/finetune/ft0/lora.py", line 231, in <module>
model, tokenizer = finetune_with_lora(model_name, output_dir)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/hail/cjs/finetune/ft0/lora.py", line 156, in finetune_with_lora
trainer = Trainer(
^^^^^^^^
File "/hail/cjs/anaconda3/envs/lora1/lib/python3.11/site-packages/transformers/utils/deprecation.py", line 172, in wrapped_func
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/hail/cjs/anaconda3/envs/lora1/lib/python3.11/site-packages/transformers/trainer.py", line 572, in __init__
raise ValueError(
ValueError: The model you are trying to fine-tune is quantized with QuantizationMethod.FP8 but that quantization method do not support training. Please open an issue on GitHub: https://github.com/huggingface/transformers to request the support for training support for QuantizationMethod.FP8
### 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
Finetune "deepseek-ai/DeepSeek-V3.1-Terminus" with lora, this error will happen.
### Expected behavior
correct this error, please.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41516/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/41516/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/41515
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41515/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41515/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41515/events
|
https://github.com/huggingface/transformers/pull/41515
| 3,504,483,436
|
PR_kwDOCUB6oc6tLSUv
| 41,515
|
[ci] Disable workflows with secrets and custom runners to run on fork
|
{
"login": "HollowMan6",
"id": 43995067,
"node_id": "MDQ6VXNlcjQzOTk1MDY3",
"avatar_url": "https://avatars.githubusercontent.com/u/43995067?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HollowMan6",
"html_url": "https://github.com/HollowMan6",
"followers_url": "https://api.github.com/users/HollowMan6/followers",
"following_url": "https://api.github.com/users/HollowMan6/following{/other_user}",
"gists_url": "https://api.github.com/users/HollowMan6/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HollowMan6/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HollowMan6/subscriptions",
"organizations_url": "https://api.github.com/users/HollowMan6/orgs",
"repos_url": "https://api.github.com/users/HollowMan6/repos",
"events_url": "https://api.github.com/users/HollowMan6/events{/privacy}",
"received_events_url": "https://api.github.com/users/HollowMan6/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T20:55:17
| 2025-10-16T14:59:25
| 2025-10-16T14:59:25
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41515",
"html_url": "https://github.com/huggingface/transformers/pull/41515",
"diff_url": "https://github.com/huggingface/transformers/pull/41515.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41515.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 -->
Skip those workflows by checking if the repository is owned by `huggingface`.
Tested by directly pushing to the fork: https://github.com/HollowMan6/transformers/commit/14be7f3c250c3ba1887877fc1df67b2e87559759
and opening a PR on the fork side:
- https://github.com/HollowMan6/transformers/pull/1
## 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
@ydshieh
|
{
"login": "HollowMan6",
"id": 43995067,
"node_id": "MDQ6VXNlcjQzOTk1MDY3",
"avatar_url": "https://avatars.githubusercontent.com/u/43995067?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/HollowMan6",
"html_url": "https://github.com/HollowMan6",
"followers_url": "https://api.github.com/users/HollowMan6/followers",
"following_url": "https://api.github.com/users/HollowMan6/following{/other_user}",
"gists_url": "https://api.github.com/users/HollowMan6/gists{/gist_id}",
"starred_url": "https://api.github.com/users/HollowMan6/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/HollowMan6/subscriptions",
"organizations_url": "https://api.github.com/users/HollowMan6/orgs",
"repos_url": "https://api.github.com/users/HollowMan6/repos",
"events_url": "https://api.github.com/users/HollowMan6/events{/privacy}",
"received_events_url": "https://api.github.com/users/HollowMan6/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41515/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/41515/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41514
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41514/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41514/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41514/events
|
https://github.com/huggingface/transformers/pull/41514
| 3,503,050,135
|
PR_kwDOCUB6oc6tGWoJ
| 41,514
|
delete some tokenizer tests using pickle
|
{
"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 |
[] | 2025-10-10T13:54:34
| 2025-10-14T12:50:53
| 2025-10-14T12:50:52
|
COLLABORATOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41514",
"html_url": "https://github.com/huggingface/transformers/pull/41514",
"diff_url": "https://github.com/huggingface/transformers/pull/41514.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41514.patch",
"merged_at": "2025-10-14T12:50:52"
}
|
# What does this PR do?
There is no room for `pickle` within `transformers`!
|
{
"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/41514/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/41514/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41513
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41513/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41513/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41513/events
|
https://github.com/huggingface/transformers/pull/41513
| 3,502,993,717
|
PR_kwDOCUB6oc6tGKkr
| 41,513
|
Remove references to AutoModelForVision2Seq
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T13:41:07
| 2025-10-13T16:00:10
| 2025-10-13T16:00:08
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41513",
"html_url": "https://github.com/huggingface/transformers/pull/41513",
"diff_url": "https://github.com/huggingface/transformers/pull/41513.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41513.patch",
"merged_at": "2025-10-13T16:00:07"
}
|
Since `AutoModelForVision2Seq` is deprecated for `v5`, this PR redirects our code to point at `AutoModelForImageTextToText` to cut down on unnecessary warnings.
Fixes #41509
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41513/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/41513/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41512
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41512/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41512/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41512/events
|
https://github.com/huggingface/transformers/pull/41512
| 3,502,706,946
|
PR_kwDOCUB6oc6tFLhZ
| 41,512
|
Remove outdated flags
|
{
"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 |
[] | 2025-10-10T12:34:34
| 2025-10-10T12:43:54
| 2025-10-10T12:34:47
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41512",
"html_url": "https://github.com/huggingface/transformers/pull/41512",
"diff_url": "https://github.com/huggingface/transformers/pull/41512.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41512.patch",
"merged_at": "2025-10-10T12:34:47"
}
|
# What does this PR do?
Those features disappeared recently, but those models were added in the middle so they still have the flags
|
{
"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/41512/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/41512/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41511
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41511/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41511/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41511/events
|
https://github.com/huggingface/transformers/pull/41511
| 3,502,609,986
|
PR_kwDOCUB6oc6tE2Kn
| 41,511
|
[don't merge yet] Remove some custom datasets defined in codebase
|
{
"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
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-10T12:07:03
| 2025-10-13T12:53:08
| null |
COLLABORATOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41511",
"html_url": "https://github.com/huggingface/transformers/pull/41511",
"diff_url": "https://github.com/huggingface/transformers/pull/41511.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41511.patch",
"merged_at": null
}
|
# What does this PR do?
It should have been gone long long time ago ...
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41511/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/41511/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41510
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41510/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41510/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41510/events
|
https://github.com/huggingface/transformers/pull/41510
| 3,502,490,958
|
PR_kwDOCUB6oc6tEb_S
| 41,510
|
Fix detectron2 import
|
{
"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 |
[] | 2025-10-10T11:31:28
| 2025-10-10T13:40:59
| 2025-10-10T11:33:47
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41510",
"html_url": "https://github.com/huggingface/transformers/pull/41510",
"diff_url": "https://github.com/huggingface/transformers/pull/41510.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41510.patch",
"merged_at": "2025-10-10T11:33:47"
}
|
# What does this PR do?
Detectron2 does not uninstall correctly for some reason (probably the setup as the lib needs to be installed from source) so we need this cc @ydshieh
|
{
"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/41510/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/41510/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41509
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41509/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41509/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41509/events
|
https://github.com/huggingface/transformers/issues/41509
| 3,502,437,111
|
I_kwDOCUB6oc7QwvL3
| 41,509
|
Warning when using the "image-to-text" pipeline
|
{
"login": "intexcor",
"id": 142020129,
"node_id": "U_kgDOCHcOIQ",
"avatar_url": "https://avatars.githubusercontent.com/u/142020129?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/intexcor",
"html_url": "https://github.com/intexcor",
"followers_url": "https://api.github.com/users/intexcor/followers",
"following_url": "https://api.github.com/users/intexcor/following{/other_user}",
"gists_url": "https://api.github.com/users/intexcor/gists{/gist_id}",
"starred_url": "https://api.github.com/users/intexcor/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/intexcor/subscriptions",
"organizations_url": "https://api.github.com/users/intexcor/orgs",
"repos_url": "https://api.github.com/users/intexcor/repos",
"events_url": "https://api.github.com/users/intexcor/events{/privacy}",
"received_events_url": "https://api.github.com/users/intexcor/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 |
[] | 2025-10-10T11:14:56
| 2025-10-13T16:00:09
| 2025-10-13T16:00:09
|
NONE
| null | null | null | null |
### System Info
transformers=4.57.0
python 3.13
### Who can help?
_No response_
### Information
- [x] The official example scripts
- [x] 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
from transformers import pipeline
pipe = pipeline("image-to-text", model="intexcp/donut")
### Expected behavior
Bug fixes
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41509/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/41509/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/41508
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41508/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41508/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41508/events
|
https://github.com/huggingface/transformers/pull/41508
| 3,502,170,916
|
PR_kwDOCUB6oc6tDV69
| 41,508
|
Bump to hfh 1.0.0.rc5 to fix test
|
{
"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 |
[] | 2025-10-10T10:00:09
| 2025-10-10T10:12:10
| 2025-10-10T10:12:09
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41508",
"html_url": "https://github.com/huggingface/transformers/pull/41508",
"diff_url": "https://github.com/huggingface/transformers/pull/41508.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41508.patch",
"merged_at": "2025-10-10T10:12:09"
}
|
Should be the current failing CI.
See https://github.com/huggingface/huggingface_hub/pull/3433 for info.
(related to [internal slack thread](https://huggingface.slack.com/archives/C01NE71C4F7/p1760022191957959))
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41508/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/41508/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41507
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41507/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41507/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41507/events
|
https://github.com/huggingface/transformers/pull/41507
| 3,502,138,986
|
PR_kwDOCUB6oc6tDO8u
| 41,507
|
[kernels] rm mra kernels
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T09:49:55
| 2025-10-14T11:34:06
| 2025-10-14T11:34:04
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41507",
"html_url": "https://github.com/huggingface/transformers/pull/41507",
"diff_url": "https://github.com/huggingface/transformers/pull/41507.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41507.patch",
"merged_at": "2025-10-14T11:34:04"
}
|
# What does this PR do?
Removes the mra kernels, and uses kernels from the hub : https://huggingface.co/kernels-community/mra instead
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41507/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/41507/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41506
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41506/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41506/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41506/events
|
https://github.com/huggingface/transformers/pull/41506
| 3,502,082,825
|
PR_kwDOCUB6oc6tDCyW
| 41,506
|
[SAM] Fix typing hints
|
{
"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 |
[] | 2025-10-10T09:31:12
| 2025-10-13T09:52:00
| 2025-10-13T09:52:00
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41506",
"html_url": "https://github.com/huggingface/transformers/pull/41506",
"diff_url": "https://github.com/huggingface/transformers/pull/41506.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41506.patch",
"merged_at": "2025-10-13T09:52:00"
}
|
# What does this PR do?
As per title
|
{
"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/41506/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/41506/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41505
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41505/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41505/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41505/events
|
https://github.com/huggingface/transformers/pull/41505
| 3,502,017,115
|
PR_kwDOCUB6oc6tC06p
| 41,505
|
🚨 [v5] `generate` delegates default cache initialization to the model
|
{
"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 |
[] | 2025-10-10T09:09:27
| 2025-10-13T12:20:52
| 2025-10-13T12:20:48
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41505",
"html_url": "https://github.com/huggingface/transformers/pull/41505",
"diff_url": "https://github.com/huggingface/transformers/pull/41505.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41505.patch",
"merged_at": "2025-10-13T12:20:48"
}
|
# What does this PR do?
See PR title.
Now that all traces of legacy caches were removed, we can trust the model to initialize its own cache! This means we no longer need to set `cache_implementation="xxx"` defaults in new models, assuming the model's forward pass defaults to the right cache class.
Also fixes related bugs, uncovered by not feeding a cache to the model.
|
{
"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/41505/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/41505/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41504
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41504/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41504/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41504/events
|
https://github.com/huggingface/transformers/pull/41504
| 3,502,004,725
|
PR_kwDOCUB6oc6tCyOc
| 41,504
|
Revert `local_rank` deletion and some cleaning
|
{
"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
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T09:05:21
| 2025-10-10T10:23:06
| 2025-10-10T10:23:04
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41504",
"html_url": "https://github.com/huggingface/transformers/pull/41504",
"diff_url": "https://github.com/huggingface/transformers/pull/41504.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41504.patch",
"merged_at": "2025-10-10T10:23:04"
}
|
# What does this PR do?
This PR removes some bits that I forgot when removing `logging_dir`. Also, we need to keep `local_rank` as torch.distributed.launch inject `local_rank` in the script. I will deprecate at once torch removes it from their codebase + we don't support this version of pytorch which is in a super long time
|
{
"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/41504/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/41504/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41503
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41503/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41503/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41503/events
|
https://github.com/huggingface/transformers/pull/41503
| 3,501,937,472
|
PR_kwDOCUB6oc6tCkWT
| 41,503
|
Fix some tests
|
{
"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 |
[] | 2025-10-10T08:41:49
| 2025-10-10T09:05:11
| 2025-10-10T09:05:09
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41503",
"html_url": "https://github.com/huggingface/transformers/pull/41503",
"diff_url": "https://github.com/huggingface/transformers/pull/41503.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41503.patch",
"merged_at": "2025-10-10T09:05:09"
}
|
# What does this PR do?
This PR removes some last remnants of the old cache format, and fixes related tests. I also updated contrastive search on the hub https://huggingface.co/transformers-community/contrastive-search/discussions/3 to fix the related tests with the new format
|
{
"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/41503/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/41503/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41502
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41502/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41502/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41502/events
|
https://github.com/huggingface/transformers/pull/41502
| 3,501,735,980
|
PR_kwDOCUB6oc6tB5fn
| 41,502
|
:globe_with_meridians: [i18n-KO] Translated `ko-LFM2.md` to Korean
|
{
"login": "ssum21",
"id": 116950962,
"node_id": "U_kgDOBviHsg",
"avatar_url": "https://avatars.githubusercontent.com/u/116950962?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ssum21",
"html_url": "https://github.com/ssum21",
"followers_url": "https://api.github.com/users/ssum21/followers",
"following_url": "https://api.github.com/users/ssum21/following{/other_user}",
"gists_url": "https://api.github.com/users/ssum21/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ssum21/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ssum21/subscriptions",
"organizations_url": "https://api.github.com/users/ssum21/orgs",
"repos_url": "https://api.github.com/users/ssum21/repos",
"events_url": "https://api.github.com/users/ssum21/events{/privacy}",
"received_events_url": "https://api.github.com/users/ssum21/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T07:36:31
| 2025-10-16T18:29:04
| 2025-10-16T18:29:04
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41502",
"html_url": "https://github.com/huggingface/transformers/pull/41502",
"diff_url": "https://github.com/huggingface/transformers/pull/41502.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41502.patch",
"merged_at": "2025-10-16T18:29:04"
}
|
<!-- PR의 제목은 ":globe_with_meridians: [i18n-KO] Translated `<your_file>.md` to Korean" 으로 부탁드립니다 -->
# What does this PR do?
Translated the `lfm2.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 (번역 누락/중복 검사)
- [x] Grammar Check (맞춤법 검사)
- [x] Review or Add new terms to glossary (용어 확인 및 추가)
- [x] Check Inline TOC (e.g. `[[lowercased-header]]`)
- [x] Check live-preview for gotchas (live-preview로 정상작동 확인)
## Who can review? (Initial)
<!-- 1. 위 체크가 모두 완료된 뒤에만 KREW 팀원들에게 리뷰를 요청하는 아래 주석을 노출해주세요!-->
May you please review this PR?
@4N3MONE, @Kim-Ju-won, @ahnjj, @FacerAin, @ssum21, @TaskerJang, @HyunZ118
## 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?
- [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? (Final)
<!-- 2. KREW 팀원들의 리뷰가 끝난 후에 아래 주석을 노출해주세요! -->
<!---transformers, course는 @stevhliu, agent-course는 @sergiopanieg smol-agents는 @albertvillanova입니다!--->
@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/41502/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/41502/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41501
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41501/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41501/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41501/events
|
https://github.com/huggingface/transformers/issues/41501
| 3,501,684,923
|
I_kwDOCUB6oc7Qt3i7
| 41,501
|
Paged attention does not work
|
{
"login": "yuyijiong",
"id": 73890704,
"node_id": "MDQ6VXNlcjczODkwNzA0",
"avatar_url": "https://avatars.githubusercontent.com/u/73890704?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yuyijiong",
"html_url": "https://github.com/yuyijiong",
"followers_url": "https://api.github.com/users/yuyijiong/followers",
"following_url": "https://api.github.com/users/yuyijiong/following{/other_user}",
"gists_url": "https://api.github.com/users/yuyijiong/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yuyijiong/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yuyijiong/subscriptions",
"organizations_url": "https://api.github.com/users/yuyijiong/orgs",
"repos_url": "https://api.github.com/users/yuyijiong/repos",
"events_url": "https://api.github.com/users/yuyijiong/events{/privacy}",
"received_events_url": "https://api.github.com/users/yuyijiong/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
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-10T07:20:29
| 2025-10-16T11:58:16
| null |
NONE
| null | null | null | null |
### System Info
- `transformers` version: 4.57.0
- Platform: Linux-5.15.0-107-generic-x86_64-with-glibc2.35
- Python version: 3.12.11
- Huggingface_hub version: 0.35.3
- Safetensors version: 0.6.2
- Accelerate version: 1.10.1
- Accelerate config: - compute_environment: LOCAL_MACHINE
- distributed_type: MULTI_GPU
- mixed_precision: bf16
- use_cpu: False
- debug: False
- num_processes: 8
- 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 (accelerator?): 2.8.0+cu128 (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 H20
### Who can help?
@vasqu @ArthurZucker @CyrilVallez
### Information
- [ ] The official example scripts
- [x] 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
When set attn_implementation to "paged_attention", I meet the error even with very simple inference code.
This error is similar to #39525
```python
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
import pandas as pd
import torch
from transformers import AutoTokenizer,Qwen3ForCausalLM,AutoConfig,AutoModelForCausalLM
if __name__ == '__main__':
model_path="/share/models/Qwen3-4B-Thinking-2507"
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True, add_bos_token=False,
add_eos_token=False)
model = AutoModelForCausalLM.from_pretrained(model_path,
dtype=torch.bfloat16,
trust_remote_code=True,
#config=config,
attn_implementation="paged_attention",#"flash_attention_2",#"sdpa",#
device_map="cuda"
).eval()
prompt="How are you today?"
chat_prompt=tokenizer.apply_chat_template([{"role":"user","content":prompt}],tokenize=False,add_generation_prompt=True)
chat_prompt_ids=tokenizer(chat_prompt,return_tensors="pt")["input_ids"].to(model.device)
output=model.generate(input_ids=chat_prompt_ids,max_new_tokens=500,num_beams=1,do_sample=False,temperature=1.0,use_cache=True,return_dict_in_generate=True,output_logits=True)
output_text=tokenizer.decode(output['sequences'][0][chat_prompt_ids.size(1):])
print(output_text)
```
Error:
```
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/transformers/models/qwen3/modeling_qwen3.py", line 260, in forward
hidden_states, _ = self.self_attn(
^^^^^^^^^^^^^^^
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/transformers/utils/deprecation.py", line 172, in wrapped_func
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/transformers/models/qwen3/modeling_qwen3.py", line 216, in forward
attn_output, attn_weights = attention_interface(
^^^^^^^^^^^^^^^^^^^^
File "/root/miniconda3/envs/yyj/lib/python3.12/site-packages/transformers/integrations/flash_paged.py", line 85, in paged_attention_forward
cu_seq_lens_q.to(torch.int32),
^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'to'
```
### Expected behavior
fix the bug
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41501/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/41501/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/41500
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41500/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41500/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41500/events
|
https://github.com/huggingface/transformers/pull/41500
| 3,501,659,884
|
PR_kwDOCUB6oc6tBpiL
| 41,500
|
[QoL] modular conversion shows LoC saved
|
{
"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 |
[] | 2025-10-10T07:10:44
| 2025-10-10T09:55:24
| 2025-10-10T09:55:23
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41500",
"html_url": "https://github.com/huggingface/transformers/pull/41500",
"diff_url": "https://github.com/huggingface/transformers/pull/41500.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41500.patch",
"merged_at": "2025-10-10T09:55:23"
}
|
# What does this PR do?
Tiny thing. Simply print the LoC saved when doing modular conversion.
<img width="1528" height="81" alt="image" src="https://github.com/user-attachments/assets/a6bab118-5dc9-4550-991e-eb8acaa657f4" />
## Why?
Because we sometimes hesitate too much (me first). Modular isn't magic, it's close,and it can't do everything. I sometimes want to optimize too much a conversion that's already leagues better than if I was using `modeling` directly.
|
{
"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/41500/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/41500/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41499
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41499/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41499/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41499/events
|
https://github.com/huggingface/transformers/pull/41499
| 3,501,646,922
|
PR_kwDOCUB6oc6tBmvP
| 41,499
|
fix bnb model loading
|
{
"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 |
[] | 2025-10-10T07:05:44
| 2025-10-10T08:28:09
| 2025-10-10T08:27:29
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41499",
"html_url": "https://github.com/huggingface/transformers/pull/41499",
"diff_url": "https://github.com/huggingface/transformers/pull/41499.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41499.patch",
"merged_at": "2025-10-10T08:27:29"
}
|
```python
from transformers import pipeline
pipe = pipeline("text-generation", model="hugging-quants/Meta-Llama-3.1-8B-Instruct-BNB-NF4")
```
Before this PR:
```
Traceback (most recent call last):
File "/home/jiqing/transformers/src/transformers/pipelines/base.py", line 262, in load_model
model = model_class.from_pretrained(model, **fp32_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiqing/transformers/src/transformers/modeling_utils.py", line 272, in _wrapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/home/jiqing/transformers/src/transformers/modeling_utils.py", line 4669, in from_pretrained
hf_quantizer, config, dtype, device_map = get_hf_quantizer(
^^^^^^^^^^^^^^^^^
File "/home/jiqing/transformers/src/transformers/quantizers/auto.py", line 301, in get_hf_quantizer
config.quantization_config = AutoHfQuantizer.merge_quantization_configs(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiqing/transformers/src/transformers/quantizers/auto.py", line 210, in merge_quantization_configs
quantization_config = AutoQuantizationConfig.from_dict(quantization_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiqing/transformers/src/transformers/quantizers/auto.py", line 130, in from_dict
raise ValueError(
ValueError: Unknown quantization type, got bitsandbytes - supported types are: ['awq', 'bitsandbytes_4bit', 'bitsandbytes_8bit', 'gptq
', 'aqlm', 'quanto', 'quark', 'fp_quant', 'eetq', 'higgs', 'hqq', 'compressed-tensors', 'fbgemm_fp8', 'torchao', 'bitnet', 'vptq', 'sp
qr', 'fp8', 'auto-round', 'mxfp4']
```
After this PR, it works.
Hi @SunMarc . Please review this PR. Thanks!
|
{
"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/41499/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/41499/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41498
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41498/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41498/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41498/events
|
https://github.com/huggingface/transformers/issues/41498
| 3,501,348,442
|
I_kwDOCUB6oc7QslZa
| 41,498
|
mamba2 mixer different outputs between fast, slow, training, and inference paths.
|
{
"login": "tsengalb99",
"id": 33385672,
"node_id": "MDQ6VXNlcjMzMzg1Njcy",
"avatar_url": "https://avatars.githubusercontent.com/u/33385672?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/tsengalb99",
"html_url": "https://github.com/tsengalb99",
"followers_url": "https://api.github.com/users/tsengalb99/followers",
"following_url": "https://api.github.com/users/tsengalb99/following{/other_user}",
"gists_url": "https://api.github.com/users/tsengalb99/gists{/gist_id}",
"starred_url": "https://api.github.com/users/tsengalb99/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/tsengalb99/subscriptions",
"organizations_url": "https://api.github.com/users/tsengalb99/orgs",
"repos_url": "https://api.github.com/users/tsengalb99/repos",
"events_url": "https://api.github.com/users/tsengalb99/events{/privacy}",
"received_events_url": "https://api.github.com/users/tsengalb99/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
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-10T04:46:26
| 2025-10-11T08:31:05
| null |
NONE
| null | null | null | null |
### System Info
```
- huggingface_hub version: 0.31.4
- Platform: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
- Python version: 3.13.0
- Running in iPython ?: No
- Running in notebook ?: No
- Running in Google Colab ?: No
- Running in Google Colab Enterprise ?: No
- Token path ?: /home/alberttseng/.cache/huggingface/token
- Has saved token ?: True
- Who am I ?: at676
- Configured git credential helpers:
- FastAI: N/A
- Tensorflow: N/A
- Torch: 2.9.0.dev20250803+cu129
- Jinja2: 3.1.6
- Graphviz: N/A
- keras: N/A
- Pydot: N/A
- Pillow: 11.2.1
- hf_transfer: N/A
- gradio: N/A
- tensorboard: N/A
- numpy: 2.2.6
- pydantic: 2.10.3
- aiohttp: 3.11.18
- hf_xet: N/A
- ENDPOINT: https://huggingface.co
- HF_HUB_CACHE: /home/alberttseng/.cache/huggingface/hub
- HF_ASSETS_CACHE: /home/alberttseng/.cache/huggingface/assets
- HF_TOKEN_PATH: /home/alberttseng/.cache/huggingface/token
- HF_STORED_TOKENS_PATH: /home/alberttseng/.cache/huggingface/stored_tokens
- HF_HUB_OFFLINE: False
- HF_HUB_DISABLE_TELEMETRY: False
- HF_HUB_DISABLE_PROGRESS_BARS: None
- HF_HUB_DISABLE_SYMLINKS_WARNING: False
- HF_HUB_DISABLE_EXPERIMENTAL_WARNING: False
- HF_HUB_DISABLE_IMPLICIT_TOKEN: False
- HF_HUB_ENABLE_HF_TRANSFER: False
- HF_HUB_ETAG_TIMEOUT: 10
- HF_HUB_DOWNLOAD_TIMEOUT: 10
```
### Who can help?
@ArthurZucker @Cyrilvallez
### 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
In https://github.com/huggingface/transformers/blob/0720e206c6ba28887e4d60ef60a6a089f6c1cc76/src/transformers/models/mamba2/modeling_mamba2.py#L656, when given the same input to `hidden_states` and setting the other arguments to `None`, I get different results for the fast path (`self.cuda_kernels_forward`) with `self.training = True`, the fast path with `self.training = False`, and the slow path.
### Expected behavior
These should all be equivalent.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41498/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/41498/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/41497
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41497/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41497/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41497/events
|
https://github.com/huggingface/transformers/pull/41497
| 3,501,186,963
|
PR_kwDOCUB6oc6tAHdQ
| 41,497
|
Add final contribution report and training file
|
{
"login": "ali1baba1",
"id": 191934682,
"node_id": "U_kgDOC3Cw2g",
"avatar_url": "https://avatars.githubusercontent.com/u/191934682?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ali1baba1",
"html_url": "https://github.com/ali1baba1",
"followers_url": "https://api.github.com/users/ali1baba1/followers",
"following_url": "https://api.github.com/users/ali1baba1/following{/other_user}",
"gists_url": "https://api.github.com/users/ali1baba1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ali1baba1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ali1baba1/subscriptions",
"organizations_url": "https://api.github.com/users/ali1baba1/orgs",
"repos_url": "https://api.github.com/users/ali1baba1/repos",
"events_url": "https://api.github.com/users/ali1baba1/events{/privacy}",
"received_events_url": "https://api.github.com/users/ali1baba1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[
{
"id": 9258341780,
"node_id": "LA_kwDOCUB6oc8AAAACJ9cVlA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Code%20agent%20slop",
"name": "Code agent slop",
"color": "C59579",
"default": false,
"description": ""
}
] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T02:59:54
| 2025-10-11T01:34:28
| 2025-10-10T12:00:37
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41497",
"html_url": "https://github.com/huggingface/transformers/pull/41497",
"diff_url": "https://github.com/huggingface/transformers/pull/41497.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41497.patch",
"merged_at": null
}
|
## What does this PR do?
This pull request adds my final contribution for the Hugging Face Transformers community project.
It includes my experiment summary and the dataset used for fine-tuning a Masked Language Model (MLM).
## Who can review?
Any member of the Transformers maintainers team.
## Motivation
This contribution documents the results of my experiment using the no-trainer MLM script.
It aims to share training data, model results, and experiment setup for community reference.
## Changes
- Added `final_contribution_report.md` containing:
- Model setup, training commands, and hyperparameters
- Results and evaluation metrics
- Observations and conclusions
- Added `train.txt` used for fine-tuning
## Additional context
This PR completes my final submission for the community contribution process.
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41497/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/41497/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41496
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41496/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41496/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41496/events
|
https://github.com/huggingface/transformers/pull/41496
| 3,501,173,681
|
PR_kwDOCUB6oc6tAEvV
| 41,496
|
Allow optuna's catch kwargs passthrough
|
{
"login": "nicha-api",
"id": 115199576,
"node_id": "U_kgDOBt3OWA",
"avatar_url": "https://avatars.githubusercontent.com/u/115199576?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/nicha-api",
"html_url": "https://github.com/nicha-api",
"followers_url": "https://api.github.com/users/nicha-api/followers",
"following_url": "https://api.github.com/users/nicha-api/following{/other_user}",
"gists_url": "https://api.github.com/users/nicha-api/gists{/gist_id}",
"starred_url": "https://api.github.com/users/nicha-api/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/nicha-api/subscriptions",
"organizations_url": "https://api.github.com/users/nicha-api/orgs",
"repos_url": "https://api.github.com/users/nicha-api/repos",
"events_url": "https://api.github.com/users/nicha-api/events{/privacy}",
"received_events_url": "https://api.github.com/users/nicha-api/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-10T02:48:40
| 2025-10-10T13:58:41
| 2025-10-10T13:58:07
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41496",
"html_url": "https://github.com/huggingface/transformers/pull/41496",
"diff_url": "https://github.com/huggingface/transformers/pull/41496.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41496.patch",
"merged_at": "2025-10-10T13:58:07"
}
|
# What does this PR do?
Allow catch kwarg passthrough in Trainer.hyperparameter_search(backend="optuna") in order to expose Optuna’s Study.optimize(..., catch=...) argument by popping out catch kwargs before calling Optuna's study.optimize().
This PR addresss feature request #41463
|
{
"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/41496/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/41496/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41495
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41495/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41495/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41495/events
|
https://github.com/huggingface/transformers/pull/41495
| 3,500,913,797
|
PR_kwDOCUB6oc6s_P5F
| 41,495
|
Rm yoso kernel
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T23:41:30
| 2025-10-10T08:50:15
| 2025-10-10T08:50:13
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41495",
"html_url": "https://github.com/huggingface/transformers/pull/41495",
"diff_url": "https://github.com/huggingface/transformers/pull/41495.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41495.patch",
"merged_at": "2025-10-10T08:50:13"
}
|
# What does this PR do?
Removing the yoso kernels, and using https://huggingface.co/kernels-community/yoso instead
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41495/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/41495/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41494
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41494/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41494/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41494/events
|
https://github.com/huggingface/transformers/issues/41494
| 3,500,893,664
|
I_kwDOCUB6oc7Qq2Xg
| 41,494
|
Incorrect tokenizer created for gemma gguf files
|
{
"login": "amychen85",
"id": 4982156,
"node_id": "MDQ6VXNlcjQ5ODIxNTY=",
"avatar_url": "https://avatars.githubusercontent.com/u/4982156?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/amychen85",
"html_url": "https://github.com/amychen85",
"followers_url": "https://api.github.com/users/amychen85/followers",
"following_url": "https://api.github.com/users/amychen85/following{/other_user}",
"gists_url": "https://api.github.com/users/amychen85/gists{/gist_id}",
"starred_url": "https://api.github.com/users/amychen85/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/amychen85/subscriptions",
"organizations_url": "https://api.github.com/users/amychen85/orgs",
"repos_url": "https://api.github.com/users/amychen85/repos",
"events_url": "https://api.github.com/users/amychen85/events{/privacy}",
"received_events_url": "https://api.github.com/users/amychen85/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
}
] |
open
| false
|
{
"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
}
|
[
{
"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
}
] | null |
[] | 2025-10-09T23:27:25
| 2025-10-27T04:51:51
| null |
NONE
| null | null | null | null |
### System Info
- `transformers` version: 4.57.0
- Platform: Linux-5.15.0-144-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- Huggingface_hub version: 0.34.4
- Safetensors version: 0.5.3
- Accelerate version: 0.34.2
- Accelerate config: not found
- DeepSpeed version: not installed
- PyTorch version (accelerator?): 2.3.1+cu121 (NA)
- Tensorflow version (GPU?): 2.17.0 (False)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: NA
### Who can help?
@yijun-lee
@Isotr0py
### 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
```
from transformers import AutoTokenizer
t1 = AutoTokenizer.from_pretrained("unsloth/gemma-3-4b-it-GGUF", gguf_file="gemma-3-4b-it-Q8_0.gguf")
x1 = t1.tokenize("<bos>What is eunoia?")
print(f"{x1=}")
t2 = AutoTokenizer.from_pretrained("google/gemma-3-4b-it")
x2 = t2.tokenize("<bos>What is eunoia?")
print(f"{x2=}")
```
### Expected behavior
The print out of the x1 and x2 should be the same. However,
```
x1=['<bos>', 'Wh', 'at', '▁is', '▁eu', 'no', 'ia', '?']
x2=['<bos>', 'What', '▁is', '▁e', 'uno', 'ia', '?']
```
Looking more into it, the tokenizer created for HF model (t2) is BPE while the tokenizer created for the GGUF model (t1) is Unigram.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41494/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/41494/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/41493
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41493/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41493/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41493/events
|
https://github.com/huggingface/transformers/pull/41493
| 3,500,778,419
|
PR_kwDOCUB6oc6s-zZZ
| 41,493
|
[kernels] Remove RWKV kernel finally !
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T22:19:42
| 2025-10-10T08:32:07
| 2025-10-10T08:32:06
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41493",
"html_url": "https://github.com/huggingface/transformers/pull/41493",
"diff_url": "https://github.com/huggingface/transformers/pull/41493.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41493.patch",
"merged_at": "2025-10-10T08:32:05"
}
|
# What does this PR do?
Cleans the rwkv kernel, after adding the kernel to `kernels-community` : https://huggingface.co/kernels-community/rwkv
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41493/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/41493/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41492
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41492/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41492/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41492/events
|
https://github.com/huggingface/transformers/issues/41492
| 3,500,775,496
|
I_kwDOCUB6oc7QqZhI
| 41,492
|
For finetuning MBart-based model, setting decoder_start_token_id in model.config is NOT ENOUGH.
|
{
"login": "Bmingg",
"id": 82941091,
"node_id": "MDQ6VXNlcjgyOTQxMDkx",
"avatar_url": "https://avatars.githubusercontent.com/u/82941091?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Bmingg",
"html_url": "https://github.com/Bmingg",
"followers_url": "https://api.github.com/users/Bmingg/followers",
"following_url": "https://api.github.com/users/Bmingg/following{/other_user}",
"gists_url": "https://api.github.com/users/Bmingg/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Bmingg/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Bmingg/subscriptions",
"organizations_url": "https://api.github.com/users/Bmingg/orgs",
"repos_url": "https://api.github.com/users/Bmingg/repos",
"events_url": "https://api.github.com/users/Bmingg/events{/privacy}",
"received_events_url": "https://api.github.com/users/Bmingg/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
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-09T22:18:01
| 2025-10-11T13:21:56
| null |
NONE
| null | null | null | null |
### System Info
Context: finetuning a MBart model with run_translation.py
Easy fix is to set it in model.generation_config as well. Both worked outside of run_translation.py, but not setting this in run_translation.py causes validation/evaluation to fail miserably.
### Who can help?
_No response_
### Information
- [x] The official example scripts
- [x] 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 run_translation.py to finetune with a validation file.
### Expected behavior
BLEU scores should be terrible due to the bug. This makes it hard to monitor finetuning.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41492/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/41492/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/41491
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41491/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41491/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41491/events
|
https://github.com/huggingface/transformers/pull/41491
| 3,500,719,119
|
PR_kwDOCUB6oc6s-nLC
| 41,491
|
Add skip_unnecessary_grad_clip to TrainingArguments for optimized gradient clipping
|
{
"login": "vaibhavgarg230",
"id": 170336277,
"node_id": "U_kgDOCicgFQ",
"avatar_url": "https://avatars.githubusercontent.com/u/170336277?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vaibhavgarg230",
"html_url": "https://github.com/vaibhavgarg230",
"followers_url": "https://api.github.com/users/vaibhavgarg230/followers",
"following_url": "https://api.github.com/users/vaibhavgarg230/following{/other_user}",
"gists_url": "https://api.github.com/users/vaibhavgarg230/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vaibhavgarg230/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vaibhavgarg230/subscriptions",
"organizations_url": "https://api.github.com/users/vaibhavgarg230/orgs",
"repos_url": "https://api.github.com/users/vaibhavgarg230/repos",
"events_url": "https://api.github.com/users/vaibhavgarg230/events{/privacy}",
"received_events_url": "https://api.github.com/users/vaibhavgarg230/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-09T21:47:27
| 2025-10-10T10:53:07
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41491",
"html_url": "https://github.com/huggingface/transformers/pull/41491",
"diff_url": "https://github.com/huggingface/transformers/pull/41491.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41491.patch",
"merged_at": null
}
|
# What does this PR do?
This PR adds an opt-in `skip_unnecessary_grad_clip` argument to `TrainingArguments`, optimizing Trainer’s gradient clipping for better efficiency. When enabled, the Trainer computes and logs the gradient norm every step, but skips calling `clip_grad_norm_` if the norm is already below `max_grad_norm`. This prevents unnecessary computation for models/trainers with consistently low gradient norms, while always maintaining logging.
**Maintainer requests addressed:**
- Grad norm is logged every step, even when clipping is skipped.
- Default behavior is unchanged (the flag is `False` by default).
**Motivation:**
- Addresses [issue #41431 ]
- Improves performance for stable gradient or CPU-bound training.
- Keeps code backward-compatible and easy to opt-in.
**Dependencies:**
- No new dependencies.
**Tests added:**
- New test: `tests/trainer/test_gradient_clipping.py`
- Verifies clipping is skipped when grad norm is under threshold.
- Verifies grad norm is still correctly logged.
- Verifies the default and opt-in cases.
**Documentation:**
- Argument and code block are documented as per contributing guidelines.
---
## Before submitting
- [x] Discussed/approved via issue or maintainer
- [x] Documentation updated
- [x] All tests and repo checks run locally
---
## Who can review?
- Trainer logic: @zach-huggingface @SunMarc
---
Thanks for reviewing! Feedback and suggestions are very welcome.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41491/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/41491/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41490
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41490/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41490/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41490/events
|
https://github.com/huggingface/transformers/pull/41490
| 3,499,987,210
|
PR_kwDOCUB6oc6s8LJg
| 41,490
|
Fix _init_weights to safely skip int8 tensors in Qwen2_5_VL model
|
{
"login": "KaparthyReddy",
"id": 166050493,
"node_id": "U_kgDOCeW6vQ",
"avatar_url": "https://avatars.githubusercontent.com/u/166050493?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KaparthyReddy",
"html_url": "https://github.com/KaparthyReddy",
"followers_url": "https://api.github.com/users/KaparthyReddy/followers",
"following_url": "https://api.github.com/users/KaparthyReddy/following{/other_user}",
"gists_url": "https://api.github.com/users/KaparthyReddy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KaparthyReddy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KaparthyReddy/subscriptions",
"organizations_url": "https://api.github.com/users/KaparthyReddy/orgs",
"repos_url": "https://api.github.com/users/KaparthyReddy/repos",
"events_url": "https://api.github.com/users/KaparthyReddy/events{/privacy}",
"received_events_url": "https://api.github.com/users/KaparthyReddy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-09T17:13:44
| 2025-10-11T13:09:59
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41490",
"html_url": "https://github.com/huggingface/transformers/pull/41490",
"diff_url": "https://github.com/huggingface/transformers/pull/41490.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41490.patch",
"merged_at": null
}
|
### Summary
This PR fixes the `_init_weights()` method in `Qwen2_5_VLForConditionalGeneration` to safely skip int8 tensors during initialization. Previously, applying `normal_()` on int8 weights caused a RuntimeError when loading quantized models.
### Changes
- Added dtype check in `_init_weights()` to only initialize floating-point tensors (`float16`, `float32`, `bfloat16`).
- Ensures int8 weights from quantized models are skipped safely.
- Verified fix by successfully loading a quantized Qwen2.5-VL model on CPU.
### Motivation
Quantized models (W8A8, int8 weights) could not be loaded directly due to the previous `_init_weights()` implementation. This fix allows smooth loading without RuntimeError, making contributions compatible with LLMCompressor quantized models.
### Verification
- Model loaded successfully on CPU after applying the fix.
- `_init_weights()` safely ignores int8 tensors.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41490/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/41490/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41489
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41489/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41489/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41489/events
|
https://github.com/huggingface/transformers/pull/41489
| 3,499,877,251
|
PR_kwDOCUB6oc6s70rh
| 41,489
|
More trainer cleaning
|
{
"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
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T16:39:15
| 2025-10-10T09:55:45
| 2025-10-10T09:55:43
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41489",
"html_url": "https://github.com/huggingface/transformers/pull/41489",
"diff_url": "https://github.com/huggingface/transformers/pull/41489.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41489.patch",
"merged_at": "2025-10-10T09:55:43"
}
|
# What does this PR do?
This PR removes 2 unused args : `_n_gpu` and `mp_parameters`
|
{
"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/41489/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/41489/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41488
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41488/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41488/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41488/events
|
https://github.com/huggingface/transformers/pull/41488
| 3,499,846,115
|
PR_kwDOCUB6oc6s7uGN
| 41,488
|
Create 1
|
{
"login": "yrabasalaev-netizen",
"id": 237085136,
"node_id": "U_kgDODiGh0A",
"avatar_url": "https://avatars.githubusercontent.com/u/237085136?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yrabasalaev-netizen",
"html_url": "https://github.com/yrabasalaev-netizen",
"followers_url": "https://api.github.com/users/yrabasalaev-netizen/followers",
"following_url": "https://api.github.com/users/yrabasalaev-netizen/following{/other_user}",
"gists_url": "https://api.github.com/users/yrabasalaev-netizen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yrabasalaev-netizen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yrabasalaev-netizen/subscriptions",
"organizations_url": "https://api.github.com/users/yrabasalaev-netizen/orgs",
"repos_url": "https://api.github.com/users/yrabasalaev-netizen/repos",
"events_url": "https://api.github.com/users/yrabasalaev-netizen/events{/privacy}",
"received_events_url": "https://api.github.com/users/yrabasalaev-netizen/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-09T16:28:59
| 2025-10-09T16:28:59
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41488",
"html_url": "https://github.com/huggingface/transformers/pull/41488",
"diff_url": "https://github.com/huggingface/transformers/pull/41488.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41488.patch",
"merged_at": null
}
|
1
# 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
- [ ] 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41488/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/41488/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41487
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41487/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41487/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41487/events
|
https://github.com/huggingface/transformers/pull/41487
| 3,499,843,152
|
PR_kwDOCUB6oc6s7tej
| 41,487
|
Migrate transformers cli to Typer
|
{
"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
}
|
[
{
"id": 9105758243,
"node_id": "LA_kwDOCUB6oc8AAAACHr7YIw",
"url": "https://api.github.com/repos/huggingface/transformers/labels/for_v5?",
"name": "for_v5?",
"color": "35BC94",
"default": false,
"description": ""
}
] |
closed
| false
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
] | null |
[] | 2025-10-09T16:28:01
| 2025-10-16T11:29:43
| 2025-10-16T11:29:42
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41487",
"html_url": "https://github.com/huggingface/transformers/pull/41487",
"diff_url": "https://github.com/huggingface/transformers/pull/41487.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41487.patch",
"merged_at": "2025-10-16T11:29:42"
}
|
This PR migrates the `transformers` CLI to Typer.
Typer is a package build on top of click by the creator of FastAPI. It is now a dependency of `huggingface_hub`, meaning it's also a dependency of `transformers`. Typer simplifies arg definition and ensures consistency using type annotations, which should help with maintenance. The benefit for users is the built-in autocompletion feature that let someone do `transformers chat [TAB][TAB]` to check what are the options. The `--help` section is also improved.
By migrating to Typer, a longer term goal is to delegate to `huggingface_hub` some aspects of the installation and auto-update of the CLI. This will come in a second time and doesn't have to be correlated with the v5 release.
### CLI `--help`
```sh
transformers --help
Usage: transformers [OPTIONS] COMMAND [ARGS]...
Transformers CLI
Options:
--install-completion Install completion for the current shell.
--show-completion Show completion for the current shell, to copy it or
customize the installation.
--help Show this message and exit.
Commands:
add-fast-image-processor Add a fast image processor to a model.
add-new-model-like Add a new model to the library, based on an...
chat Chat with a model from the command line.
download Download a model and its tokenizer from the Hub.
env Print information about the environment.
run Run a pipeline on a given input file.
serve Run a FastAPI server to serve models...
version Print CLI version.
```
### Side notes
Noted down some stuff while working on it. Can be addressed in later PRs.
1. Any command, even a simple `transformers env` is currently very long due both in previous and new CLI. This is due to `torch` import, no matter if it's used or not. I do think this is not good UX, especially if we want to have something like `transformers chat` as entrypoint for any openai-compatible server.
This is not really related to CLI only, but to lazy imports in general. I broke it down to `is_torch_available` actually importing the package, not just checking it's existence.
2. The current `transformers serve` + `transformers chat` twin commands are really nice. One to start a server, the other one launch a chat interface. However, I feel that the current UX for `chat` is too bloated since it covers both the case where a server is already started AND started a new server from a model id (or path). I do think that `transformers chat` should only be to consume an existing API. It would make the whole implementation much cleaner and the interface less bloated for the end user (currently having 4-5 arguments only to provide model name, path, address, port, and host, instead of a single "url" argument).
Since this would be a breaking change, I think addressing it for v5 is the perfect timing.
=> **EDIT:** this is now done in this PR. `transformers chat` does not serve a model now (making it much simpler)
```
transformers chat https://router.huggingface.co/v1 HuggingFaceTB/SmolLM3-3B
```
3. The `transformers serve` feature is currently only available as a CLI. I do believe it would be best to move it to its own module so that someone could call it programmatically (e.g. `from transformers import serve` in a notebook).
### TODO
- [x] (minor) do not import `typer_factory` from private internal (requires an update in `huggingface_hub` first)
- [x] delete `./commands` folder and remove `transformers-legacy` CLI (that I currently use for testing)
- [x] adapt remaining CLI tests
- [x] `transformers chat` UI-only (not serving)
- [ ] do not use classes for `chat` and `serve`? + expose them as modules (e.g. `from transformers import chat, serve`)
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41487/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/41487/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41486
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41486/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41486/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41486/events
|
https://github.com/huggingface/transformers/pull/41486
| 3,499,829,317
|
PR_kwDOCUB6oc6s7qjz
| 41,486
|
[`CI`] Fix copies on main
|
{
"login": "vasqu",
"id": 73884904,
"node_id": "MDQ6VXNlcjczODg0OTA0",
"avatar_url": "https://avatars.githubusercontent.com/u/73884904?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vasqu",
"html_url": "https://github.com/vasqu",
"followers_url": "https://api.github.com/users/vasqu/followers",
"following_url": "https://api.github.com/users/vasqu/following{/other_user}",
"gists_url": "https://api.github.com/users/vasqu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vasqu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vasqu/subscriptions",
"organizations_url": "https://api.github.com/users/vasqu/orgs",
"repos_url": "https://api.github.com/users/vasqu/repos",
"events_url": "https://api.github.com/users/vasqu/events{/privacy}",
"received_events_url": "https://api.github.com/users/vasqu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T16:24:23
| 2025-10-09T16:38:19
| 2025-10-09T16:38:14
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41486",
"html_url": "https://github.com/huggingface/transformers/pull/41486",
"diff_url": "https://github.com/huggingface/transformers/pull/41486.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41486.patch",
"merged_at": "2025-10-09T16:38:14"
}
|
As per title
|
{
"login": "vasqu",
"id": 73884904,
"node_id": "MDQ6VXNlcjczODg0OTA0",
"avatar_url": "https://avatars.githubusercontent.com/u/73884904?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vasqu",
"html_url": "https://github.com/vasqu",
"followers_url": "https://api.github.com/users/vasqu/followers",
"following_url": "https://api.github.com/users/vasqu/following{/other_user}",
"gists_url": "https://api.github.com/users/vasqu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vasqu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vasqu/subscriptions",
"organizations_url": "https://api.github.com/users/vasqu/orgs",
"repos_url": "https://api.github.com/users/vasqu/repos",
"events_url": "https://api.github.com/users/vasqu/events{/privacy}",
"received_events_url": "https://api.github.com/users/vasqu/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41486/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/41486/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41485
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41485/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41485/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41485/events
|
https://github.com/huggingface/transformers/pull/41485
| 3,499,818,961
|
PR_kwDOCUB6oc6s7ocD
| 41,485
|
Fix smolvlm2 dtype mismatch final
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-09T16:21:42
| 2025-10-10T18:04:35
| null |
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41485",
"html_url": "https://github.com/huggingface/transformers/pull/41485",
"diff_url": "https://github.com/huggingface/transformers/pull/41485.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41485.patch",
"merged_at": null
}
|
# Fix SmolVLM2 quantization dtype mismatch
## What does this PR do?
Fixes #41453 - SmolVLM2 cannot be used with quantization due to dtype mismatch error.
**Problem**: When loading SmolVLM2 with BitsAndBytesConfig and bfloat16, the `inputs_merger` function fails with:
```
RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.
```
**Root Cause**:
- Quantization forces `inputs_embeds` to `torch.bfloat16` (from BitsAndBytesConfig)
- Vision encoder outputs `image_hidden_states` in `torch.float32`
- Direct assignment between incompatible dtypes causes the crash
**Solution**: Added dtype conversion to ensure `image_hidden_states` matches `inputs_embeds` dtype before assignment:
```python
# Before (BROKEN):
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
# After (FIXED):
# Ensure dtype compatibility for quantization
image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
**Changes**:
- Modified `src/transformers/models/smolvlm/modeling_smolvlm.py` - Added dtype conversion in `inputs_merger` function
- Updated `src/transformers/models/smolvlm/modular_smolvlm.py` - Aligned modular file with same fix
- Added test in `tests/models/smolvlm/test_modeling_smolvlm.py` - `test_quantization_dtype_compatibility()` with `@slow` decorator
**Testing**: The fix has been thoroughly tested and verified to resolve the quantization dtype mismatch issue without breaking existing functionality.
Fixes #41453
## 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?
@yonigozlan @molbap - This affects vision models and quantization functionality
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41485/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/41485/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41484
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41484/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41484/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41484/events
|
https://github.com/huggingface/transformers/pull/41484
| 3,499,676,992
|
PR_kwDOCUB6oc6s7KHP
| 41,484
|
add Trainer import to .md in appropriate cell block for training.ipynb transformers_doc
|
{
"login": "benkeene",
"id": 39920749,
"node_id": "MDQ6VXNlcjM5OTIwNzQ5",
"avatar_url": "https://avatars.githubusercontent.com/u/39920749?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/benkeene",
"html_url": "https://github.com/benkeene",
"followers_url": "https://api.github.com/users/benkeene/followers",
"following_url": "https://api.github.com/users/benkeene/following{/other_user}",
"gists_url": "https://api.github.com/users/benkeene/gists{/gist_id}",
"starred_url": "https://api.github.com/users/benkeene/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/benkeene/subscriptions",
"organizations_url": "https://api.github.com/users/benkeene/orgs",
"repos_url": "https://api.github.com/users/benkeene/repos",
"events_url": "https://api.github.com/users/benkeene/events{/privacy}",
"received_events_url": "https://api.github.com/users/benkeene/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T15:41:42
| 2025-10-10T12:04:50
| 2025-10-10T12:04:07
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41484",
"html_url": "https://github.com/huggingface/transformers/pull/41484",
"diff_url": "https://github.com/huggingface/transformers/pull/41484.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41484.patch",
"merged_at": "2025-10-10T12:04:07"
}
|
# What does this PR do?
transformers_doc/en/training.ipynb does not import Trainer in the code block
```python
trainer = Trainer(
model=model,
args=training_args,
train_dataset=dataset["train"],
eval_dataset=dataset["test"],
compute_metrics=compute_metrics,
)
trainer.train()
```
Which is imported from huggingface/transformers repo docs/source/en/training.md. Added the import in the source .md file.
Current behavior:
---------------------------------------------------------------------------
NameError Traceback (most recent call last)
Cell In[7], line 1
----> 1 trainer = Trainer(
2 model=model,
3 args=training_args,
4 train_dataset=dataset["train"],
5 eval_dataset=dataset["test"],
6 compute_metrics=compute_metrics,
7 )
8 trainer.train()
NameError: name 'Trainer' is not defined
Expected behavior:
The training should begin. Adding the import directly to transformers_doc/en/training.ipynb fixes the issue, but as mentioned in the pull request prompt: changes should be made in the huggingface/transformers repository.
<!--
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)
Missing import in transformers_doc/en/training.ipynb: `from transformers import Trainer`.
## 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41484/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/41484/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41483
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41483/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41483/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41483/events
|
https://github.com/huggingface/transformers/pull/41483
| 3,499,644,613
|
PR_kwDOCUB6oc6s7DFS
| 41,483
|
Fixed tiny incorrect imports in `glm4v`
|
{
"login": "Sai-Suraj-27",
"id": 87087741,
"node_id": "MDQ6VXNlcjg3MDg3NzQx",
"avatar_url": "https://avatars.githubusercontent.com/u/87087741?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Sai-Suraj-27",
"html_url": "https://github.com/Sai-Suraj-27",
"followers_url": "https://api.github.com/users/Sai-Suraj-27/followers",
"following_url": "https://api.github.com/users/Sai-Suraj-27/following{/other_user}",
"gists_url": "https://api.github.com/users/Sai-Suraj-27/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Sai-Suraj-27/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Sai-Suraj-27/subscriptions",
"organizations_url": "https://api.github.com/users/Sai-Suraj-27/orgs",
"repos_url": "https://api.github.com/users/Sai-Suraj-27/repos",
"events_url": "https://api.github.com/users/Sai-Suraj-27/events{/privacy}",
"received_events_url": "https://api.github.com/users/Sai-Suraj-27/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T15:32:45
| 2025-10-10T08:59:50
| 2025-10-10T08:57:01
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41483",
"html_url": "https://github.com/huggingface/transformers/pull/41483",
"diff_url": "https://github.com/huggingface/transformers/pull/41483.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41483.patch",
"merged_at": "2025-10-10T08:57:01"
}
|
# What does this PR do?
Fixes tiny incorrect imports in `glm4v`. These are the correct classes
https://github.com/huggingface/transformers/blob/34b861abd11074fd32362b9a25c1cc582fa0b941/src/transformers/models/qwen2_vl/processing_qwen2_vl.py#L39
https://github.com/huggingface/transformers/blob/34b861abd11074fd32362b9a25c1cc582fa0b941/src/transformers/models/qwen2_vl/processing_qwen2_vl.py#L48
## 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?
cc: @zucchini-nlp @gante @Cyrilvallez (Similar to https://github.com/huggingface/transformers/pull/41354)
|
{
"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/41483/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/41483/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41482
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41482/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41482/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41482/events
|
https://github.com/huggingface/transformers/pull/41482
| 3,499,602,994
|
PR_kwDOCUB6oc6s66LP
| 41,482
|
Fix smolvlm2 dtype mismatch
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T15:20:29
| 2025-10-09T16:19:51
| 2025-10-09T16:19:51
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41482",
"html_url": "https://github.com/huggingface/transformers/pull/41482",
"diff_url": "https://github.com/huggingface/transformers/pull/41482.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41482.patch",
"merged_at": null
}
|
# Fix SmolVLM2 quantization dtype mismatch
## What does this PR do?
Fixes #41453 - SmolVLM2 cannot be used with quantization due to dtype mismatch error.
**Problem**: When loading SmolVLM2 with BitsAndBytesConfig and bfloat16, the `inputs_merger` function fails with:
```
RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.
```
**Root Cause**:
- Quantization forces `inputs_embeds` to `torch.bfloat16` (from BitsAndBytesConfig)
- Vision encoder outputs `image_hidden_states` in `torch.float32`
- Direct assignment between incompatible dtypes causes the crash
**Solution**: Added dtype conversion to ensure `image_hidden_states` matches `inputs_embeds` dtype before assignment:
```python
# Before (BROKEN):
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
# After (FIXED):
# Ensure dtype compatibility for quantization
image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
**Changes**:
- Modified `src/transformers/models/smolvlm/modeling_smolvlm.py` - Added dtype conversion in `inputs_merger` function
- Updated `src/transformers/models/smolvlm/modular_smolvlm.py` - Aligned modular file with same fix
- Added test in `tests/models/smolvlm/test_modeling_smolvlm.py` - `test_quantization_dtype_compatibility()` with `@slow` decorator
**Testing**: The fix has been thoroughly tested and verified to resolve the quantization dtype mismatch issue without breaking existing functionality.
Fixes #41453
## 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?
@yonigozlan @molbap - This affects vision models and quantization functionality
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41482/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/41482/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41481
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41481/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41481/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41481/events
|
https://github.com/huggingface/transformers/pull/41481
| 3,499,517,628
|
PR_kwDOCUB6oc6s6n5c
| 41,481
|
Fix/smolvlm2 quantization dtype mismatch clean
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T14:52:50
| 2025-10-09T15:02:16
| 2025-10-09T15:02:16
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41481",
"html_url": "https://github.com/huggingface/transformers/pull/41481",
"diff_url": "https://github.com/huggingface/transformers/pull/41481.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41481.patch",
"merged_at": null
}
|
# Fix SmolVLM2 quantization dtype mismatch
## What does this PR do?
Fixes #41453 - SmolVLM2 cannot be used with quantization due to dtype mismatch error.
**Problem**: When loading SmolVLM2 with BitsAndBytesConfig and bfloat16, the `inputs_merger` function fails with:
```
RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.
```
**Root Cause**:
- Quantization forces `inputs_embeds` to `torch.bfloat16` (from BitsAndBytesConfig)
- Vision encoder outputs `image_hidden_states` in `torch.float32`
- Direct assignment between incompatible dtypes causes the crash
**Solution**: Added dtype conversion to ensure `image_hidden_states` matches `inputs_embeds` dtype before assignment:
```python
# Before (BROKEN):
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
# After (FIXED):
# Ensure dtype compatibility for quantization
image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
**Changes**:
- Modified `src/transformers/models/smolvlm/modeling_smolvlm.py` - Added dtype conversion in `inputs_merger` function
- Updated `src/transformers/models/smolvlm/modular_smolvlm.py` - Aligned modular file with same fix
- Added test in `tests/models/smolvlm/test_modeling_smolvlm.py` - `test_quantization_dtype_compatibility()` with `@slow` decorator
**Testing**: The fix has been thoroughly tested and verified to resolve the quantization dtype mismatch issue without breaking existing functionality.
Fixes #41453
## 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?
@yonigozlan @molbap @zucchini-nlp - This affects vision models and quantization functionality
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41481/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/41481/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41480
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41480/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41480/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41480/events
|
https://github.com/huggingface/transformers/pull/41480
| 3,499,484,892
|
PR_kwDOCUB6oc6s6g-R
| 41,480
|
Fix/smolvlm2 quantization dtype mismatch
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T14:42:20
| 2025-10-09T15:23:46
| 2025-10-09T15:00:09
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41480",
"html_url": "https://github.com/huggingface/transformers/pull/41480",
"diff_url": "https://github.com/huggingface/transformers/pull/41480.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41480.patch",
"merged_at": null
}
|
# Fix SmolVLM2 quantization dtype mismatch
## What does this PR do?
Fixes #41453 - SmolVLM2 cannot be used with quantization due to dtype mismatch error.
**Problem**: When loading SmolVLM2 with BitsAndBytesConfig and bfloat16, the `inputs_merger` function fails with:
```
RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.
```
**Root Cause**:
- Quantization forces `inputs_embeds` to `torch.bfloat16` (from BitsAndBytesConfig)
- Vision encoder outputs `image_hidden_states` in `torch.float32`
- Direct assignment between incompatible dtypes causes the crash
**Solution**: Added dtype conversion to ensure `image_hidden_states` matches `inputs_embeds` dtype before assignment:
```python
# Before (BROKEN):
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
# After (FIXED):
# Ensure dtype compatibility for quantization
image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
**Changes**:
- Modified `src/transformers/models/smolvlm/modeling_smolvlm.py` - Added dtype conversion in `inputs_merger` function
- Updated `src/transformers/models/smolvlm/modular_smolvlm.py` - Aligned modular file with same fix
- Added test in `tests/models/smolvlm/test_modeling_smolvlm.py` - `test_quantization_dtype_compatibility()` with `@slow` decorator
**Testing**: The fix has been thoroughly tested and verified to resolve the quantization dtype mismatch issue without breaking existing functionality.
Fixes #41453
## 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?
@yonigozlan @molbap @zucchini-nlp - This affects vision models and quantization functionality
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41480/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/41480/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41479
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41479/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41479/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41479/events
|
https://github.com/huggingface/transformers/pull/41479
| 3,499,421,064
|
PR_kwDOCUB6oc6s6TI5
| 41,479
|
Remove SigOpt
|
{
"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
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T14:26:00
| 2025-10-09T16:05:57
| 2025-10-09T16:05:55
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41479",
"html_url": "https://github.com/huggingface/transformers/pull/41479",
"diff_url": "https://github.com/huggingface/transformers/pull/41479.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41479.patch",
"merged_at": "2025-10-09T16:05:55"
}
|
# What does this PR do?
This PR removes SigOpt integration as the library is not maintained anymore by intel. Also, users who used this library had to pass a API key, so since the service is down, the API won't work at all. I disabled the tests a few months ago since they started failing and now it is better to just remove the integration.
|
{
"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/41479/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/41479/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41478
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41478/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41478/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41478/events
|
https://github.com/huggingface/transformers/pull/41478
| 3,499,415,486
|
PR_kwDOCUB6oc6s6R6z
| 41,478
|
Fix/smolvlm2 quantization dtype mismatch
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T14:24:47
| 2025-10-09T15:23:46
| 2025-10-09T14:38:22
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41478",
"html_url": "https://github.com/huggingface/transformers/pull/41478",
"diff_url": "https://github.com/huggingface/transformers/pull/41478.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41478.patch",
"merged_at": null
}
|
# Fix SmolVLM2 quantization dtype mismatch
## What does this PR do?
Fixes #41453 - SmolVLM2 cannot be used with quantization due to dtype mismatch error.
**Problem**: When loading SmolVLM2 with BitsAndBytesConfig and bfloat16, the `inputs_merger` function fails with:
```
RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.
```
**Root Cause**:
- Quantization forces `inputs_embeds` to `torch.bfloat16` (from BitsAndBytesConfig)
- Vision encoder outputs `image_hidden_states` in `torch.float32`
- Direct assignment between incompatible dtypes causes the crash
**Solution**: Added dtype conversion to ensure `image_hidden_states` matches `inputs_embeds` dtype before assignment:
```python
# Before (BROKEN):
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
# After (FIXED):
# Ensure dtype compatibility for quantization
image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
**Changes**:
- Modified `src/transformers/models/smolvlm/modeling_smolvlm.py` - Added dtype conversion in `inputs_merger` function
- Updated `src/transformers/models/smolvlm/modular_smolvlm.py` - Aligned modular file with same fix
- Added test in `tests/models/smolvlm/test_modeling_smolvlm.py` - `test_quantization_dtype_compatibility()` with `@slow` decorator
**Testing**: The fix has been thoroughly tested and verified to resolve the quantization dtype mismatch issue without breaking existing functionality.
Fixes #41453
## 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?
@yonigozlan @molbap @zucchini-nlp - This affects vision models and quantization functionality
|
{
"login": "omsherikar",
"id": 180152315,
"node_id": "U_kgDOCrzn-w",
"avatar_url": "https://avatars.githubusercontent.com/u/180152315?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/omsherikar",
"html_url": "https://github.com/omsherikar",
"followers_url": "https://api.github.com/users/omsherikar/followers",
"following_url": "https://api.github.com/users/omsherikar/following{/other_user}",
"gists_url": "https://api.github.com/users/omsherikar/gists{/gist_id}",
"starred_url": "https://api.github.com/users/omsherikar/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/omsherikar/subscriptions",
"organizations_url": "https://api.github.com/users/omsherikar/orgs",
"repos_url": "https://api.github.com/users/omsherikar/repos",
"events_url": "https://api.github.com/users/omsherikar/events{/privacy}",
"received_events_url": "https://api.github.com/users/omsherikar/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41478/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/41478/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41477
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41477/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41477/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41477/events
|
https://github.com/huggingface/transformers/pull/41477
| 3,499,314,590
|
PR_kwDOCUB6oc6s58AQ
| 41,477
|
Cleaning hub kernels
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T14:00:06
| 2025-10-09T14:32:20
| 2025-10-09T14:32:18
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41477",
"html_url": "https://github.com/huggingface/transformers/pull/41477",
"diff_url": "https://github.com/huggingface/transformers/pull/41477.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41477.patch",
"merged_at": "2025-10-09T14:32:18"
}
|
# What does this PR do?
Cleaning some comments, and renaming all instances of `load_and_register_kernel` to `load_and_register_attn_kernel` since it's used only for attention kernels
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41477/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/41477/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41476
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41476/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41476/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41476/events
|
https://github.com/huggingface/transformers/pull/41476
| 3,499,137,716
|
PR_kwDOCUB6oc6s5VRJ
| 41,476
|
Pickle - part 2
|
{
"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 |
[] | 2025-10-09T13:16:11
| 2025-10-09T13:46:59
| 2025-10-09T13:46:54
|
COLLABORATOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41476",
"html_url": "https://github.com/huggingface/transformers/pull/41476",
"diff_url": "https://github.com/huggingface/transformers/pull/41476.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41476.patch",
"merged_at": "2025-10-09T13:46:54"
}
|
# What does this PR do?
Require
```
if not strtobool(os.environ.get("TRUST_REMOTE_CODE", "False")):
raise ValueError(
"This part uses `pickle.load` which is insecure and will execute arbitrary code that is potentially "
"malicious. It's recommended to never unpickle data that could have come from an untrusted source, or "
"that could have been tampered with. If you already verified the pickle data and decided to use it, "
"you can set the environment variable `TRUST_REMOTE_CODE` to `True` to allow it."
)
```
I limit this PR to conversion scripts only. For other usage of pickle in our codebase, I will try to see if there is any alternative.
|
{
"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/41476/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/41476/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41475
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41475/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41475/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41475/events
|
https://github.com/huggingface/transformers/pull/41475
| 3,499,088,182
|
PR_kwDOCUB6oc6s5KhH
| 41,475
|
Remove DISABLE_KERNEL_MAPPING flag
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T13:03:01
| 2025-10-10T08:19:27
| 2025-10-10T08:19:25
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41475",
"html_url": "https://github.com/huggingface/transformers/pull/41475",
"diff_url": "https://github.com/huggingface/transformers/pull/41475.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41475.patch",
"merged_at": "2025-10-10T08:19:25"
}
|
# What does this PR do?
Removes `DISABLE_KERNEL_MAPPING` from docker files since it's an opt in for now, and we don't use kernels by default
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41475/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/41475/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41474
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41474/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41474/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41474/events
|
https://github.com/huggingface/transformers/pull/41474
| 3,498,885,786
|
PR_kwDOCUB6oc6s4eyj
| 41,474
|
🚨 [v5] Toggle the serialization format in processors
|
{
"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 |
[] | 2025-10-09T12:06:14
| 2025-10-16T08:19:22
| 2025-10-16T08:19:22
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41474",
"html_url": "https://github.com/huggingface/transformers/pull/41474",
"diff_url": "https://github.com/huggingface/transformers/pull/41474.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41474.patch",
"merged_at": "2025-10-16T08:19:22"
}
|
# What does this PR do?
As per title, in v5 we will start saving in the new nested format. The loading is still supported though for old-format configs
Also, we don't need an `optional_attributes` field that does not do what it is called. Since it is only `chat template`, let's just manually pop and set it. For audio tokenizers, it'd be nice to refactor a bit, I think we could try to move it in `self.attributes`
Fixes https://github.com/huggingface/transformers/issues/40447 by deleting ambiguously named `optional_attributes`
|
{
"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/41474/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/41474/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41473
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41473/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41473/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41473/events
|
https://github.com/huggingface/transformers/pull/41473
| 3,498,828,973
|
PR_kwDOCUB6oc6s4ScC
| 41,473
|
Set `truncation` to `False` in Qwen3Omni to avoid default truncation
|
{
"login": "BakerBunker",
"id": 17872844,
"node_id": "MDQ6VXNlcjE3ODcyODQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/17872844?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/BakerBunker",
"html_url": "https://github.com/BakerBunker",
"followers_url": "https://api.github.com/users/BakerBunker/followers",
"following_url": "https://api.github.com/users/BakerBunker/following{/other_user}",
"gists_url": "https://api.github.com/users/BakerBunker/gists{/gist_id}",
"starred_url": "https://api.github.com/users/BakerBunker/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/BakerBunker/subscriptions",
"organizations_url": "https://api.github.com/users/BakerBunker/orgs",
"repos_url": "https://api.github.com/users/BakerBunker/repos",
"events_url": "https://api.github.com/users/BakerBunker/events{/privacy}",
"received_events_url": "https://api.github.com/users/BakerBunker/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T11:48:53
| 2025-10-10T09:56:04
| 2025-10-10T09:55:18
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41473",
"html_url": "https://github.com/huggingface/transformers/pull/41473",
"diff_url": "https://github.com/huggingface/transformers/pull/41473.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41473.patch",
"merged_at": "2025-10-10T09:55:18"
}
|
# 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/QwenLM/Qwen3-Omni/issues/40
## 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
|
{
"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/41473/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/41473/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41472
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41472/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41472/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41472/events
|
https://github.com/huggingface/transformers/issues/41472
| 3,498,589,872
|
I_kwDOCUB6oc7QiD6w
| 41,472
|
Python 3.13.8 breaks DeBERTa-v2 and SEW-D model loading
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/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 |
[] | 2025-10-09T10:34:40
| 2025-10-16T08:10:10
| 2025-10-16T08:10:10
|
MEMBER
| null | null | null | null |
Python 3.13.8 breaks DeBERTa-v2 and SEW-D model loading.
### Describe the bug
Python 3.13.8 contains a regression in the `inspect` module that breaks PyTorch JIT script compilation when comments appear between decorators and function definitions. This makes DeBERTa-v2 and SEW-D models completely unusable on Python 3.13.8.
### Reproduction
```python
# Python 3.13.8
from transformers import AutoModel
# This fails with IndentationError
model = AutoModel.from_pretrained("microsoft/deberta-v2-xlarge")
```
Error:
```python
IndentationError: expected an indented block after function definition on line 3
```
### Root Cause
The issue is in these files:
- transformers/models/deberta_v2/modeling_deberta_v2.py (lines 105-108, 111-114, 117-120)
- transformers/models/sew_d/modeling_sew_d.py (similar pattern)
Both use:
```python
@torch.jit.script
# Copied from transformers.models.deberta.modeling_deberta.xxx
def function_name(...):
...
```
Python 3.13.8's `inspect.BlockFinder` incorrectly treats the comment as the start of a lambda/generator expression, truncating the source before the function body, which then fails to parse.
- ✅ Works in Python 3.13.7 and earlier
- ❌ Broken in Python 3.13.8
- 🔄 Expected to be fixed in Python 3.13.9
### Upstream Bug
This is a Python bug tracked at:
- https://github.com/python/cpython/issues/139783
### Impact
Affected:
- ❌ All DeBERTa-v2 models cannot be loaded
- ❌ All SEW-D models cannot be loaded
- ❌ Any downstream library using these models (e.g., trl, llm-blender)
Not Affected:
- ✅ Other model architectures work fine
- ✅ Config loading works for all models
### Workaround
- For users: Use Python 3.13.7 or earlier, or wait for Python 3.13.9
- For transformers maintainers (optional): Consider removing the comment between decorator and function: move comment above the decorator
```python
# Copied from transformers.models.deberta.modeling_deberta.xxx
@torch.jit.script
def function_name(...):
...
```
### Environment
- Python: 3.13.8
- transformers version: 4.57.0
- PyTorch: 2.8.0
### Additional Context
This issue was discovered while using trl library's PairRMJudge, which uses llm-blender, which loads DeBERTa models via AutoModel.from_pretrained().
Related discussions:
- TRL issue:
- https://github.com/huggingface/trl/issues/4239
- CPython bug:
- https://github.com/python/cpython/issues/139783
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41472/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/transformers/issues/41472/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/41471
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41471/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41471/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41471/events
|
https://github.com/huggingface/transformers/pull/41471
| 3,498,546,564
|
PR_kwDOCUB6oc6s3Ugt
| 41,471
|
Update GLM-4.6 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 |
[] | 2025-10-09T10:22:19
| 2025-10-11T05:35:07
| 2025-10-09T16:18:05
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41471",
"html_url": "https://github.com/huggingface/transformers/pull/41471",
"diff_url": "https://github.com/huggingface/transformers/pull/41471.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41471.patch",
"merged_at": "2025-10-09T16:18:05"
}
|
Updated the simple description and technical blog address for GLM-4.6.
|
{
"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/41471/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/41471/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41470
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41470/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41470/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41470/events
|
https://github.com/huggingface/transformers/pull/41470
| 3,498,394,828
|
PR_kwDOCUB6oc6s2zkZ
| 41,470
|
[kernels] Cleanup deta kernel
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T09:38:55
| 2025-10-09T11:17:44
| 2025-10-09T11:17:42
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41470",
"html_url": "https://github.com/huggingface/transformers/pull/41470",
"diff_url": "https://github.com/huggingface/transformers/pull/41470.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41470.patch",
"merged_at": "2025-10-09T11:17:42"
}
|
# What does this PR do?
Cleanup the `deta` kernel since it's the same as `deformable_detr` kernel cleaned here : https://github.com/huggingface/transformers/pull/36853
And do the necessary modeling changes (even though the model is deprecated)
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41470/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/41470/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41469
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41469/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41469/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41469/events
|
https://github.com/huggingface/transformers/pull/41469
| 3,498,372,645
|
PR_kwDOCUB6oc6s2usV
| 41,469
|
Remove check_quantized_param
|
{
"login": "cyyever",
"id": 17618148,
"node_id": "MDQ6VXNlcjE3NjE4MTQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/17618148?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cyyever",
"html_url": "https://github.com/cyyever",
"followers_url": "https://api.github.com/users/cyyever/followers",
"following_url": "https://api.github.com/users/cyyever/following{/other_user}",
"gists_url": "https://api.github.com/users/cyyever/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cyyever/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cyyever/subscriptions",
"organizations_url": "https://api.github.com/users/cyyever/orgs",
"repos_url": "https://api.github.com/users/cyyever/repos",
"events_url": "https://api.github.com/users/cyyever/events{/privacy}",
"received_events_url": "https://api.github.com/users/cyyever/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T09:32:42
| 2025-10-18T00:43:23
| 2025-10-18T00:43:17
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41469",
"html_url": "https://github.com/huggingface/transformers/pull/41469",
"diff_url": "https://github.com/huggingface/transformers/pull/41469.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41469.patch",
"merged_at": null
}
|
# What does this PR do?
As the title says.
|
{
"login": "cyyever",
"id": 17618148,
"node_id": "MDQ6VXNlcjE3NjE4MTQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/17618148?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cyyever",
"html_url": "https://github.com/cyyever",
"followers_url": "https://api.github.com/users/cyyever/followers",
"following_url": "https://api.github.com/users/cyyever/following{/other_user}",
"gists_url": "https://api.github.com/users/cyyever/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cyyever/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cyyever/subscriptions",
"organizations_url": "https://api.github.com/users/cyyever/orgs",
"repos_url": "https://api.github.com/users/cyyever/repos",
"events_url": "https://api.github.com/users/cyyever/events{/privacy}",
"received_events_url": "https://api.github.com/users/cyyever/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41469/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/41469/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41468
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41468/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41468/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41468/events
|
https://github.com/huggingface/transformers/pull/41468
| 3,498,339,171
|
PR_kwDOCUB6oc6s2net
| 41,468
|
Remove KERAS_NLP_IMPORT_ERROR
|
{
"login": "cyyever",
"id": 17618148,
"node_id": "MDQ6VXNlcjE3NjE4MTQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/17618148?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cyyever",
"html_url": "https://github.com/cyyever",
"followers_url": "https://api.github.com/users/cyyever/followers",
"following_url": "https://api.github.com/users/cyyever/following{/other_user}",
"gists_url": "https://api.github.com/users/cyyever/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cyyever/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cyyever/subscriptions",
"organizations_url": "https://api.github.com/users/cyyever/orgs",
"repos_url": "https://api.github.com/users/cyyever/repos",
"events_url": "https://api.github.com/users/cyyever/events{/privacy}",
"received_events_url": "https://api.github.com/users/cyyever/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T09:22:23
| 2025-10-09T12:01:43
| 2025-10-09T11:58:31
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41468",
"html_url": "https://github.com/huggingface/transformers/pull/41468",
"diff_url": "https://github.com/huggingface/transformers/pull/41468.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41468.patch",
"merged_at": "2025-10-09T11:58:31"
}
|
# What does this PR do?
Because `KERAS_NLP_IMPORT_ERROR` is not used.
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41468/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/41468/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41467
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41467/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41467/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41467/events
|
https://github.com/huggingface/transformers/pull/41467
| 3,498,332,026
|
PR_kwDOCUB6oc6s2l8H
| 41,467
|
Add check for kernel config
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T09:20:01
| 2025-10-16T13:01:50
| 2025-10-16T13:01:50
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41467",
"html_url": "https://github.com/huggingface/transformers/pull/41467",
"diff_url": "https://github.com/huggingface/transformers/pull/41467.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41467.patch",
"merged_at": null
}
|
# What does this PR do?
Erroring when `kernel_config` is provided without specifically setting `use_kernels` to `True`
|
{
"login": "MekkCyber",
"id": 93391238,
"node_id": "U_kgDOBZEJhg",
"avatar_url": "https://avatars.githubusercontent.com/u/93391238?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/MekkCyber",
"html_url": "https://github.com/MekkCyber",
"followers_url": "https://api.github.com/users/MekkCyber/followers",
"following_url": "https://api.github.com/users/MekkCyber/following{/other_user}",
"gists_url": "https://api.github.com/users/MekkCyber/gists{/gist_id}",
"starred_url": "https://api.github.com/users/MekkCyber/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/MekkCyber/subscriptions",
"organizations_url": "https://api.github.com/users/MekkCyber/orgs",
"repos_url": "https://api.github.com/users/MekkCyber/repos",
"events_url": "https://api.github.com/users/MekkCyber/events{/privacy}",
"received_events_url": "https://api.github.com/users/MekkCyber/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41467/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/41467/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41466
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41466/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41466/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41466/events
|
https://github.com/huggingface/transformers/pull/41466
| 3,498,303,181
|
PR_kwDOCUB6oc6s2f5O
| 41,466
|
Try to remove `pickle` - `BloomTokenizerFast`
|
{
"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 |
[] | 2025-10-09T09:10:39
| 2025-10-10T08:52:53
| 2025-10-10T08:52:51
|
COLLABORATOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41466",
"html_url": "https://github.com/huggingface/transformers/pull/41466",
"diff_url": "https://github.com/huggingface/transformers/pull/41466.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41466.patch",
"merged_at": "2025-10-10T08:52:51"
}
|
# What does this PR do?
I can see the effect as the value is changed as expected from
```
(Pdb) self.backend_tokenizer.pre_tokenizer
Sequence(pretokenizers=[Split(pattern=Regex(" ?[^(\s|[.,!?…。，、।۔،])]+"), behavior=Isolated, invert=False), ByteLevel(add_prefix_space=False, trim_offsets=True, use_regex=False)])
```
to
```
(Pdb) self.backend_tokenizer.pre_tokenizer
Sequence(pretokenizers=[Split(pattern=Regex(" ?[^(\s|[.,!?…。，、।۔،])]+"), behavior=Isolated, invert=False), ByteLevel(add_prefix_space=True, trim_offsets=True, use_regex=False)])
```
|
{
"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/41466/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/41466/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41465
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41465/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41465/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41465/events
|
https://github.com/huggingface/transformers/issues/41465
| 3,498,216,107
|
I_kwDOCUB6oc7Qgoqr
| 41,465
|
Support s_aux in GPT-OSS
|
{
"login": "LoserCheems",
"id": 124847097,
"node_id": "U_kgDOB3ED-Q",
"avatar_url": "https://avatars.githubusercontent.com/u/124847097?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LoserCheems",
"html_url": "https://github.com/LoserCheems",
"followers_url": "https://api.github.com/users/LoserCheems/followers",
"following_url": "https://api.github.com/users/LoserCheems/following{/other_user}",
"gists_url": "https://api.github.com/users/LoserCheems/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LoserCheems/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LoserCheems/subscriptions",
"organizations_url": "https://api.github.com/users/LoserCheems/orgs",
"repos_url": "https://api.github.com/users/LoserCheems/repos",
"events_url": "https://api.github.com/users/LoserCheems/events{/privacy}",
"received_events_url": "https://api.github.com/users/LoserCheems/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 |
[] | 2025-10-09T08:47:28
| 2025-10-09T14:51:11
| null |
CONTRIBUTOR
| null | null | null | null |
### Feature request
Flash Attention currently can’t consume the `s_aux`/learnable bias head that many architectures (e.g. GPT-OSS) rely on—see the open discussion in [Dao-AILab/flash-attention#1797](https://github.com/Dao-AILab/flash-attention/issues/1797). We’d like Transformers to expose an optional backend hook for [Flash-DMATTN](https://github.com/SmallDoges/flash-dmattn), which already supports the same Flash-Attention-style API plus learnable bias tensors out of the box. Integrating Flash-DMATTN as a selectable backend would unblock models that need `s_aux` today while keeping the existing Flash Attention pathways unchanged for users who don’t opt in.
### Motivation
Right now, Flash Attention is the default fast path for many HF models, but the missing `s_aux` handling prevents us from enabling learnable-bias features (attention sinks, trainable mask bias, flash-decoding helpers) in production. Downstream projects either fall back to slower SDPA or maintain private forks. Flash-DMATTN already ships kernels with full bias gradients and has parity with Flash Attention on the rest of the API surface, so wiring it in would immediately restore feature coverage without waiting on upstream CUDA updates. This request is prompted by the stalled upstream issue above and by our own deployments that need learnable bias with long-context inference.
### Your contribution
We can provide a PR that:
- Adds a new optional backend selector (similar to `flash_attn_2` / `flash_attn_3`) that dispatches to Flash-DMATTN when installed.
- Introduces lightweight capability checks so models that request learnable bias automatically choose the compatible backend.
- Supplies documentation snippets and equivalence benchmarks (matching the Flash-DMATTN suite) to ease review.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41465/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/41465/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/41464
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41464/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41464/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41464/events
|
https://github.com/huggingface/transformers/pull/41464
| 3,497,862,133
|
PR_kwDOCUB6oc6s1AWK
| 41,464
|
Fix auto model configuration for encoder of perceptionlm
|
{
"login": "fschlatt",
"id": 23191892,
"node_id": "MDQ6VXNlcjIzMTkxODky",
"avatar_url": "https://avatars.githubusercontent.com/u/23191892?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/fschlatt",
"html_url": "https://github.com/fschlatt",
"followers_url": "https://api.github.com/users/fschlatt/followers",
"following_url": "https://api.github.com/users/fschlatt/following{/other_user}",
"gists_url": "https://api.github.com/users/fschlatt/gists{/gist_id}",
"starred_url": "https://api.github.com/users/fschlatt/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/fschlatt/subscriptions",
"organizations_url": "https://api.github.com/users/fschlatt/orgs",
"repos_url": "https://api.github.com/users/fschlatt/repos",
"events_url": "https://api.github.com/users/fschlatt/events{/privacy}",
"received_events_url": "https://api.github.com/users/fschlatt/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-09T06:50:18
| 2025-10-09T12:08:04
| 2025-10-09T12:08:03
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41464",
"html_url": "https://github.com/huggingface/transformers/pull/41464",
"diff_url": "https://github.com/huggingface/transformers/pull/41464.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41464.patch",
"merged_at": "2025-10-09T12:08:03"
}
|
# 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 #41387
## 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?
- [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).
- [ ] 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.
@ArthurZucker @Cyrilvallez
|
{
"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/41464/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/41464/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41463
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41463/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41463/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41463/events
|
https://github.com/huggingface/transformers/issues/41463
| 3,497,524,185
|
I_kwDOCUB6oc7Qd_vZ
| 41,463
|
Pop catch kwarg in run_hp_search_optuna() for flexible Optuna error handling
|
{
"login": "nicha-api",
"id": 115199576,
"node_id": "U_kgDOBt3OWA",
"avatar_url": "https://avatars.githubusercontent.com/u/115199576?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/nicha-api",
"html_url": "https://github.com/nicha-api",
"followers_url": "https://api.github.com/users/nicha-api/followers",
"following_url": "https://api.github.com/users/nicha-api/following{/other_user}",
"gists_url": "https://api.github.com/users/nicha-api/gists{/gist_id}",
"starred_url": "https://api.github.com/users/nicha-api/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/nicha-api/subscriptions",
"organizations_url": "https://api.github.com/users/nicha-api/orgs",
"repos_url": "https://api.github.com/users/nicha-api/repos",
"events_url": "https://api.github.com/users/nicha-api/events{/privacy}",
"received_events_url": "https://api.github.com/users/nicha-api/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 |
[] | 2025-10-09T03:39:46
| 2025-10-14T14:03:25
| 2025-10-14T14:03:25
|
CONTRIBUTOR
| null | null | null | null |
### Feature request
Allow `catch` kwarg passthrough in `Trainer.hyperparameter_search(backend="optuna")` in order to expose Optuna’s `Study.optimize(..., catch=...)` argument by popping out catch kwargs before calling Optuna's `study.optimize()`.
### Motivation
The current Hugging Face Transformers integration with Optuna does not expose Optuna’s Study.optimize(..., catch=...) argument.
As a result, when a trial raises an exception (e.g., CUDA OOM, data mismatch, tokenizer failure), the entire hyperparameter search stops prematurely instead of marking the trial as failed or pruned.
Adding support for a catch parameter—forwarded to Optuna’s backend—would allow users to continue tuning even when individual trials encounter errors.
### Your contribution
[PR#41496](https://github.com/huggingface/transformers/pull/41496)
|
{
"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/41463/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/41463/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/41462
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41462/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41462/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41462/events
|
https://github.com/huggingface/transformers/issues/41462
| 3,497,371,098
|
I_kwDOCUB6oc7QdaXa
| 41,462
|
Support encoder text classification for sequence to sequence models like BART and T5
|
{
"login": "cbhyphen",
"id": 12734117,
"node_id": "MDQ6VXNlcjEyNzM0MTE3",
"avatar_url": "https://avatars.githubusercontent.com/u/12734117?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cbhyphen",
"html_url": "https://github.com/cbhyphen",
"followers_url": "https://api.github.com/users/cbhyphen/followers",
"following_url": "https://api.github.com/users/cbhyphen/following{/other_user}",
"gists_url": "https://api.github.com/users/cbhyphen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cbhyphen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cbhyphen/subscriptions",
"organizations_url": "https://api.github.com/users/cbhyphen/orgs",
"repos_url": "https://api.github.com/users/cbhyphen/repos",
"events_url": "https://api.github.com/users/cbhyphen/events{/privacy}",
"received_events_url": "https://api.github.com/users/cbhyphen/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 |
[] | 2025-10-09T01:59:49
| 2025-10-22T20:18:38
| null |
NONE
| null | null | null | null |
### Feature request
Add support for encoder-only text classification for models like BART and T5.
### Motivation
The current sequence classifiers for BART and T5 include the full encoder-decoder. While the decoder is used to produce the sentence representation for the classification head, the following paper shows that mean pooling of encoder final hidden states may be a better option (at least for T5): https://arxiv.org/abs/2108.08877. In addition to this, not using the decoder would reduce the size of the model and speedup fine-tuning as well as inference.
### Your contribution
PR Submission
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41462/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/41462/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/41461
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41461/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41461/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41461/events
|
https://github.com/huggingface/transformers/pull/41461
| 3,496,894,449
|
PR_kwDOCUB6oc6sx1Rt
| 41,461
|
Fix Marian ONNX export to match PyTorch greedy decoding
|
{
"login": "prajeeta15",
"id": 96904203,
"node_id": "U_kgDOBcakCw",
"avatar_url": "https://avatars.githubusercontent.com/u/96904203?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/prajeeta15",
"html_url": "https://github.com/prajeeta15",
"followers_url": "https://api.github.com/users/prajeeta15/followers",
"following_url": "https://api.github.com/users/prajeeta15/following{/other_user}",
"gists_url": "https://api.github.com/users/prajeeta15/gists{/gist_id}",
"starred_url": "https://api.github.com/users/prajeeta15/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/prajeeta15/subscriptions",
"organizations_url": "https://api.github.com/users/prajeeta15/orgs",
"repos_url": "https://api.github.com/users/prajeeta15/repos",
"events_url": "https://api.github.com/users/prajeeta15/events{/privacy}",
"received_events_url": "https://api.github.com/users/prajeeta15/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[
{
"id": 9258341780,
"node_id": "LA_kwDOCUB6oc8AAAACJ9cVlA",
"url": "https://api.github.com/repos/huggingface/transformers/labels/Code%20agent%20slop",
"name": "Code agent slop",
"color": "C59579",
"default": false,
"description": ""
}
] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T21:04:45
| 2025-10-09T11:43:05
| 2025-10-09T11:43:05
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41461",
"html_url": "https://github.com/huggingface/transformers/pull/41461",
"diff_url": "https://github.com/huggingface/transformers/pull/41461.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41461.patch",
"merged_at": null
}
|
issue #41430
Summary:
This PR resolves the mismatch between Marian ONNX decoding and Hugging Face generate(). ONNX outputs now match PyTorch token-by-token for greedy decoding.
What changed:
- Added MarianDecoderOnnxWrapper (decoder-only) to:
1. Pass encoder_hidden_states and encoder_attention_mask at every step (cross-attention enabled)
2. Apply lm_head + final_logits_bias to logits (required by Marian)
3. Provide a clean, ONNX-friendly forward method for export
- Added a self-contained ONNX parity test (tests/onnx/marian_onnx_test.py):
1. Exports decoder-only ONNX with dynamic axes
2. Starts decoding from decoder_start_token_id
3. Greedily decodes by feeding the full decoded sequence step-by-step (no cache reuse)
4. Asserts token-by-token equality between ONNX and PyTorch outputs
Why this (might) work :
- Marian’s learned bias and cross-attention are required to replicate greedy decoding exactly.
- The test mirrors generate()’s path, ensuring ONNX produces identical tokens.
Impact:
- No changes to core model behavior.
- Wrapper is opt-in and only used for ONNX export.
Additional files / examples:
- examples/onnx/run_local_marian_decoder_test.py – A standalone script to test Marian ONNX decoder locally with any input sentence. Demonstrates step-by-step decoding and token-by-token parity with PyTorch generate() output.
- onnx/run_local_decoder_test_mbart.py - same as above but for Mbart, had similar issues.
Resolves the reported Marian ONNX mismatch.
Before:
<img width="1556" height="217" alt="Screenshot 2025-10-05 132250" src="https://github.com/user-attachments/assets/518b98bf-8bf7-408e-972e-10de93aa8da2" />
After:
<img width="1362" height="174" alt="Screenshot 2025-10-09 015547" src="https://github.com/user-attachments/assets/95049844-5748-4cf2-979d-57ded75a8b20" />
<img width="1026" height="710" alt="Screenshot 2025-10-09 002838" src="https://github.com/user-attachments/assets/4b8e1c4a-4d75-4349-8d61-673b74ab3732" />
I am just trying to learn and thought this might work... this might have some errors, but let me know how to improve on those. thank you :)
|
{
"login": "Rocketknight1",
"id": 12866554,
"node_id": "MDQ6VXNlcjEyODY2NTU0",
"avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Rocketknight1",
"html_url": "https://github.com/Rocketknight1",
"followers_url": "https://api.github.com/users/Rocketknight1/followers",
"following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}",
"gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions",
"organizations_url": "https://api.github.com/users/Rocketknight1/orgs",
"repos_url": "https://api.github.com/users/Rocketknight1/repos",
"events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Rocketknight1/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41461/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/41461/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41460
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41460/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41460/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41460/events
|
https://github.com/huggingface/transformers/pull/41460
| 3,496,750,095
|
PR_kwDOCUB6oc6sxWJe
| 41,460
|
Apply PR #40546
|
{
"login": "Galigator",
"id": 9264392,
"node_id": "MDQ6VXNlcjkyNjQzOTI=",
"avatar_url": "https://avatars.githubusercontent.com/u/9264392?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Galigator",
"html_url": "https://github.com/Galigator",
"followers_url": "https://api.github.com/users/Galigator/followers",
"following_url": "https://api.github.com/users/Galigator/following{/other_user}",
"gists_url": "https://api.github.com/users/Galigator/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Galigator/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Galigator/subscriptions",
"organizations_url": "https://api.github.com/users/Galigator/orgs",
"repos_url": "https://api.github.com/users/Galigator/repos",
"events_url": "https://api.github.com/users/Galigator/events{/privacy}",
"received_events_url": "https://api.github.com/users/Galigator/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T20:12:52
| 2025-10-08T20:15:26
| 2025-10-08T20:15:26
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41460",
"html_url": "https://github.com/huggingface/transformers/pull/41460",
"diff_url": "https://github.com/huggingface/transformers/pull/41460.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41460.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
- [ ] 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 @Cyrilvallez
- vision models: @yonigozlan @molbap
- audio models: @eustlb @ebezzam @vasqu
- multimodal models: @zucchini-nlp
- graph models: @clefourrier
Library:
- generate: @zucchini-nlp (visual-language models) or @gante (all others)
- continuous batching: @remi-or @ArthurZucker @McPatate
- pipelines: @Rocketknight1
- tokenizers: @ArthurZucker and @itazap
- trainer: @zach-huggingface @SunMarc
- attention: @vasqu @ArthurZucker @CyrilVallez
- model loading (from pretrained, etc): @CyrilVallez
- distributed: @3outeille @ArthurZucker @S1ro1
- CIs: @ydshieh
Integrations:
- deepspeed: HF Trainer/Accelerate: @SunMarc @zach-huggingface
- ray/raytune: @richardliaw, @amogkam
- Big Model Inference: @SunMarc
- quantization (bitsandbytes, autogpt): @SunMarc @MekkCyber
- kernels: @MekkCyber @drbh
- peft: @BenjaminBossan @githubnemo
Devices/Backends:
- AMD ROCm: @ivarflakstad
- Intel XPU: @IlyasMoutawwakil
- Ascend NPU: @ivarflakstad
Documentation: @stevhliu
Research projects are not maintained and should be taken as is.
-->
|
{
"login": "Galigator",
"id": 9264392,
"node_id": "MDQ6VXNlcjkyNjQzOTI=",
"avatar_url": "https://avatars.githubusercontent.com/u/9264392?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Galigator",
"html_url": "https://github.com/Galigator",
"followers_url": "https://api.github.com/users/Galigator/followers",
"following_url": "https://api.github.com/users/Galigator/following{/other_user}",
"gists_url": "https://api.github.com/users/Galigator/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Galigator/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Galigator/subscriptions",
"organizations_url": "https://api.github.com/users/Galigator/orgs",
"repos_url": "https://api.github.com/users/Galigator/repos",
"events_url": "https://api.github.com/users/Galigator/events{/privacy}",
"received_events_url": "https://api.github.com/users/Galigator/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41460/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/41460/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41459
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41459/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41459/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41459/events
|
https://github.com/huggingface/transformers/pull/41459
| 3,496,484,488
|
PR_kwDOCUB6oc6swctc
| 41,459
|
[docs] Standardize model docs
|
{
"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
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-08T18:36:51
| 2025-10-28T08:19:19
| null |
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41459",
"html_url": "https://github.com/huggingface/transformers/pull/41459",
"diff_url": "https://github.com/huggingface/transformers/pull/41459.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41459.patch",
"merged_at": null
}
|
- standardizes all model docs to include a generated summary of the abstract/blog, a `Pipeline` and `AutoModel` or `ModelForTask` example and usage tips
- updates `add_dates.py` to also add contributor names at the top
- removes PyTorch badges because we are PyTorch-first now
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41459/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/41459/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41458
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41458/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41458/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41458/events
|
https://github.com/huggingface/transformers/pull/41458
| 3,496,389,947
|
PR_kwDOCUB6oc6swIst
| 41,458
|
Adding ScatterMoE kernel support for Granite models.
|
{
"login": "shawntan",
"id": 119799,
"node_id": "MDQ6VXNlcjExOTc5OQ==",
"avatar_url": "https://avatars.githubusercontent.com/u/119799?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/shawntan",
"html_url": "https://github.com/shawntan",
"followers_url": "https://api.github.com/users/shawntan/followers",
"following_url": "https://api.github.com/users/shawntan/following{/other_user}",
"gists_url": "https://api.github.com/users/shawntan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/shawntan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/shawntan/subscriptions",
"organizations_url": "https://api.github.com/users/shawntan/orgs",
"repos_url": "https://api.github.com/users/shawntan/repos",
"events_url": "https://api.github.com/users/shawntan/events{/privacy}",
"received_events_url": "https://api.github.com/users/shawntan/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-08T18:05:32
| 2025-10-20T17:34:31
| null |
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41458",
"html_url": "https://github.com/huggingface/transformers/pull/41458",
"diff_url": "https://github.com/huggingface/transformers/pull/41458.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41458.patch",
"merged_at": null
}
|
# What does this PR do?
Adds ScatterMoE kernel support for Granite MoE models.
Started in #40365 but has significantly deviated in approach, so starting a new pull request.
## Before submitting
- [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#create-a-pull-request),
Pull Request section?
- [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
- [ ] 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?
@MekkCyber already started to provide some comments in #40365.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41458/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/41458/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41457
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41457/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41457/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41457/events
|
https://github.com/huggingface/transformers/pull/41457
| 3,496,373,582
|
PR_kwDOCUB6oc6swFW0
| 41,457
|
Fix doc
|
{
"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 |
[] | 2025-10-08T17:59:30
| 2025-10-08T18:15:58
| 2025-10-08T18:13:21
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41457",
"html_url": "https://github.com/huggingface/transformers/pull/41457",
"diff_url": "https://github.com/huggingface/transformers/pull/41457.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41457.patch",
"merged_at": "2025-10-08T18:13:21"
}
|
# What does this PR do?
|
{
"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/41457/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/41457/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41456
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41456/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41456/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41456/events
|
https://github.com/huggingface/transformers/pull/41456
| 3,496,115,546
|
PR_kwDOCUB6oc6svOOr
| 41,456
|
Add vision contribution guide
|
{
"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 |
[] | 2025-10-08T16:24:56
| 2025-10-20T16:56:50
| 2025-10-20T16:56:48
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41456",
"html_url": "https://github.com/huggingface/transformers/pull/41456",
"diff_url": "https://github.com/huggingface/transformers/pull/41456.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41456.patch",
"merged_at": "2025-10-20T16:56:48"
}
|
# What does this PR do?
This updates the vision contribution guide, to help external contributors merge their work faster and reduce the load on maintainers with a checklist. It is a very minimal checklist that will need to be completed with specific model implementation guides following the new API.
|
{
"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/41456/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/41456/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41455
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41455/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41455/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41455/events
|
https://github.com/huggingface/transformers/issues/41455
| 3,495,880,613
|
I_kwDOCUB6oc7QXuel
| 41,455
|
RuntimeError with dtype=torch.float16 and dtype=torch.bfloat16 in HQQ 4-bit, 8-bit quantized Granite4 models on RTX3050 TI
|
{
"login": "prathamesh-chavan-22",
"id": 158864599,
"node_id": "U_kgDOCXgU1w",
"avatar_url": "https://avatars.githubusercontent.com/u/158864599?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/prathamesh-chavan-22",
"html_url": "https://github.com/prathamesh-chavan-22",
"followers_url": "https://api.github.com/users/prathamesh-chavan-22/followers",
"following_url": "https://api.github.com/users/prathamesh-chavan-22/following{/other_user}",
"gists_url": "https://api.github.com/users/prathamesh-chavan-22/gists{/gist_id}",
"starred_url": "https://api.github.com/users/prathamesh-chavan-22/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/prathamesh-chavan-22/subscriptions",
"organizations_url": "https://api.github.com/users/prathamesh-chavan-22/orgs",
"repos_url": "https://api.github.com/users/prathamesh-chavan-22/repos",
"events_url": "https://api.github.com/users/prathamesh-chavan-22/events{/privacy}",
"received_events_url": "https://api.github.com/users/prathamesh-chavan-22/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-08T15:16:56
| 2025-10-17T06:24:13
| null |
CONTRIBUTOR
| null | null | null | null |
### Environment
- Transformers version: 4.57.0
- PyTorch version: 2.8.0+cu128
- Accelerate version: 1.10.1
- GPU: Nvidia RTX 3050Ti 4GB VRAM
### Steps to reproduce
1. Load HQQ 4-bit Granite model using dtype=torch.float16
```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, HqqConfig
# 1️⃣ Quantization config
quant_config = HqqConfig(nbits=8, group_size=64)
# 2️⃣ Load model
model_name = "ibm-granite/granite-4.0-h-micro"
model = AutoModelForCausalLM.from_pretrained(
model_name,
dtype=torch.bfloat16,
device_map="cuda", # "auto or "cuda" if single GPU
quantization_config=quant_config
)
# 3️⃣ Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)
# 4️⃣ Prepare prompt
prompt = "Translate the following English sentence to French: Hello, how are you?"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
# 5️⃣ Generate output
with torch.no_grad():
outputs = model.generate(
**inputs,
max_new_tokens=50,
do_sample=True, # optional: enable sampling for more diverse output
temperature=0.7, # optional: adjust creativity
top_p=0.9 # optional: nucleus sampling
)
# 6️⃣ Decode output
generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(generated_text)
```
Observed behavior
RuntimeError: expected mat1 and mat2 to have the same dtype (struct c10::BFloat16 != float)
Expected behavior
Model should generate outputs without dtype errors
Workaround
1. Cast inputs to float32 before generation:
```python
inputs = {k: v.float() for k, v in inputs.items()}
```
2. Use dtype=torch.float32 or don't mention it at all
**Notes**
1. Docs show dtype=torch.float16
2. Error occurs in some PyTorch/Accelerate setups even on GPUs that support float16/bfloat16
@MekkCyber
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41455/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/41455/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/41454
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41454/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41454/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41454/events
|
https://github.com/huggingface/transformers/issues/41454
| 3,495,742,177
|
I_kwDOCUB6oc7QXMrh
| 41,454
|
Huggingface wav2vec2-Conformer conversion script fails for Fairseq models
|
{
"login": "Getmany1",
"id": 26164540,
"node_id": "MDQ6VXNlcjI2MTY0NTQw",
"avatar_url": "https://avatars.githubusercontent.com/u/26164540?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Getmany1",
"html_url": "https://github.com/Getmany1",
"followers_url": "https://api.github.com/users/Getmany1/followers",
"following_url": "https://api.github.com/users/Getmany1/following{/other_user}",
"gists_url": "https://api.github.com/users/Getmany1/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Getmany1/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Getmany1/subscriptions",
"organizations_url": "https://api.github.com/users/Getmany1/orgs",
"repos_url": "https://api.github.com/users/Getmany1/repos",
"events_url": "https://api.github.com/users/Getmany1/events{/privacy}",
"received_events_url": "https://api.github.com/users/Getmany1/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": ""
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-08T14:40:56
| 2025-10-13T08:37:11
| null |
NONE
| null | null | null | null |
### System Info
- `transformers` version: 4.57.0.dev0
- Platform: Linux-4.18.0-553.74.1.el8_10.x86_64-x86_64-with-glibc2.28
- Python version: 3.10.14
- Huggingface_hub version: 1.0.0.rc2
- Safetensors version: 0.6.2
- Accelerate version: not installed
- Accelerate config: not found
- DeepSpeed version: not installed
- PyTorch version (accelerator?): 2.5.1+cu124 (CUDA)
- Using distributed or parallel set-up in script?: <fill in>
- Using GPU in script?: No
- GPU type: Tesla V100-SXM2-32GB
### Who can help?
@eustlb @amyeroberts
### 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
The official conversion script (`transformers/models/wav2vec2_conformer/convert_wav2vec2_conformer_original_pytorch_checkpoint_to_pytorch.py`) fails to properly convert Fairseq wav2vec2-Conformer models. Note that standard Wav2Vec2 models with the Transformer encoder convert successfully - this issue is specific to models with the Conformer encoder.
**Test Model:** [GetmanY1/wav2vec2-conformer-rel-pos-base-fi-ft](https://huggingface.co/GetmanY1/wav2vec2-conformer-rel-pos-base-fi-ft) (includes both original Fairseq checkpoint and converted model for reproducibility)
**Fairseq Model Configuration:**
```yaml
model:
_name: wav2vec2
quantize_targets: true
latent_temp: [2.0, 0.5, 0.999995]
extractor_mode: default
layer_norm_first: false
final_dim: 256
dropout_input: 0.1
dropout_features: 0.1
feature_grad_mult: 0.1
encoder_embed_dim: 768
encoder_layers: 12
encoder_ffn_embed_dim: 3072
encoder_attention_heads: 12
activation_fn: gelu
dropout: 0.1
attention_dropout: 0.1
activation_dropout: 0.0
encoder_layerdrop: 0.0
layer_type: conformer
attn_type: espnet
pos_enc_type: rel_pos
```
**Steps to reproduce the behavior**
1. Run the conversion script:
```bash
python3 transformers/models/wav2vec2_conformer/convert_wav2vec2_conformer_original_pytorch_checkpoint_to_pytorch.py \
--pytorch_dump_folder_path="/converted" \
--checkpoint_path="/path/to/checkpoint_best.pt" \
--dict_path="/path/to/dict.ltr.txt"
```
2. **Error 1:**` AttributeError: 'ParametrizedConv1d' object has no attribute 'weight_g'`
**Fix:** Update lines 99 and 101 in `convert_wav2vec2_conformer_original_pytorch_checkpoint_to_pytorch.py`:
```python
# Line 99: Change
hf_pointer.weight_g.data = value
# To:
hf_pointer.parametrizations.weight.original0.data = value
# Line 101: Change
hf_pointer.weight_v.data = value
# To:
hf_pointer.parametrizations.weight.original1.data = value
```
(Related: [Issue #26796](https://github.com/huggingface/transformers/issues/26796), which probably wasn't fixed in this particular conversion script. )
3. **Error 2:** Shape mismatch for pos_conv weights:
`ValueError: Shape of hf wav2vec2_conformer.encoder.pos_conv_embed.conv.weight_g is torch.Size([768, 48, 128]), but should be torch.Size([1, 1, 128]) for w2v_encoder.w2v_model.encoder.pos_conv.0.weight_g`
**Temporary workaround:** Comment out shape validation (lines 90-94) in `convert_wav2vec2_conformer_original_pytorch_checkpoint_to_pytorch.py`:
```python
if hf_shape != value.shape:
raise ValueError(
f"Shape of hf {key + '.' + weight_type if weight_type is not None else ''} is {hf_shape}, but should be"
f" {value.shape} for {full_name}"
)
```
which results in the script running without errors, but the saved converted model doesn't work correctly and only outputs blanks.
---
**P.S. Common Fairseq Setup Issues:**
If you encounter:
`omegaconf.errors.ConfigKeyError: Key 'multi_corpus_keys' not in 'AudioFinetuningConfig'`
**Fix:** Install the latest Fairseq from source: `pip install --no-build-isolation git+https://github.com/pytorch/fairseq`
If you get:
`TypeError: ConformerEncoder.build_encoder_layer() got an unexpected keyword argument 'layer_idx'`
**Fix:** Add `**kwargs` to two methods in `fairseq/models/wav2vec/wav2vec2.py`:
```python
# Line 1182: Change
def build_encoder_layer(self, args):
# To:
def build_encoder_layer(self, args, **kwargs):
# Line 1225: Change
def extract_features(self, x, padding_mask=None, tgt_layer=None):
# To:
def extract_features(self, x, padding_mask=None, tgt_layer=None, **kwargs):
```
(Fix was proposed in https://github.com/facebookresearch/fairseq/issues/5386 , but still not merged with the main branch)
### Expected behavior
The converted HuggingFace model should produce outputs matching the Fairseq model.
The original Fairseq model runs correctly. Here's a minimal example:
```python
import torch
import fairseq
from datasets import load_dataset
# Paths (update these)
MODEL_PATH = "path/to/checkpoint_best.pt"
DICT_PATH = "path/to/dict.ltr.txt"
def load_dict(path):
vocab = {0: '<s>', 1: '<pad>', 2: '</s>', 3: '<unk>'}
with open(path) as f:
for line in f:
token = line.strip().split()[0]
vocab[len(vocab)] = token
return vocab
def decode_ctc(logits, vocab):
preds = logits.argmax(dim=-1)[0]
blank_idx = logits.shape[-1] - 1
tokens = []
prev = None
for p in preds:
p = p.item()
if p != blank_idx and p != prev and p >= 4:
tokens.append(vocab[p])
prev = p
return ''.join(tokens).replace('|', ' ').strip()
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
dataset = load_dataset("google/fleurs", "fi_fi", split="test", streaming=True)
sample = next(iter(dataset))
audio = sample["audio"]["array"]
print(f"Reference: {sample['transcription']}")
models, _, _ = fairseq.checkpoint_utils.load_model_ensemble_and_task([MODEL_PATH])
model = models[0].eval().to(device)
audio_tensor = torch.from_numpy(audio).float().unsqueeze(0).to(device)
with torch.no_grad():
out = model(source=audio_tensor, padding_mask=None)
logits = out["encoder_out"].transpose(0, 1) if isinstance(out, dict) else out[0]
vocab = load_dict(DICT_PATH)
transcript = decode_ctc(logits, vocab)
print(f"Fairseq transcription: {transcript}")
```
I also tried to convert a wav2vec2-conformer-base with rotary positional embeddings instead of relative embeddings, but the converted model still works incorrectly, yielding>80% WER.
### **Impact**
This issue affects anyone trying to convert Fairseq wav2vec2-Conformer models to HuggingFace format, making it difficult to share these models with the broader research and developer community through HuggingFace Hub.
---
I've tried to include all necessary information to verify and reproduce the issue. Please let me know if you need any additional details
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41454/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/41454/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/41453
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41453/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41453/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41453/events
|
https://github.com/huggingface/transformers/issues/41453
| 3,495,732,527
|
I_kwDOCUB6oc7QXKUv
| 41,453
|
SmolVLM2 cannot be used quantized
|
{
"login": "anderlem",
"id": 120205010,
"node_id": "U_kgDOByou0g",
"avatar_url": "https://avatars.githubusercontent.com/u/120205010?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/anderlem",
"html_url": "https://github.com/anderlem",
"followers_url": "https://api.github.com/users/anderlem/followers",
"following_url": "https://api.github.com/users/anderlem/following{/other_user}",
"gists_url": "https://api.github.com/users/anderlem/gists{/gist_id}",
"starred_url": "https://api.github.com/users/anderlem/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/anderlem/subscriptions",
"organizations_url": "https://api.github.com/users/anderlem/orgs",
"repos_url": "https://api.github.com/users/anderlem/repos",
"events_url": "https://api.github.com/users/anderlem/events{/privacy}",
"received_events_url": "https://api.github.com/users/anderlem/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
}
] |
open
| false
| null |
[] | null |
[] | 2025-10-08T14:38:43
| 2025-10-13T09:49:10
| null |
NONE
| null | null | null | null |
### System Info
When loading SmolVLM2 quantized with a BitsAndBytesConfig and in bf16 and training, I get
`RuntimeError: Index put requires the source and destination dtypes match, got BFloat16 for the destination and Float for the source.` at
```
File "lib/python3.12/site-packages/transformers/models/smolvlm/modeling_smolvlm.py", line 648, in inputs_merger
image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
```
image_hidden_states seems not to match the dtype of image_embeds. I fixed it by inserting `image_hidden_states = image_hidden_states.to(dtype=inputs_embeds.dtype)` above that line.
### Who can help?
@yonigozlan @molbap
### 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
```
bnb_config: BitsAndBytesConfig = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4",
bnb_4bit_compute_dtype=torch.bfloat16
)
AutoModelForImageTextToText.from_pretrained(
"HuggingFaceTB/SmolVLM2-2.2B-Instruct",
quantization_config=bnb_config,
torch_dtype=torch.bfloat16
)
```
### Expected behavior
No error.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41453/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/41453/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/41452
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41452/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41452/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41452/events
|
https://github.com/huggingface/transformers/pull/41452
| 3,495,622,741
|
PR_kwDOCUB6oc6stjPO
| 41,452
|
Update hqq.md
|
{
"login": "prathamesh-chavan-22",
"id": 158864599,
"node_id": "U_kgDOCXgU1w",
"avatar_url": "https://avatars.githubusercontent.com/u/158864599?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/prathamesh-chavan-22",
"html_url": "https://github.com/prathamesh-chavan-22",
"followers_url": "https://api.github.com/users/prathamesh-chavan-22/followers",
"following_url": "https://api.github.com/users/prathamesh-chavan-22/following{/other_user}",
"gists_url": "https://api.github.com/users/prathamesh-chavan-22/gists{/gist_id}",
"starred_url": "https://api.github.com/users/prathamesh-chavan-22/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/prathamesh-chavan-22/subscriptions",
"organizations_url": "https://api.github.com/users/prathamesh-chavan-22/orgs",
"repos_url": "https://api.github.com/users/prathamesh-chavan-22/repos",
"events_url": "https://api.github.com/users/prathamesh-chavan-22/events{/privacy}",
"received_events_url": "https://api.github.com/users/prathamesh-chavan-22/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T14:12:14
| 2025-10-08T14:44:57
| 2025-10-08T14:44:56
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41452",
"html_url": "https://github.com/huggingface/transformers/pull/41452",
"diff_url": "https://github.com/huggingface/transformers/pull/41452.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41452.patch",
"merged_at": "2025-10-08T14:44:56"
}
|
Fixed a typo in documentation of hqq, quantization method
Documentation: @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/41452/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/41452/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41451
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41451/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41451/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41451/events
|
https://github.com/huggingface/transformers/pull/41451
| 3,495,582,298
|
PR_kwDOCUB6oc6stadX
| 41,451
|
Build a CI docker image with FA3
|
{
"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
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-08T14:02:03
| 2025-10-09T12:43:02
| null |
COLLABORATOR
| null | null | true
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41451",
"html_url": "https://github.com/huggingface/transformers/pull/41451",
"diff_url": "https://github.com/huggingface/transformers/pull/41451.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41451.patch",
"merged_at": null
}
|
# What does this PR do?
Build a CI docker image with FA3.
It is available as
huggingface/transformers-all-latest-gpu:fa3
a build is trigged at [this run](https://github.com/huggingface/transformers/actions/runs/18347217144/job/52257062261)
Let's see how it ends.
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41451/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 1,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/transformers/issues/41451/timeline
| null | null | null | null | true
| false
|
https://api.github.com/repos/huggingface/transformers/issues/41450
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41450/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41450/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41450/events
|
https://github.com/huggingface/transformers/pull/41450
| 3,495,413,243
|
PR_kwDOCUB6oc6ss1u4
| 41,450
|
Allow modifications of attributes in modular
|
{
"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 |
[] | 2025-10-08T13:17:08
| 2025-10-08T15:34:36
| 2025-10-08T15:34:35
|
CONTRIBUTOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41450",
"html_url": "https://github.com/huggingface/transformers/pull/41450",
"diff_url": "https://github.com/huggingface/transformers/pull/41450.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41450.patch",
"merged_at": null
}
|
# What does this PR do?
Often, attributes are not derived from model names. Modular files based on a given model are then bloated because there is a tiny change in an attribute, needing the entire thing to be rewritten. See https://github.com/huggingface/transformers/pull/41224/files here for instance.
This introduces an attribute renamer, from an arbitrary attribute name to another.
```python
class DINOv3ViTForImageClassification(Dinov2ForImageClassification):
_attribute_renames = {"dinov2": "dinov3"}
```
On the attached PR, this expands the modeling file correctly.
|
{
"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/41450/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/41450/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41449
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41449/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41449/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41449/events
|
https://github.com/huggingface/transformers/pull/41449
| 3,495,305,196
|
PR_kwDOCUB6oc6sseW-
| 41,449
|
Fix trainer simple tests
|
{
"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
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T12:49:08
| 2025-10-16T14:19:08
| 2025-10-15T12:09:00
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41449",
"html_url": "https://github.com/huggingface/transformers/pull/41449",
"diff_url": "https://github.com/huggingface/transformers/pull/41449.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41449.patch",
"merged_at": "2025-10-15T12:09:00"
}
|
# What does this PR do?
This PR should all simple trainer tests and some deepspeed tests.
The only remaining tests to fix are deepspeed z2 grad acc tests but this is strange why it is failing ... cc @IlyasMoutawwakil maybe you have an idea as you worked on it for `HPU`
|
{
"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/41449/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/41449/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41448
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41448/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41448/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41448/events
|
https://github.com/huggingface/transformers/pull/41448
| 3,495,225,330
|
PR_kwDOCUB6oc6ssM9p
| 41,448
|
🚨 [v5] Rename left traces of `past_key_value` in BERT-like models
|
{
"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 |
[] | 2025-10-08T12:25:14
| 2025-10-09T14:39:01
| 2025-10-09T08:44:44
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41448",
"html_url": "https://github.com/huggingface/transformers/pull/41448",
"diff_url": "https://github.com/huggingface/transformers/pull/41448.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41448.patch",
"merged_at": "2025-10-09T08:44:44"
}
|
# What does this PR do?
As per title, finalizes the work from https://github.com/huggingface/transformers/pull/41425 and renames `past_key_value` as `past_key_values` in all files without deprecation cycle
It was a find and replace, so tests might fail 😅
|
{
"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/41448/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/41448/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41447
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41447/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41447/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41447/events
|
https://github.com/huggingface/transformers/issues/41447
| 3,495,127,956
|
I_kwDOCUB6oc7QU2uU
| 41,447
|
ImportError: cannot import name 'Qwen3VLMoeForConditionalGeneration' from 'transformers'
|
{
"login": "EIIvy",
"id": 48089890,
"node_id": "MDQ6VXNlcjQ4MDg5ODkw",
"avatar_url": "https://avatars.githubusercontent.com/u/48089890?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/EIIvy",
"html_url": "https://github.com/EIIvy",
"followers_url": "https://api.github.com/users/EIIvy/followers",
"following_url": "https://api.github.com/users/EIIvy/following{/other_user}",
"gists_url": "https://api.github.com/users/EIIvy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/EIIvy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/EIIvy/subscriptions",
"organizations_url": "https://api.github.com/users/EIIvy/orgs",
"repos_url": "https://api.github.com/users/EIIvy/repos",
"events_url": "https://api.github.com/users/EIIvy/events{/privacy}",
"received_events_url": "https://api.github.com/users/EIIvy/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2025-10-08T11:56:39
| 2025-10-15T09:24:55
| null |
NONE
| null | null | null | null |
I download the latest transformers==4.57.0, the code get this error :
from transformers import Qwen3VLMoeForConditionalGeneration
ImportError: cannot import name 'Qwen3VLMoeForConditionalGeneration' from 'transformers'
| null |
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41447/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/41447/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/41446
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41446/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41446/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41446/events
|
https://github.com/huggingface/transformers/pull/41446
| 3,495,117,304
|
PR_kwDOCUB6oc6sr1fq
| 41,446
|
Enable non-streaming mode in `transformers serve`
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T11:53:19
| 2025-10-15T07:37:28
| 2025-10-15T07:37:26
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41446",
"html_url": "https://github.com/huggingface/transformers/pull/41446",
"diff_url": "https://github.com/huggingface/transformers/pull/41446.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41446.patch",
"merged_at": "2025-10-15T07:37:26"
}
|
Needs this to be merged first: https://github.com/huggingface/transformers/pull/41444
Tests and docs need to be added before undraft
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41446/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/41446/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41445
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41445/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41445/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41445/events
|
https://github.com/huggingface/transformers/pull/41445
| 3,495,093,267
|
PR_kwDOCUB6oc6srwXq
| 41,445
|
[`from_pretrained`] Small refactor `from_pretrained`: move around unrelated stuff
|
{
"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 |
[] | 2025-10-08T11:45:17
| 2025-10-15T13:13:10
| 2025-10-13T14:33:32
|
COLLABORATOR
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41445",
"html_url": "https://github.com/huggingface/transformers/pull/41445",
"diff_url": "https://github.com/huggingface/transformers/pull/41445.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41445.patch",
"merged_at": "2025-10-13T14:33:32"
}
|
# What does this PR do?
As we are moving towards having a dynamic weight loader, we need to have a simple `from_pretrained` call.
This should help
|
{
"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/41445/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/41445/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41444
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41444/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41444/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41444/events
|
https://github.com/huggingface/transformers/pull/41444
| 3,495,048,352
|
PR_kwDOCUB6oc6srmov
| 41,444
|
Streaming should be handled at the request-level rather than at the istance level
|
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T11:30:55
| 2025-10-10T08:24:57
| 2025-10-10T08:24:55
|
MEMBER
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41444",
"html_url": "https://github.com/huggingface/transformers/pull/41444",
"diff_url": "https://github.com/huggingface/transformers/pull/41444.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41444.patch",
"merged_at": "2025-10-10T08:24:55"
}
| null |
{
"login": "LysandreJik",
"id": 30755778,
"node_id": "MDQ6VXNlcjMwNzU1Nzc4",
"avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LysandreJik",
"html_url": "https://github.com/LysandreJik",
"followers_url": "https://api.github.com/users/LysandreJik/followers",
"following_url": "https://api.github.com/users/LysandreJik/following{/other_user}",
"gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}",
"starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions",
"organizations_url": "https://api.github.com/users/LysandreJik/orgs",
"repos_url": "https://api.github.com/users/LysandreJik/repos",
"events_url": "https://api.github.com/users/LysandreJik/events{/privacy}",
"received_events_url": "https://api.github.com/users/LysandreJik/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41444/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/41444/timeline
| null | null | null | null | true
| true
|
https://api.github.com/repos/huggingface/transformers/issues/41443
|
https://api.github.com/repos/huggingface/transformers
|
https://api.github.com/repos/huggingface/transformers/issues/41443/labels{/name}
|
https://api.github.com/repos/huggingface/transformers/issues/41443/comments
|
https://api.github.com/repos/huggingface/transformers/issues/41443/events
|
https://github.com/huggingface/transformers/pull/41443
| 3,495,038,108
|
PR_kwDOCUB6oc6srkbY
| 41,443
|
Qwen3: fix bool attention_mask in eager_attention_forward
|
{
"login": "ZhengKai91",
"id": 18749478,
"node_id": "MDQ6VXNlcjE4NzQ5NDc4",
"avatar_url": "https://avatars.githubusercontent.com/u/18749478?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ZhengKai91",
"html_url": "https://github.com/ZhengKai91",
"followers_url": "https://api.github.com/users/ZhengKai91/followers",
"following_url": "https://api.github.com/users/ZhengKai91/following{/other_user}",
"gists_url": "https://api.github.com/users/ZhengKai91/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ZhengKai91/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ZhengKai91/subscriptions",
"organizations_url": "https://api.github.com/users/ZhengKai91/orgs",
"repos_url": "https://api.github.com/users/ZhengKai91/repos",
"events_url": "https://api.github.com/users/ZhengKai91/events{/privacy}",
"received_events_url": "https://api.github.com/users/ZhengKai91/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
[] |
closed
| false
| null |
[] | null |
[] | 2025-10-08T11:27:37
| 2025-10-08T12:34:23
| 2025-10-08T12:34:23
|
NONE
| null | null | false
|
{
"url": "https://api.github.com/repos/huggingface/transformers/pulls/41443",
"html_url": "https://github.com/huggingface/transformers/pull/41443",
"diff_url": "https://github.com/huggingface/transformers/pull/41443.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/41443.patch",
"merged_at": null
}
|
This PR fixes a silent numerical/logic bug that occurs when a boolean attention_mask is supplied to the eager attention path.
The current implementation simply adds the mask to the attention scores, which is correct for float masks that already contain −∞ values, but produces completely wrong logits when the mask is of type torch.bool.
The fix converts a boolean mask to float and fills the masked-out positions with −∞ before the addition, guaranteeing that softmax will assign them zero probability.
No breaking change.
@vasqu @ArthurZucker @Cyrilvallez
|
{
"login": "ZhengKai91",
"id": 18749478,
"node_id": "MDQ6VXNlcjE4NzQ5NDc4",
"avatar_url": "https://avatars.githubusercontent.com/u/18749478?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ZhengKai91",
"html_url": "https://github.com/ZhengKai91",
"followers_url": "https://api.github.com/users/ZhengKai91/followers",
"following_url": "https://api.github.com/users/ZhengKai91/following{/other_user}",
"gists_url": "https://api.github.com/users/ZhengKai91/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ZhengKai91/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ZhengKai91/subscriptions",
"organizations_url": "https://api.github.com/users/ZhengKai91/orgs",
"repos_url": "https://api.github.com/users/ZhengKai91/repos",
"events_url": "https://api.github.com/users/ZhengKai91/events{/privacy}",
"received_events_url": "https://api.github.com/users/ZhengKai91/received_events",
"type": "User",
"user_view_type": "public",
"site_admin": false
}
|
{
"url": "https://api.github.com/repos/huggingface/transformers/issues/41443/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/41443/timeline
| null | null | null | null | true
| true
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.