Datasets:

DanielPFlorian

/

Transformers-Github-Issues

like 0

{ "login": "dependabot[bot]", "id": 49699333, "node_id": "MDM6Qm90NDk2OTkzMzM=", "avatar_url": "https://avatars.githubusercontent.com/in/29110?v=4", "gravatar_id": "", "url": "https://api.github.com/users/dependabot%5Bbot%5D", "html_url": "https://github.com/apps/dependabot", "followers_url": "https://api.github.com/users/dependabot%5Bbot%5D/followers", "following_url": "https://api.github.com/users/dependabot%5Bbot%5D/following{/other_user}", "gists_url": "https://api.github.com/users/dependabot%5Bbot%5D/gists{/gist_id}", "starred_url": "https://api.github.com/users/dependabot%5Bbot%5D/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/dependabot%5Bbot%5D/subscriptions", "organizations_url": "https://api.github.com/users/dependabot%5Bbot%5D/orgs", "repos_url": "https://api.github.com/users/dependabot%5Bbot%5D/repos", "events_url": "https://api.github.com/users/dependabot%5Bbot%5D/events{/privacy}", "received_events_url": "https://api.github.com/users/dependabot%5Bbot%5D/received_events", "type": "Bot", "site_admin": false }

[ { "id": 1905493434, "node_id": "MDU6TGFiZWwxOTA1NDkzNDM0", "url": "https://api.github.com/repos/huggingface/transformers/labels/dependencies", "name": "dependencies", "color": "0366d6", "default": false, "description": "Pull requests that update a dependency file" }, { "id": 6410654816, "node_id": "LA_kwDOCUB6oc8AAAABfhrUYA", "url": "https://api.github.com/repos/huggingface/transformers/labels/python", "name": "python", "color": "2b67c6", "default": false, "description": "Pull requests that update Python code" } ]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28457). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "FYI @Rocketknight1 if we see failures" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28457/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28457", "html_url": "https://github.com/huggingface/transformers/pull/28457", "diff_url": "https://github.com/huggingface/transformers/pull/28457.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28457.patch", "merged_at": 1705069735000 }

{ "login": "pseudotensor", "id": 2249614, "node_id": "MDQ6VXNlcjIyNDk2MTQ=", "avatar_url": "https://avatars.githubusercontent.com/u/2249614?v=4", "gravatar_id": "", "url": "https://api.github.com/users/pseudotensor", "html_url": "https://github.com/pseudotensor", "followers_url": "https://api.github.com/users/pseudotensor/followers", "following_url": "https://api.github.com/users/pseudotensor/following{/other_user}", "gists_url": "https://api.github.com/users/pseudotensor/gists{/gist_id}", "starred_url": "https://api.github.com/users/pseudotensor/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/pseudotensor/subscriptions", "organizations_url": "https://api.github.com/users/pseudotensor/orgs", "repos_url": "https://api.github.com/users/pseudotensor/repos", "events_url": "https://api.github.com/users/pseudotensor/events{/privacy}", "received_events_url": "https://api.github.com/users/pseudotensor/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey! Thanks for raising this, I have no idea how to fix this 🤣 python is trying to convert the object to a boolean, but I don't know what's happening other than that! 🤗 ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28456/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "nakranivaibhav", "id": 67785830, "node_id": "MDQ6VXNlcjY3Nzg1ODMw", "avatar_url": "https://avatars.githubusercontent.com/u/67785830?v=4", "gravatar_id": "", "url": "https://api.github.com/users/nakranivaibhav", "html_url": "https://github.com/nakranivaibhav", "followers_url": "https://api.github.com/users/nakranivaibhav/followers", "following_url": "https://api.github.com/users/nakranivaibhav/following{/other_user}", "gists_url": "https://api.github.com/users/nakranivaibhav/gists{/gist_id}", "starred_url": "https://api.github.com/users/nakranivaibhav/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/nakranivaibhav/subscriptions", "organizations_url": "https://api.github.com/users/nakranivaibhav/orgs", "repos_url": "https://api.github.com/users/nakranivaibhav/repos", "events_url": "https://api.github.com/users/nakranivaibhav/events{/privacy}", "received_events_url": "https://api.github.com/users/nakranivaibhav/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28455/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28455", "html_url": "https://github.com/huggingface/transformers/pull/28455", "diff_url": "https://github.com/huggingface/transformers/pull/28455.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28455.patch", "merged_at": null }

{ "login": "SamSJackson", "id": 86316114, "node_id": "MDQ6VXNlcjg2MzE2MTE0", "avatar_url": "https://avatars.githubusercontent.com/u/86316114?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SamSJackson", "html_url": "https://github.com/SamSJackson", "followers_url": "https://api.github.com/users/SamSJackson/followers", "following_url": "https://api.github.com/users/SamSJackson/following{/other_user}", "gists_url": "https://api.github.com/users/SamSJackson/gists{/gist_id}", "starred_url": "https://api.github.com/users/SamSJackson/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SamSJackson/subscriptions", "organizations_url": "https://api.github.com/users/SamSJackson/orgs", "repos_url": "https://api.github.com/users/SamSJackson/repos", "events_url": "https://api.github.com/users/SamSJackson/events{/privacy}", "received_events_url": "https://api.github.com/users/SamSJackson/received_events", "type": "User", "site_admin": false }

[]

[]

[ "I spent far too long on this. I had forgotten to instantiate my custom processor class.", "Thanks for sharing! 🤗 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28454/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "jimmymanianchira", "id": 9268915, "node_id": "MDQ6VXNlcjkyNjg5MTU=", "avatar_url": "https://avatars.githubusercontent.com/u/9268915?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jimmymanianchira", "html_url": "https://github.com/jimmymanianchira", "followers_url": "https://api.github.com/users/jimmymanianchira/followers", "following_url": "https://api.github.com/users/jimmymanianchira/following{/other_user}", "gists_url": "https://api.github.com/users/jimmymanianchira/gists{/gist_id}", "starred_url": "https://api.github.com/users/jimmymanianchira/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jimmymanianchira/subscriptions", "organizations_url": "https://api.github.com/users/jimmymanianchira/orgs", "repos_url": "https://api.github.com/users/jimmymanianchira/repos", "events_url": "https://api.github.com/users/jimmymanianchira/events{/privacy}", "received_events_url": "https://api.github.com/users/jimmymanianchira/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks for reporting, pinging @Narsil for visibility, do you have any reproducers / examples of inputs that can lead to this? ", "@jimmymanianchira this is quite unusual as safetensors do not interact with CUDA right @Narsil ? \r\nAre you sure you are using the same transformers version across your tests? How did you moved your tests to use the safetensors version? ", "Safetensors doesn' t interact with it in anyway. I supposed something else is at play.", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28453/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "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", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28452/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28452", "html_url": "https://github.com/huggingface/transformers/pull/28452", "diff_url": "https://github.com/huggingface/transformers/pull/28452.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28452.patch", "merged_at": 1704983646000 }

{ "login": "keenranger", "id": 18392918, "node_id": "MDQ6VXNlcjE4MzkyOTE4", "avatar_url": "https://avatars.githubusercontent.com/u/18392918?v=4", "gravatar_id": "", "url": "https://api.github.com/users/keenranger", "html_url": "https://github.com/keenranger", "followers_url": "https://api.github.com/users/keenranger/followers", "following_url": "https://api.github.com/users/keenranger/following{/other_user}", "gists_url": "https://api.github.com/users/keenranger/gists{/gist_id}", "starred_url": "https://api.github.com/users/keenranger/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/keenranger/subscriptions", "organizations_url": "https://api.github.com/users/keenranger/orgs", "repos_url": "https://api.github.com/users/keenranger/repos", "events_url": "https://api.github.com/users/keenranger/events{/privacy}", "received_events_url": "https://api.github.com/users/keenranger/received_events", "type": "User", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28451). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28451/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28451", "html_url": "https://github.com/huggingface/transformers/pull/28451", "diff_url": "https://github.com/huggingface/transformers/pull/28451.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28451.patch", "merged_at": 1704993973000 }

{ "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", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28450/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 1, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28450", "html_url": "https://github.com/huggingface/transformers/pull/28450", "diff_url": "https://github.com/huggingface/transformers/pull/28450.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28450.patch", "merged_at": 1704993822000 }

{ "login": "kadirnar", "id": 36204372, "node_id": "MDQ6VXNlcjM2MjA0Mzcy", "avatar_url": "https://avatars.githubusercontent.com/u/36204372?v=4", "gravatar_id": "", "url": "https://api.github.com/users/kadirnar", "html_url": "https://github.com/kadirnar", "followers_url": "https://api.github.com/users/kadirnar/followers", "following_url": "https://api.github.com/users/kadirnar/following{/other_user}", "gists_url": "https://api.github.com/users/kadirnar/gists{/gist_id}", "starred_url": "https://api.github.com/users/kadirnar/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/kadirnar/subscriptions", "organizations_url": "https://api.github.com/users/kadirnar/orgs", "repos_url": "https://api.github.com/users/kadirnar/repos", "events_url": "https://api.github.com/users/kadirnar/events{/privacy}", "received_events_url": "https://api.github.com/users/kadirnar/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This model doesn't work either.\r\n\r\nhttps://huggingface.co./Intel/dpt-beit-base-384", "cc @NielsRogge as well ", "Hi,\r\n\r\nLooks like you may need to update your Transformers version. Works for me on main", "> Hi,\r\n> \r\n> Looks like you may need to update your Transformers version. Works for me on main\r\n\r\nI updated the package version.\r\n\r\n- `transformers` version: 4.36.2\r\n- Platform: Windows-10-10.0.22631-SP0\r\n- Python version: 3.8.18\r\n- Huggingface_hub version: 0.19.4\r\n- Safetensors version: 0.4.1\r\n- Accelerate version: 0.25.0\r\n- Accelerate config: not found\r\n- PyTorch version (GPU?): 2.1.1+cu118 (True)\r\n- Tensorflow version (GPU?): not installed (NA)\r\n- Flax version (CPU?/GPU?/TPU?): not installed (NA)\r\n- Jax version: not installed\r\n- JaxLib version: not installed\r\n- Using GPU in script?: <fill in>\r\n- Using distributed or parallel set-up in script?: <fill in>\r\n\r\nThe \"Intel/dpt-beit-base-384\" model works, but the \"swinv2\" model doesn't.\r\n\r\n<img width=\"1619\" alt=\"image\" src=\"https://github.com/huggingface/transformers/assets/36204372/98ea8310-8746-4180-a696-71ccd94d87d2\">\r\n\r\n", "Ok, that's because https://github.com/huggingface/transformers/pull/27742 was merged 3 weeks ago, which probably was not part yet of 4.36.2. Hence this one will be part of the next Transformers release." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28449/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "SangbumChoi", "id": 34004152, "node_id": "MDQ6VXNlcjM0MDA0MTUy", "avatar_url": "https://avatars.githubusercontent.com/u/34004152?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SangbumChoi", "html_url": "https://github.com/SangbumChoi", "followers_url": "https://api.github.com/users/SangbumChoi/followers", "following_url": "https://api.github.com/users/SangbumChoi/following{/other_user}", "gists_url": "https://api.github.com/users/SangbumChoi/gists{/gist_id}", "starred_url": "https://api.github.com/users/SangbumChoi/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SangbumChoi/subscriptions", "organizations_url": "https://api.github.com/users/SangbumChoi/orgs", "repos_url": "https://api.github.com/users/SangbumChoi/repos", "events_url": "https://api.github.com/users/SangbumChoi/events{/privacy}", "received_events_url": "https://api.github.com/users/SangbumChoi/received_events", "type": "User", "site_admin": false }

[ { "id": 1843244711, "node_id": "MDU6TGFiZWwxODQzMjQ0NzEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/New%20model", "name": "New model", "color": "fbca04", "default": false, "description": "" } ]

[]

[ "Hi @SangbumChoi thanks for this amazing draft!\r\n\r\nIs the YOLOv6 model trained from scratch or are you using any pre-trained weights?\r\n\r\nRegarding the design, that looks great already, however we would need to include the preparation of the targets inside the image processor, such that users can pass images + targets to it, such that the image processor then outputs a `BatchFeature` containing both `pixel_values` and `labels`.\r\n\r\nSame goes for the postprocessing, we would need to make it conform our existing models, meaning that a `post_process_object_detection` method would need to be implemented. This would also allow the model to be compatible with the pipeline API, making sure it will work with the inference widgets on the hub, etc.\r\n\r\nI'll discuss with team regarding the addition of the model :)", "Hi @NielsRogge!\r\n\r\nCurrent pipeline of model yolov6n and yolov6s is all public pre-trained weights and I also tested tolerence (1e-3 rather than 1e-4). You may check [convert_yolov6_to_pytorch.py] (https://github.com/SangbumChoi/transformers/blob/yolov6/src/transformers/models/yolov6/convert_yolov6_to_pytorch.py) however you might need to install YOLOv6 and unwrap the model.pt and store pure stated_dict since it is wrapped with the python class yolo.\r\n\r\nhttps://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6n.pt\r\n\r\nAlso I think understand more than 90% of whole pipeline in transformers. So currently i'm working on training pipeline and also including minority features for `BatchFeature`, `post_process_object_detection`, and etc...\r\n\r\nI think perfect or PR-ready version might take some more time but feel free to discuss with your team and happy to get feedbacks of mandatory requirements 😄 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28448/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "fxmarty", "id": 9808326, "node_id": "MDQ6VXNlcjk4MDgzMjY=", "avatar_url": "https://avatars.githubusercontent.com/u/9808326?v=4", "gravatar_id": "", "url": "https://api.github.com/users/fxmarty", "html_url": "https://github.com/fxmarty", "followers_url": "https://api.github.com/users/fxmarty/followers", "following_url": "https://api.github.com/users/fxmarty/following{/other_user}", "gists_url": "https://api.github.com/users/fxmarty/gists{/gist_id}", "starred_url": "https://api.github.com/users/fxmarty/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/fxmarty/subscriptions", "organizations_url": "https://api.github.com/users/fxmarty/orgs", "repos_url": "https://api.github.com/users/fxmarty/repos", "events_url": "https://api.github.com/users/fxmarty/events{/privacy}", "received_events_url": "https://api.github.com/users/fxmarty/received_events", "type": "User", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28447). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28447/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28447", "html_url": "https://github.com/huggingface/transformers/pull/28447", "diff_url": "https://github.com/huggingface/transformers/pull/28447.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28447.patch", "merged_at": 1705488654000 }

{ "login": "andysingal", "id": 20493493, "node_id": "MDQ6VXNlcjIwNDkzNDkz", "avatar_url": "https://avatars.githubusercontent.com/u/20493493?v=4", "gravatar_id": "", "url": "https://api.github.com/users/andysingal", "html_url": "https://github.com/andysingal", "followers_url": "https://api.github.com/users/andysingal/followers", "following_url": "https://api.github.com/users/andysingal/following{/other_user}", "gists_url": "https://api.github.com/users/andysingal/gists{/gist_id}", "starred_url": "https://api.github.com/users/andysingal/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/andysingal/subscriptions", "organizations_url": "https://api.github.com/users/andysingal/orgs", "repos_url": "https://api.github.com/users/andysingal/repos", "events_url": "https://api.github.com/users/andysingal/events{/privacy}", "received_events_url": "https://api.github.com/users/andysingal/received_events", "type": "User", "site_admin": false }

[]

[]

[ "TransfoXL was deprecated and is now in the legacy folder see `/transformers/src/transformers/models/deprecated/transfo_xl` as it is no longer maintained", "> TransfoXL was deprecated and is now in the legacy folder see `/transformers/src/transformers/models/deprecated/transfo_xl` as it is no longer maintained\r\n\r\nAny workaround for above code to run? ", "Just here because I am having this same issue right now :( ", "Same here, i just did from transformers import Train\r\n\r\nOn Fri, Jan 12, 2024 at 9:07 AM RichardAragon ***@***.***>\r\nwrote:\r\n\r\n> Just here because I am having this same issue right now :(\r\n>\r\n> —\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/transformers/issues/28446#issuecomment-1888382637>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/AE4LJNKV23HELVQMCBAHRILYOCVXZAVCNFSM6AAAAABBWGNETKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOBYGM4DENRTG4>\r\n> .\r\n> You are receiving this because you authored the thread.Message ID:\r\n> ***@***.***>\r\n>\r\n", "` from transformers import TransfoXLForSequenceClassification` should help you. \r\ncc @ydshieh this is a regression and should throw a deprecated warning not an error! Can you have a look as you did the deprecation cycle! ", "OK, taking look into this", "@andysingal \r\n\r\nI am running\r\n\r\n```\r\nckpt = \"transfo-xl-wt103\"\r\n\r\nfrom transformers import AutoModelForSequenceClassification\r\n\r\nmodel = AutoModelForSequenceClassification.from_pretrained(ckpt)\r\n```\r\nand it works.\r\n\r\nCould you share the colab that can produce the issue?", "I got the same error after I upgraded the transformers package. If you are downloading the files from a hugging face repo, can you try removing the local model cache files, and redownload them? That worked for me. ", "following this issue...", "@andysingal @sanalsprasad @nandakishorebellammuralidhar \r\n\r\nAs mentioned, I tried on colab and I am not able to reproduce the error.\r\n\r\nCould you provide your system information by running `transformers-cli env` (command) as well as a code snippet.\r\nOr you can try to reproduce it on colab.\r\n\r\nOtherwise, I'm afraid that I won't be able to help on this.", "I tried the same processes again yesterday or the day before that threw\r\nthis error for me before. I did get the configuration error again, I had to\r\nuninstall transformers and install huggingface transformers, that fixed it\r\nthis time.\r\n\r\nThe use case was that I was trying to find tune an already quantized model.\r\nIt was a model I had already fine tuned, and I wanted to find tune it\r\nagain. If memory serves, that is the issue that brought me here too. I\r\nthink then I was attempting to merge two already merged models via mergekit.\r\n\r\nOn Mon, Jan 22, 2024, 6:26 AM Yih-Dar ***@***.***> wrote:\r\n\r\n> @andysingal <https://github.com/andysingal> @sanalsprasad\r\n> <https://github.com/sanalsprasad> @nandakishorebellammuralidhar\r\n> <https://github.com/nandakishorebellammuralidhar>\r\n>\r\n> As mentioned, I tried on colab and I am not able to reproduce the error.\r\n>\r\n> Could you provide your system information by running transformers-cli env\r\n> (command) as well as a code snippet.\r\n> Or you can try to reproduce it on colab.\r\n>\r\n> Otherwise, I'm afraid that I won't be able to help on this.\r\n>\r\n> —\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/transformers/issues/28446#issuecomment-1904115533>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/BA44S7PLUJZ3YNDOD6MUCRLYPZZKPAVCNFSM6AAAAABBWGNETKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMBUGEYTKNJTGM>\r\n> .\r\n> You are receiving this because you commented.Message ID:\r\n> ***@***.***>\r\n>\r\n", "transformers.models.transfo_xl.configuration_transfo_xl is deprecated from transformers v.4.36\r\nso install version 4.35\r\n!pip install -q -U git+https://github.com/huggingface/[email protected]\r\nand restart colab kernel.", "@kasiwoos it is deprecated, but it will continue to work. We just don't run any test against this model anymore and it won't be maintained.\r\n\r\nBut I can't reproduce the issue people reported here.", "Same issue for me, @kasiwoos fix worked. To reiterate the issue for me was if you are loading a fine tuned llama 2 8bit quantized from 2 weeks ago it won't work with the latest transformers release.", "@patruff Could you give more details on how to reproduce, please. That would be really helpful.", "> @patruff Could you give more details on how to reproduce, please. That would be really helpful.\n\n\nSure, run this on a T4 in Colab with the latest transformers\n\n# name can be any 8bit model\nname='patruff/chucklesEFT1'\n\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\nimport torch\n\nmodel_8bit = AutoModelForCausalLM.from_pretrained(name, device_map=\"auto\", load_in_8bit=True)\ntokenizer = AutoTokenizer.from_pretrained(name)", "@patruff First thanks for sharing. I am still not able to reproduce however.\r\n\r\n`name='patruff/chucklesEFT1'` is a dataset, so I changed it to `name='patruff/toxic-llama2-7b-tuneEFT1'`.\r\n\r\nOn colab, it works (even if I upgrade `transformers` to v4.37).\r\n\r\n", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28446/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "statelesshz", "id": 28150734, "node_id": "MDQ6VXNlcjI4MTUwNzM0", "avatar_url": "https://avatars.githubusercontent.com/u/28150734?v=4", "gravatar_id": "", "url": "https://api.github.com/users/statelesshz", "html_url": "https://github.com/statelesshz", "followers_url": "https://api.github.com/users/statelesshz/followers", "following_url": "https://api.github.com/users/statelesshz/following{/other_user}", "gists_url": "https://api.github.com/users/statelesshz/gists{/gist_id}", "starred_url": "https://api.github.com/users/statelesshz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/statelesshz/subscriptions", "organizations_url": "https://api.github.com/users/statelesshz/orgs", "repos_url": "https://api.github.com/users/statelesshz/repos", "events_url": "https://api.github.com/users/statelesshz/events{/privacy}", "received_events_url": "https://api.github.com/users/statelesshz/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28445/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28445", "html_url": "https://github.com/huggingface/transformers/pull/28445", "diff_url": "https://github.com/huggingface/transformers/pull/28445.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28445.patch", "merged_at": null }

{ "login": "kimbaang", "id": 154123661, "node_id": "U_kgDOCS-9jQ", "avatar_url": "https://avatars.githubusercontent.com/u/154123661?v=4", "gravatar_id": "", "url": "https://api.github.com/users/kimbaang", "html_url": "https://github.com/kimbaang", "followers_url": "https://api.github.com/users/kimbaang/followers", "following_url": "https://api.github.com/users/kimbaang/following{/other_user}", "gists_url": "https://api.github.com/users/kimbaang/gists{/gist_id}", "starred_url": "https://api.github.com/users/kimbaang/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/kimbaang/subscriptions", "organizations_url": "https://api.github.com/users/kimbaang/orgs", "repos_url": "https://api.github.com/users/kimbaang/repos", "events_url": "https://api.github.com/users/kimbaang/events{/privacy}", "received_events_url": "https://api.github.com/users/kimbaang/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey @kimbaang 👋 \r\n\r\nThis issue (https://github.com/huggingface/transformers/issues/25084) seems related -- can you check if the solutions presented there help with your problem? 🤗 ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28444/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "hackyon", "id": 1557853, "node_id": "MDQ6VXNlcjE1NTc4NTM=", "avatar_url": "https://avatars.githubusercontent.com/u/1557853?v=4", "gravatar_id": "", "url": "https://api.github.com/users/hackyon", "html_url": "https://github.com/hackyon", "followers_url": "https://api.github.com/users/hackyon/followers", "following_url": "https://api.github.com/users/hackyon/following{/other_user}", "gists_url": "https://api.github.com/users/hackyon/gists{/gist_id}", "starred_url": "https://api.github.com/users/hackyon/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/hackyon/subscriptions", "organizations_url": "https://api.github.com/users/hackyon/orgs", "repos_url": "https://api.github.com/users/hackyon/repos", "events_url": "https://api.github.com/users/hackyon/events{/privacy}", "received_events_url": "https://api.github.com/users/hackyon/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks @ArthurZucker. \r\n\r\nThe CIs are green now. Please take a look when you get a chance. Thanks!", "Yes! Will review 🤗 ", "Just rebasing the changes on top of the latest head", "Thanks for adding the tests! It seems to have caused you a lot of extra work 😓 so let' s keep it as is. A lot of tests were missing\r\n", "Thanks. \r\n\r\nRan all 3 edited tests locally with RUN_SLOW=1 and they all passed. Also rebased off of main and re-pushed to do another run of CI checks", "Test failure from tests_exotic_models seems unrelated:\r\n\r\nE Failed to import NATTEN's CPP backend. This could be due to an invalid/incomplete install. Please uninstall NATTEN (pip uninstall natten) and re-install with the correct torch build: shi-labs.com/natten\r\n\r\nERROR tests/models/dinat/test_modeling_dinat.py - RuntimeError: Failed to import transformers.models.dinat.modeling_dinat because of the following error (look up to see its traceback):\r\nERROR tests/models/nat/test_modeling_nat.py - RuntimeError: Failed to import transformers.models.nat.modeling_nat because of the following error (look up to see its traceback):\r\n\r\nIf we need the tests to pass, I can do another merge to upstream and re-push to trigger another test run.", "Don't worry it is unrelated! 🤗 we can merge without ", "> this might slow the CI a tad bit 😉\r\n\r\nbecause of the new model classes (which is of course fine) or anything else?\r\n", "Copied from of the full test suite of T5! " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28443/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28443", "html_url": "https://github.com/huggingface/transformers/pull/28443", "diff_url": "https://github.com/huggingface/transformers/pull/28443.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28443.patch", "merged_at": 1706756029000 }

{ "login": "zhuhong61", "id": 95205772, "node_id": "U_kgDOBay5jA", "avatar_url": "https://avatars.githubusercontent.com/u/95205772?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zhuhong61", "html_url": "https://github.com/zhuhong61", "followers_url": "https://api.github.com/users/zhuhong61/followers", "following_url": "https://api.github.com/users/zhuhong61/following{/other_user}", "gists_url": "https://api.github.com/users/zhuhong61/gists{/gist_id}", "starred_url": "https://api.github.com/users/zhuhong61/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zhuhong61/subscriptions", "organizations_url": "https://api.github.com/users/zhuhong61/orgs", "repos_url": "https://api.github.com/users/zhuhong61/repos", "events_url": "https://api.github.com/users/zhuhong61/events{/privacy}", "received_events_url": "https://api.github.com/users/zhuhong61/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28442/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28442", "html_url": "https://github.com/huggingface/transformers/pull/28442", "diff_url": "https://github.com/huggingface/transformers/pull/28442.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28442.patch", "merged_at": null }

{ "login": "gmftbyGMFTBY", "id": 27548710, "node_id": "MDQ6VXNlcjI3NTQ4NzEw", "avatar_url": "https://avatars.githubusercontent.com/u/27548710?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gmftbyGMFTBY", "html_url": "https://github.com/gmftbyGMFTBY", "followers_url": "https://api.github.com/users/gmftbyGMFTBY/followers", "following_url": "https://api.github.com/users/gmftbyGMFTBY/following{/other_user}", "gists_url": "https://api.github.com/users/gmftbyGMFTBY/gists{/gist_id}", "starred_url": "https://api.github.com/users/gmftbyGMFTBY/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gmftbyGMFTBY/subscriptions", "organizations_url": "https://api.github.com/users/gmftbyGMFTBY/orgs", "repos_url": "https://api.github.com/users/gmftbyGMFTBY/repos", "events_url": "https://api.github.com/users/gmftbyGMFTBY/events{/privacy}", "received_events_url": "https://api.github.com/users/gmftbyGMFTBY/received_events", "type": "User", "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" } ]

[]

[ "Good ISSUE! \r\nSimilar to the `learning_rate_target` in trlx config: https://trlx.readthedocs.io/en/docs/configs.html#trlx.data.configs.TrainConfig\r\n\r\n```\r\nlearning_rate_init (float) – Initial learning rate after ramp up\r\nlearning_rate_target (float) – Target learning rate after decay\r\n```\r\n\r\nNow the minimum learning rate cannot be configured in the transformers, it is hard-coded to 0, such as cosine_schedule or linear_schedule\r\n\r\nhttps://github.com/huggingface/transformers/blob/main/src/transformers/optimization.py#L140\r\n```\r\nreturn max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))\r\n```\r\nhttps://github.com/huggingface/transformers/blob/main/src/transformers/optimization.py#L104\r\n```\r\nreturn max(0.0, float(num_training_steps - current_step) / float(max(1, num_training_steps - num_warmup_steps)))\r\n```\r\n", "### Temporary Implementation\r\n\r\n```python\r\nimport math\r\nimport transformers\r\n\r\nlr_decay_steps = 1500\r\nmin_lr_ratio = 0.1\r\n\r\n\r\ndef _get_cosine_schedule_with_warmup_lr_lambda(\r\n current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_cycles: float\r\n):\r\n if current_step < num_warmup_steps:\r\n return float(current_step) / float(max(1, num_warmup_steps))\r\n if current_step > lr_decay_steps:\r\n return min_lr_ratio\r\n progress = float(current_step - num_warmup_steps) / float(max(1, lr_decay_steps - num_warmup_steps))\r\n coefficient = max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))\r\n return min_lr_ratio + coefficient * (1.0 - min_lr_ratio)\r\n\r\n\r\ndef add_lr_decay_limit_for_cosine_schedule():\r\n transformers.optimization._get_cosine_schedule_with_warmup_lr_lambda = _get_cosine_schedule_with_warmup_lr_lambda\r\n\r\n```\r\n\r\nOur implementation builds upon the cosine scheduler provided by `transformers.optimization`. We have introduced two new parameters:\r\n\r\n1. **`lr_decay_steps`**: This parameter signifies the maximum number of iterations for learning rate decay.\r\n \r\n2. **`min_lr_ratio`**: This parameter represents the proportion of the constant learning rate in comparison to `TrainingArguments.learning_rate` after reaching the specified `lr_decay_steps`.\r\n\r\nThe practical impact of the aforementioned scheduler is illustrated in the following figure.\r\n<img width=\"1637\" alt=\"Screenshot 2024-01-11 at 09 25 31\" src=\"https://github.com/huggingface/transformers/assets/91840504/4908e3ee-7fe9-4542-854b-21f5a3821f03\">\r\n\r\nAlternatively, we can adopt the following implementation, which features fewer parameters and a slightly different curve.\r\n```python\r\ndef _get_cosine_schedule_with_warmup_lr_lambda(\r\n current_step: int, *, num_warmup_steps: int, num_training_steps: int, num_cycles: float\r\n):\r\n if current_step < num_warmup_steps:\r\n return float(current_step) / float(max(1, num_warmup_steps))\r\n progress = float(current_step - num_warmup_steps) / float(max(1, num_training_steps - num_warmup_steps))\r\n return max(min_lr_ratio, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress)))\r\n```\r\n", "cc @muellerzr sounds good no? \r\n", "@muellerz Hello, could you please review our issue and contribution?" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28441/reactions", "total_count": 2, "+1": 2, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Moreh-LeeJunhyeok", "id": 99154015, "node_id": "U_kgDOBej4Xw", "avatar_url": "https://avatars.githubusercontent.com/u/99154015?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Moreh-LeeJunhyeok", "html_url": "https://github.com/Moreh-LeeJunhyeok", "followers_url": "https://api.github.com/users/Moreh-LeeJunhyeok/followers", "following_url": "https://api.github.com/users/Moreh-LeeJunhyeok/following{/other_user}", "gists_url": "https://api.github.com/users/Moreh-LeeJunhyeok/gists{/gist_id}", "starred_url": "https://api.github.com/users/Moreh-LeeJunhyeok/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Moreh-LeeJunhyeok/subscriptions", "organizations_url": "https://api.github.com/users/Moreh-LeeJunhyeok/orgs", "repos_url": "https://api.github.com/users/Moreh-LeeJunhyeok/repos", "events_url": "https://api.github.com/users/Moreh-LeeJunhyeok/events{/privacy}", "received_events_url": "https://api.github.com/users/Moreh-LeeJunhyeok/received_events", "type": "User", "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" } ]

[]

[ "what is a status of this issue? is this on progress? thanks in advance ", "Hi @Moreh-LeeJunhyeok, thanks for opening an issue! \r\n\r\nNote - making experiments easier isn't in and of itself enough of a reason to add something to a model. However, as there was the equivalent added to Llama, this seems reasonable. Could you open a PR and we can review your proposed changes? " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28440/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "amyeroberts", "id": 22614925, "node_id": "MDQ6VXNlcjIyNjE0OTI1", "avatar_url": "https://avatars.githubusercontent.com/u/22614925?v=4", "gravatar_id": "", "url": "https://api.github.com/users/amyeroberts", "html_url": "https://github.com/amyeroberts", "followers_url": "https://api.github.com/users/amyeroberts/followers", "following_url": "https://api.github.com/users/amyeroberts/following{/other_user}", "gists_url": "https://api.github.com/users/amyeroberts/gists{/gist_id}", "starred_url": "https://api.github.com/users/amyeroberts/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/amyeroberts/subscriptions", "organizations_url": "https://api.github.com/users/amyeroberts/orgs", "repos_url": "https://api.github.com/users/amyeroberts/repos", "events_url": "https://api.github.com/users/amyeroberts/events{/privacy}", "received_events_url": "https://api.github.com/users/amyeroberts/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks for this work ! I don't see the PR's documentation though, would be nice to take a look at the rendered stuff.\r\n\r\nAlso would this be interesting @amyeroberts : https://github.com/huggingface/doc-builder/issues/465 ? (I've seen a discrepancy between docstring and typing in safetensors, and thought this might be interesting in transformers if done in core specific places potentially).", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28439). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "@pcuenca Thank you for your review! I believe I've addressed all of your comments. \r\n\r\n@Narsil We finally have the docs 🥳 Re typing and docstring checks, I'm definitely pro when the types are given. We currently something which does this in the check_docstrings - [specifically here](https://github.com/huggingface/transformers/blob/708b19eb093bcfb4efcfa229925b249397c2bd5e/utils/check_docstrings.py#L976). However, all of the pipeline classes are ignored because they accept arguments which aren't documented e.g. most accept `tokenizer` even if it isn't used. Once possibility is explicitly setting the documented kwargs in the `__init__` and passing the rest through `**kwargs` in this or a follow up PR. WDYT? " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28439/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28439", "html_url": "https://github.com/huggingface/transformers/pull/28439", "diff_url": "https://github.com/huggingface/transformers/pull/28439.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28439.patch", "merged_at": 1706633697000 }

{ "login": "jeffhataws", "id": 56947987, "node_id": "MDQ6VXNlcjU2OTQ3OTg3", "avatar_url": "https://avatars.githubusercontent.com/u/56947987?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jeffhataws", "html_url": "https://github.com/jeffhataws", "followers_url": "https://api.github.com/users/jeffhataws/followers", "following_url": "https://api.github.com/users/jeffhataws/following{/other_user}", "gists_url": "https://api.github.com/users/jeffhataws/gists{/gist_id}", "starred_url": "https://api.github.com/users/jeffhataws/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jeffhataws/subscriptions", "organizations_url": "https://api.github.com/users/jeffhataws/orgs", "repos_url": "https://api.github.com/users/jeffhataws/repos", "events_url": "https://api.github.com/users/jeffhataws/events{/privacy}", "received_events_url": "https://api.github.com/users/jeffhataws/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28438/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "JuanFKurucz", "id": 31422367, "node_id": "MDQ6VXNlcjMxNDIyMzY3", "avatar_url": "https://avatars.githubusercontent.com/u/31422367?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JuanFKurucz", "html_url": "https://github.com/JuanFKurucz", "followers_url": "https://api.github.com/users/JuanFKurucz/followers", "following_url": "https://api.github.com/users/JuanFKurucz/following{/other_user}", "gists_url": "https://api.github.com/users/JuanFKurucz/gists{/gist_id}", "starred_url": "https://api.github.com/users/JuanFKurucz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JuanFKurucz/subscriptions", "organizations_url": "https://api.github.com/users/JuanFKurucz/orgs", "repos_url": "https://api.github.com/users/JuanFKurucz/repos", "events_url": "https://api.github.com/users/JuanFKurucz/events{/privacy}", "received_events_url": "https://api.github.com/users/JuanFKurucz/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28437/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28437", "html_url": "https://github.com/huggingface/transformers/pull/28437", "diff_url": "https://github.com/huggingface/transformers/pull/28437.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28437.patch", "merged_at": 1704906547000 }

{ "login": "JustinLin610", "id": 27664428, "node_id": "MDQ6VXNlcjI3NjY0NDI4", "avatar_url": "https://avatars.githubusercontent.com/u/27664428?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JustinLin610", "html_url": "https://github.com/JustinLin610", "followers_url": "https://api.github.com/users/JustinLin610/followers", "following_url": "https://api.github.com/users/JustinLin610/following{/other_user}", "gists_url": "https://api.github.com/users/JustinLin610/gists{/gist_id}", "starred_url": "https://api.github.com/users/JustinLin610/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JustinLin610/subscriptions", "organizations_url": "https://api.github.com/users/JustinLin610/orgs", "repos_url": "https://api.github.com/users/JustinLin610/repos", "events_url": "https://api.github.com/users/JustinLin610/events{/privacy}", "received_events_url": "https://api.github.com/users/JustinLin610/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Feel free to rebase on main as well if some CIs are unrelated to your PR, and ping me whenever for a final review! 🤗 ", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28436). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "Thanks a lot for this pr and bearing with me! 🤗", "Qwen1.5-0.5B-Chat lose output.weight after load from_pretrained and save_pretrained\r\n\r\n\r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28436/reactions", "total_count": 5, "+1": 5, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28436", "html_url": "https://github.com/huggingface/transformers/pull/28436", "diff_url": "https://github.com/huggingface/transformers/pull/28436.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28436.patch", "merged_at": 1705503742000 }

{ "login": "gregor-ge", "id": 7710563, "node_id": "MDQ6VXNlcjc3MTA1NjM=", "avatar_url": "https://avatars.githubusercontent.com/u/7710563?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gregor-ge", "html_url": "https://github.com/gregor-ge", "followers_url": "https://api.github.com/users/gregor-ge/followers", "following_url": "https://api.github.com/users/gregor-ge/following{/other_user}", "gists_url": "https://api.github.com/users/gregor-ge/gists{/gist_id}", "starred_url": "https://api.github.com/users/gregor-ge/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gregor-ge/subscriptions", "organizations_url": "https://api.github.com/users/gregor-ge/orgs", "repos_url": "https://api.github.com/users/gregor-ge/repos", "events_url": "https://api.github.com/users/gregor-ge/events{/privacy}", "received_events_url": "https://api.github.com/users/gregor-ge/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Is this what you are looking for:\r\nhttps://github.com/huggingface/transformers/blob/a3b00030c96be4c66e818cf9a22040dbe41006d8/src/transformers/modeling_utils.py#L1165\r\n", "Not quite, no, but it is similar. `_keep_in_fp32_modules ` is set on the library side (e.g. for the T5 modules) and it also is limited to fp32. I am looking for a solution similar to the initial suggestion in #20683, that is for example `Blip2ForConditionalGeneration.from_pretrained(..., load_in_8bit=True, skip_for_8bit_modules=[\"qformer\", \"vision_encoder\", \"language_projection\"])` which loads those modules in the checkpoint format (fp16, fp32, bf16, ...) without quantization.\r\n\r\nI think the following could work as a (dirty) way of implementing what I want with it though but I have to check later:\r\n```python\r\nfrom transformers import Blip2ForConditionalGeneration\r\n# add modules to list so they are skipped for quantization\r\nBlip2ForConditionalGeneration._keep_in_fp32_modules = [\"qformer\", \"vision_encoder\", \"language_projection\"]\r\nmodel = Blip2ForConditionalGeneration.from_pretrained(..., load_in_8bit=True)\r\n\r\n# manually cast back to fp16 if wanted\r\nBlip2ForConditionalGeneration._keep_in_fp32_modules = []\r\nmodel.qformer = model.qformer.half()\r\nmodel.vision_encoder= model.vision_encoder.half()\r\n...\r\n```", "I dug around in `load_pretrained()` and realized there *is* a config option `quantization_config.llm_int8_skip_modules` [here](https://github.com/huggingface/transformers/blob/a3b00030c96be4c66e818cf9a22040dbe41006d8/src/transformers/modeling_utils.py#L3510) which seems to do exactly what I want. I will test this out on the weekend and report back. One thing I noticed is that when you set this parameter, you have to also manually add parameters like the `lm_head`which are skipped on default otherwise.", "Yes, it works just as I want. It also work for int4. The parameter is documented [here](https://huggingface.co./docs/transformers/main_classes/quantization#transformers.BitsAndBytesConfig.llm_int8_skip_modules) which I seem to have missed in my search before.\r\n\r\nFor posterity, the following snippet is how to use it:\r\n```python\r\nfrom transformers import Blip2ForConditionalGeneration\r\nskip_int8_modules = [\"lm_head\", \"vision_encoder\", \"language_projection\", \"qformer\"] if skip_vision_int8 else None\r\nself.model = Blip2ForConditionalGeneration.from_pretrained(\r\n checkpoint,\r\n load_in_8bit=True,\r\n llm_int8_skip_modules=skip_int8_modules\r\n)\r\n´´´\r\n\r\nThanks for helping me!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28435/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "JeevanBhoot", "id": 64039772, "node_id": "MDQ6VXNlcjY0MDM5Nzcy", "avatar_url": "https://avatars.githubusercontent.com/u/64039772?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JeevanBhoot", "html_url": "https://github.com/JeevanBhoot", "followers_url": "https://api.github.com/users/JeevanBhoot/followers", "following_url": "https://api.github.com/users/JeevanBhoot/following{/other_user}", "gists_url": "https://api.github.com/users/JeevanBhoot/gists{/gist_id}", "starred_url": "https://api.github.com/users/JeevanBhoot/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JeevanBhoot/subscriptions", "organizations_url": "https://api.github.com/users/JeevanBhoot/orgs", "repos_url": "https://api.github.com/users/JeevanBhoot/repos", "events_url": "https://api.github.com/users/JeevanBhoot/events{/privacy}", "received_events_url": "https://api.github.com/users/JeevanBhoot/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @JeevanBhoot \r\nThis is not supported in PyTorch, when passing input_ids they need to stay in `torch.int64`, the only way to run inference in bf16 is the following: \r\n```diff\r\nmodel_path = \"meta-llama/Llama-2-7b-chat-hf\"\r\nmodel = AutoModelForCausalLM.from_pretrained(model_path, device_map=\"auto\", trust_remote_code=False, revision=\"main\", torch_dtype=torch.bfloat16).to(\"cuda\")\r\ntokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True)\r\n\r\n- input_ids = tokenizer(\"Tell me an interesting fact!\", return_tensors=\"pt\").input_ids.to(\"cuda\").to(torch.bfloat16)\r\n+ input_ids = tokenizer(\"Tell me an interesting fact!\", return_tensors=\"pt\").input_ids.to(\"cuda\")\r\noutput = model.generate(input_ids)\r\n```", "@younesbelkada When input_ids in `torch.int64`, the output from the model is also `torch.int64`. Is there no way to have the model activations and outputs be `torch.bfloat16`?", "@JeevanBhoot if you pass `torch.LongTensor` to a `nn.Embedding` layer the output should be a floating point. If you load your model in bf16 it should be `int64` -> `bfloat16`, you can inspect that by doing `model.get_input_embeddings()(input_ids).dtype`", "@younesbelkada Ok, I can confirm that `model.get_input_embeddings()(input_ids).dtype` returns `bfloat16` but `output.dtype` is `int64`. \r\n\r\nIs the final output of the model cast from `bfloat16` to `int64`, or is the dtype changing at some point in the model?", "@JeevanBhoot `output` will always be in `int64` since it corresponds to the predicted tokens. There is no need to cast the output indices in `bfloat16`, what is your motivation behind getting `output` in bf16?", "@younesbelkada I don't need the final output to be in bf16, but I want to confirm that EVERY operation and output within the model is bf16.", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28434/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "amyeroberts", "id": 22614925, "node_id": "MDQ6VXNlcjIyNjE0OTI1", "avatar_url": "https://avatars.githubusercontent.com/u/22614925?v=4", "gravatar_id": "", "url": "https://api.github.com/users/amyeroberts", "html_url": "https://github.com/amyeroberts", "followers_url": "https://api.github.com/users/amyeroberts/followers", "following_url": "https://api.github.com/users/amyeroberts/following{/other_user}", "gists_url": "https://api.github.com/users/amyeroberts/gists{/gist_id}", "starred_url": "https://api.github.com/users/amyeroberts/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/amyeroberts/subscriptions", "organizations_url": "https://api.github.com/users/amyeroberts/orgs", "repos_url": "https://api.github.com/users/amyeroberts/repos", "events_url": "https://api.github.com/users/amyeroberts/events{/privacy}", "received_events_url": "https://api.github.com/users/amyeroberts/received_events", "type": "User", "site_admin": false }

[]

[]

[ "@Narsil - I've added applying sigmoid function to the model outputs for image classification matching the text classification pipeline, per your [suggestion here](https://github.com/huggingface/huggingface.js/issues/429#issuecomment-1880625080). \r\n\r\nAgree controlling through a config would be a good idea -- it is partially controlled through `model.config.problem_type` here. Is this enough or would you like the activation function to be more explicitly defined in the config? " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28433/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28433", "html_url": "https://github.com/huggingface/transformers/pull/28433", "diff_url": "https://github.com/huggingface/transformers/pull/28433.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28433.patch", "merged_at": 1704968979000 }

{ "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", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28432/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28432", "html_url": "https://github.com/huggingface/transformers/pull/28432", "diff_url": "https://github.com/huggingface/transformers/pull/28432.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28432.patch", "merged_at": 1704890345000 }

{ "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", "site_admin": false }

[]

[]

[ "@stevhliu please help review the doc", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28431). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "Hi @stevhliu . Thanks for your advice, I have fixed these comments. Please take a look, thx!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28431/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28431", "html_url": "https://github.com/huggingface/transformers/pull/28431", "diff_url": "https://github.com/huggingface/transformers/pull/28431.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28431.patch", "merged_at": 1704992148000 }

{ "login": "bayllama", "id": 142558246, "node_id": "U_kgDOCH9EJg", "avatar_url": "https://avatars.githubusercontent.com/u/142558246?v=4", "gravatar_id": "", "url": "https://api.github.com/users/bayllama", "html_url": "https://github.com/bayllama", "followers_url": "https://api.github.com/users/bayllama/followers", "following_url": "https://api.github.com/users/bayllama/following{/other_user}", "gists_url": "https://api.github.com/users/bayllama/gists{/gist_id}", "starred_url": "https://api.github.com/users/bayllama/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/bayllama/subscriptions", "organizations_url": "https://api.github.com/users/bayllama/orgs", "repos_url": "https://api.github.com/users/bayllama/repos", "events_url": "https://api.github.com/users/bayllama/events{/privacy}", "received_events_url": "https://api.github.com/users/bayllama/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28430/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28430", "html_url": "https://github.com/huggingface/transformers/pull/28430", "diff_url": "https://github.com/huggingface/transformers/pull/28430.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28430.patch", "merged_at": 1704885069000 }

{ "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", "site_admin": false }

[]

[]

[ "Hi, I don't think this change should be done. jit traced model still uses SDPA that may dispatch to memory-efficient backend, which requires to have no fully masked rows (see https://github.com/pytorch/pytorch/issues/110213), which is what `_unmask_unattended` avoids. I'd recommend to trace models with `query_length > 1`.\r\n\r\n@jiqing-feng Could you clarify the motivation and what is not working as intended?", "> Hi, I don't think this change should be done. jit traced model still uses SDPA that may dispatch to memory-efficient backend, which requires to have no fully masked rows (see [pytorch/pytorch#110213](https://github.com/pytorch/pytorch/issues/110213)), which is what `_unmask_unattended` avoids. I'd recommend to trace models with `query_length > 1`.\r\n> \r\n> @jiqing-feng Could you clarify the motivation and what is not working as intended?\r\n\r\nSorry for that I didn't make it clear. Yes, we trace the model with `query_length > 1`, but when we use the trace model to do generation tasks, the inputs do not contain `past_key_values` when inference the 1st token. We used a tricky way that input a `past_key_values` with 0 length, see [here](https://github.com/huggingface/optimum-intel/blob/main/optimum/intel/generation/modeling.py#L311-L313). In this way, we could avoid input `None` to the traced model (None input will cause traced model error), but it does not satisfy `query_length > 0` so the traced model forward failed in the `_unmask_unattended`. \r\n\r\nIn conclusion, [TSModelForCausalLM](https://github.com/huggingface/optimum-intel/blob/main/optimum/intel/generation/modeling.py#L331) in `optimum-intel` will fail if I don't make this change. \r\n \r\nDo you have any good ideas to avoid it? Thx!\r\n\r\nPlease ask me if there is anything I didn't make clear.\r\n", "@jiqing-feng `query_length` is taken from `input_ids` (https://github.com/huggingface/transformers/blob/ee2482b6f8eca320f1eff112673adae215dbc927/src/transformers/models/llama/modeling_llama.py#L1037), not from the sequence length in the KV cache so I am not sure to understand, how is PKV impacting the controlflow `if query_length > 1:`? Feel free to share a repro of the unexpected behavior/bug if that can help.", "Though I agree that jit.trace / dynamo / symbolic_trace are likely not able to properly trace https://github.com/huggingface/transformers/blob/6c78bbcb8320d316434262ef003251ca997db0d1/src/transformers/modeling_attn_mask_utils.py#L244-L245", "Hi @fxmarty Thanks for your advice, I get your point now. The problem should be fixed in `optimum-intel` instead of transformers because it is related to our traced model inputs.", "@jiqing-feng I actually apply the same fix as you in https://github.com/huggingface/transformers/pull/28447 as `_unmask_unattended` can not be properly traced with torch.jit.trace or symbolic_trace. As this method was only implemented as a workaround to a limitation in pytorch with a specific backend on a specific hardware, I think it is fine.\r\n\r\nWe may want to have a tracable `_unmask_unattended`." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28429/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28429", "html_url": "https://github.com/huggingface/transformers/pull/28429", "diff_url": "https://github.com/huggingface/transformers/pull/28429.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28429.patch", "merged_at": null }

{ "login": "andysingal", "id": 20493493, "node_id": "MDQ6VXNlcjIwNDkzNDkz", "avatar_url": "https://avatars.githubusercontent.com/u/20493493?v=4", "gravatar_id": "", "url": "https://api.github.com/users/andysingal", "html_url": "https://github.com/andysingal", "followers_url": "https://api.github.com/users/andysingal/followers", "following_url": "https://api.github.com/users/andysingal/following{/other_user}", "gists_url": "https://api.github.com/users/andysingal/gists{/gist_id}", "starred_url": "https://api.github.com/users/andysingal/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/andysingal/subscriptions", "organizations_url": "https://api.github.com/users/andysingal/orgs", "repos_url": "https://api.github.com/users/andysingal/repos", "events_url": "https://api.github.com/users/andysingal/events{/privacy}", "received_events_url": "https://api.github.com/users/andysingal/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey, this is not related to `transformers` but the inference API ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28428/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "MohammadDara", "id": 6161219, "node_id": "MDQ6VXNlcjYxNjEyMTk=", "avatar_url": "https://avatars.githubusercontent.com/u/6161219?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MohammadDara", "html_url": "https://github.com/MohammadDara", "followers_url": "https://api.github.com/users/MohammadDara/followers", "following_url": "https://api.github.com/users/MohammadDara/following{/other_user}", "gists_url": "https://api.github.com/users/MohammadDara/gists{/gist_id}", "starred_url": "https://api.github.com/users/MohammadDara/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/MohammadDara/subscriptions", "organizations_url": "https://api.github.com/users/MohammadDara/orgs", "repos_url": "https://api.github.com/users/MohammadDara/repos", "events_url": "https://api.github.com/users/MohammadDara/events{/privacy}", "received_events_url": "https://api.github.com/users/MohammadDara/received_events", "type": "User", "site_admin": false }

[]

[]

[ "cc @ydshieh if you can have a look 🤗 ", "Hi.\r\n\r\nWell , when I tried it on Colab\r\n\r\n> Downloading data: 80% 3.75G/4.69G [01:16<00:22, 41.7MB/s]\r\n\r\nSo it's big `4.69 G` ... On colab, it's fast. In your case, it depends on your network :-)", "Sorry, your are right, something is wrong! I will check", "Well, it is 51 files of 1.33 GB to download, and after that 20M examples to generate and takes 14 hours ....\r\n\r\n> Generating train split 1647286/21015300 [1:11:27<13:45:48, 390.89 examples/s\r\n\r\n\r\n", "So it will definitely finish (in theory). There is nothing we can do but just keep patient 😭 ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28427/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Adonai02", "id": 70610799, "node_id": "MDQ6VXNlcjcwNjEwNzk5", "avatar_url": "https://avatars.githubusercontent.com/u/70610799?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Adonai02", "html_url": "https://github.com/Adonai02", "followers_url": "https://api.github.com/users/Adonai02/followers", "following_url": "https://api.github.com/users/Adonai02/following{/other_user}", "gists_url": "https://api.github.com/users/Adonai02/gists{/gist_id}", "starred_url": "https://api.github.com/users/Adonai02/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Adonai02/subscriptions", "organizations_url": "https://api.github.com/users/Adonai02/orgs", "repos_url": "https://api.github.com/users/Adonai02/repos", "events_url": "https://api.github.com/users/Adonai02/events{/privacy}", "received_events_url": "https://api.github.com/users/Adonai02/received_events", "type": "User", "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" } ]

[]

[ "That would be nice but a bit outside the scope of transformers! Would be nice if you have a working example! \r\nWhat I recommend is to register a load_state_dict hook that converts the checkpoints on the fly. \r\nThe benchmark should run on different num kv heads as some shapes might be less optimal ? That would be my intuitiion. Also a single head (MQA) should be always faster than MHA", "If I understand correctly, here is displayed an attempt to implement GQA on a non GQA Llama 2 13b model?\nIf that's the case, and despite the slight loss of performance observed, does the context size in VRAM gets diminished as GQA allows, and is the perplexity of the model affected?\nIf that's not the case, sorry for misunderstanding!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28425/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "hi-sushanta", "id": 93595990, "node_id": "U_kgDOBZQpVg", "avatar_url": "https://avatars.githubusercontent.com/u/93595990?v=4", "gravatar_id": "", "url": "https://api.github.com/users/hi-sushanta", "html_url": "https://github.com/hi-sushanta", "followers_url": "https://api.github.com/users/hi-sushanta/followers", "following_url": "https://api.github.com/users/hi-sushanta/following{/other_user}", "gists_url": "https://api.github.com/users/hi-sushanta/gists{/gist_id}", "starred_url": "https://api.github.com/users/hi-sushanta/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/hi-sushanta/subscriptions", "organizations_url": "https://api.github.com/users/hi-sushanta/orgs", "repos_url": "https://api.github.com/users/hi-sushanta/repos", "events_url": "https://api.github.com/users/hi-sushanta/events{/privacy}", "received_events_url": "https://api.github.com/users/hi-sushanta/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Would you be able to check my solution?", "I don't understand why this error occurs, but my code does the same work to define the issue section.", "@hi-sushanta To figure out the errors, you'll need to look at the CI runs and debug locally so that the currently failing tests pass. For the quality checks, you'll need to fun `make fix-copies` and `make fixup`", "But this command returns to the original codebase. that does not make sense because my code works exactly like the issues section mentioned." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28424/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28424", "html_url": "https://github.com/huggingface/transformers/pull/28424", "diff_url": "https://github.com/huggingface/transformers/pull/28424.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28424.patch", "merged_at": null }

{ "login": "JuanFKurucz", "id": 31422367, "node_id": "MDQ6VXNlcjMxNDIyMzY3", "avatar_url": "https://avatars.githubusercontent.com/u/31422367?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JuanFKurucz", "html_url": "https://github.com/JuanFKurucz", "followers_url": "https://api.github.com/users/JuanFKurucz/followers", "following_url": "https://api.github.com/users/JuanFKurucz/following{/other_user}", "gists_url": "https://api.github.com/users/JuanFKurucz/gists{/gist_id}", "starred_url": "https://api.github.com/users/JuanFKurucz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JuanFKurucz/subscriptions", "organizations_url": "https://api.github.com/users/JuanFKurucz/orgs", "repos_url": "https://api.github.com/users/JuanFKurucz/repos", "events_url": "https://api.github.com/users/JuanFKurucz/events{/privacy}", "received_events_url": "https://api.github.com/users/JuanFKurucz/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28423/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28423", "html_url": "https://github.com/huggingface/transformers/pull/28423", "diff_url": "https://github.com/huggingface/transformers/pull/28423.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28423.patch", "merged_at": 1705306163000 }

{ "login": "aphedges", "id": 14283972, "node_id": "MDQ6VXNlcjE0MjgzOTcy", "avatar_url": "https://avatars.githubusercontent.com/u/14283972?v=4", "gravatar_id": "", "url": "https://api.github.com/users/aphedges", "html_url": "https://github.com/aphedges", "followers_url": "https://api.github.com/users/aphedges/followers", "following_url": "https://api.github.com/users/aphedges/following{/other_user}", "gists_url": "https://api.github.com/users/aphedges/gists{/gist_id}", "starred_url": "https://api.github.com/users/aphedges/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/aphedges/subscriptions", "organizations_url": "https://api.github.com/users/aphedges/orgs", "repos_url": "https://api.github.com/users/aphedges/repos", "events_url": "https://api.github.com/users/aphedges/events{/privacy}", "received_events_url": "https://api.github.com/users/aphedges/received_events", "type": "User", "site_admin": false }

[]

[]

[ "I would appreciate if you could make sure to preserve my commit message (with any needed changes) when merging so it ends up in the Git log." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28422/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28422", "html_url": "https://github.com/huggingface/transformers/pull/28422", "diff_url": "https://github.com/huggingface/transformers/pull/28422.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28422.patch", "merged_at": 1704983924000 }

{ "login": "muellerzr", "id": 7831895, "node_id": "MDQ6VXNlcjc4MzE4OTU=", "avatar_url": "https://avatars.githubusercontent.com/u/7831895?v=4", "gravatar_id": "", "url": "https://api.github.com/users/muellerzr", "html_url": "https://github.com/muellerzr", "followers_url": "https://api.github.com/users/muellerzr/followers", "following_url": "https://api.github.com/users/muellerzr/following{/other_user}", "gists_url": "https://api.github.com/users/muellerzr/gists{/gist_id}", "starred_url": "https://api.github.com/users/muellerzr/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/muellerzr/subscriptions", "organizations_url": "https://api.github.com/users/muellerzr/orgs", "repos_url": "https://api.github.com/users/muellerzr/repos", "events_url": "https://api.github.com/users/muellerzr/events{/privacy}", "received_events_url": "https://api.github.com/users/muellerzr/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28421/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28421", "html_url": "https://github.com/huggingface/transformers/pull/28421", "diff_url": "https://github.com/huggingface/transformers/pull/28421.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28421.patch", "merged_at": 1704884551000 }

{ "login": "harisankar95", "id": 58052269, "node_id": "MDQ6VXNlcjU4MDUyMjY5", "avatar_url": "https://avatars.githubusercontent.com/u/58052269?v=4", "gravatar_id": "", "url": "https://api.github.com/users/harisankar95", "html_url": "https://github.com/harisankar95", "followers_url": "https://api.github.com/users/harisankar95/followers", "following_url": "https://api.github.com/users/harisankar95/following{/other_user}", "gists_url": "https://api.github.com/users/harisankar95/gists{/gist_id}", "starred_url": "https://api.github.com/users/harisankar95/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/harisankar95/subscriptions", "organizations_url": "https://api.github.com/users/harisankar95/orgs", "repos_url": "https://api.github.com/users/harisankar95/repos", "events_url": "https://api.github.com/users/harisankar95/events{/privacy}", "received_events_url": "https://api.github.com/users/harisankar95/received_events", "type": "User", "site_admin": false }

[]

[]

[ "@amyeroberts Can you please review the PR? MobileViT image preprocessor is updated to accept segmentation maps inline to other preprocessors for segmentation models like that of Segformer. ", "I have corrected the error in the tests_torch. I will wait for the https://github.com/huggingface/transformers/pull/28432 to be merged to fix the remaining CI tests", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28420). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28420/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28420", "html_url": "https://github.com/huggingface/transformers/pull/28420", "diff_url": "https://github.com/huggingface/transformers/pull/28420.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28420.patch", "merged_at": 1704984734000 }

{ "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", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28419). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "I tried, but I couldn't figure out a way to test it, and I don't think the previous test authors could either! The test would need to check that `input()` is called, but we can't really check that - we can only check for the error that's raised when it times out. This is what happens in [the existing tests](https://github.com/huggingface/transformers/blob/main/tests/models/auto/test_tokenization_auto.py#L306), and that's why the tests didn't catch the problem - the error is raised without ever calling `input()`, and the test only checks for the error, so it passed." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28419/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28419", "html_url": "https://github.com/huggingface/transformers/pull/28419", "diff_url": "https://github.com/huggingface/transformers/pull/28419.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28419.patch", "merged_at": 1704985929000 }

{ "login": "NoB0", "id": 28621493, "node_id": "MDQ6VXNlcjI4NjIxNDkz", "avatar_url": "https://avatars.githubusercontent.com/u/28621493?v=4", "gravatar_id": "", "url": "https://api.github.com/users/NoB0", "html_url": "https://github.com/NoB0", "followers_url": "https://api.github.com/users/NoB0/followers", "following_url": "https://api.github.com/users/NoB0/following{/other_user}", "gists_url": "https://api.github.com/users/NoB0/gists{/gist_id}", "starred_url": "https://api.github.com/users/NoB0/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/NoB0/subscriptions", "organizations_url": "https://api.github.com/users/NoB0/orgs", "repos_url": "https://api.github.com/users/NoB0/repos", "events_url": "https://api.github.com/users/NoB0/events{/privacy}", "received_events_url": "https://api.github.com/users/NoB0/received_events", "type": "User", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28418). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28418/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28418", "html_url": "https://github.com/huggingface/transformers/pull/28418", "diff_url": "https://github.com/huggingface/transformers/pull/28418.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28418.patch", "merged_at": null }

{ "login": "dependabot[bot]", "id": 49699333, "node_id": "MDM6Qm90NDk2OTkzMzM=", "avatar_url": "https://avatars.githubusercontent.com/in/29110?v=4", "gravatar_id": "", "url": "https://api.github.com/users/dependabot%5Bbot%5D", "html_url": "https://github.com/apps/dependabot", "followers_url": "https://api.github.com/users/dependabot%5Bbot%5D/followers", "following_url": "https://api.github.com/users/dependabot%5Bbot%5D/following{/other_user}", "gists_url": "https://api.github.com/users/dependabot%5Bbot%5D/gists{/gist_id}", "starred_url": "https://api.github.com/users/dependabot%5Bbot%5D/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/dependabot%5Bbot%5D/subscriptions", "organizations_url": "https://api.github.com/users/dependabot%5Bbot%5D/orgs", "repos_url": "https://api.github.com/users/dependabot%5Bbot%5D/repos", "events_url": "https://api.github.com/users/dependabot%5Bbot%5D/events{/privacy}", "received_events_url": "https://api.github.com/users/dependabot%5Bbot%5D/received_events", "type": "Bot", "site_admin": false }

[ { "id": 1905493434, "node_id": "MDU6TGFiZWwxOTA1NDkzNDM0", "url": "https://api.github.com/repos/huggingface/transformers/labels/dependencies", "name": "dependencies", "color": "0366d6", "default": false, "description": "Pull requests that update a dependency file" }, { "id": 6410654816, "node_id": "LA_kwDOCUB6oc8AAAABfhrUYA", "url": "https://api.github.com/repos/huggingface/transformers/labels/python", "name": "python", "color": "2b67c6", "default": false, "description": "Pull requests that update Python code" } ]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28417/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28417", "html_url": "https://github.com/huggingface/transformers/pull/28417", "diff_url": "https://github.com/huggingface/transformers/pull/28417.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28417.patch", "merged_at": 1704882164000 }

{ "login": "gabeorlanski", "id": 18234433, "node_id": "MDQ6VXNlcjE4MjM0NDMz", "avatar_url": "https://avatars.githubusercontent.com/u/18234433?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gabeorlanski", "html_url": "https://github.com/gabeorlanski", "followers_url": "https://api.github.com/users/gabeorlanski/followers", "following_url": "https://api.github.com/users/gabeorlanski/following{/other_user}", "gists_url": "https://api.github.com/users/gabeorlanski/gists{/gist_id}", "starred_url": "https://api.github.com/users/gabeorlanski/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gabeorlanski/subscriptions", "organizations_url": "https://api.github.com/users/gabeorlanski/orgs", "repos_url": "https://api.github.com/users/gabeorlanski/repos", "events_url": "https://api.github.com/users/gabeorlanski/events{/privacy}", "received_events_url": "https://api.github.com/users/gabeorlanski/received_events", "type": "User", "site_admin": false }

[]

[]

[ "From the [model card](https://huggingface.co./microsoft/phi-2):\r\n> If you are using transformers>=4.36.0, always load the model with trust_remote_code=True to prevent side-effects.\r\n\r\nYou'll find a few examples on how to load the model there. The following should work:\r\n```python\r\nfrom transformers import AutoModelForCausalLM\r\nmodel = AutoModelForCausalLM.from_pretrained(\"microsoft/phi-2\", trust_remote_code=True)\r\n```", "I tried that and also tried deleting the downloaded file then running the code, issue still persists only with Phi1.5. ", "Hey, that is expected, `model = AutoModelForCausalLM.from_pretrained(\"microsoft/phi-2\", use_safetensors= True, trust_remote_code =True)` will allow you to do that, but if you want to use the converted checkpoints you should use the ones share by @susnato here hf.co/susnato/phi-2 \r\n\r\nwill be fixed by #28392 ", "So with Phi-2 that does work. But with Phi 1.5 the model hangs and never loads. I tried deleting the downloaded model from before, and it only downloads the config this time but, again, does not load.\r\n\r\nI updated the issue accordingly to specify this is only happening with Phi 1.5.", "What checkpoint are you using for phi 1.5? @gabeorlanski ", "I am using [microsoft/phi-1_5](https://huggingface.co./microsoft/phi-1_5)\r\n\r\nTo update, it _does_ end up finishing, but gave me this error:\r\n```shell\r\nTraceback (most recent call last):\r\n File \"/u/g/o/gorlanski/miniconda3/envs/syncode/lib/python3.10/site-packages/huggingface_hub/utils/_errors.py\", line 286, in hf_raise_for_status\r\n response.raise_for_status()\r\n File \"/u/g/o/gorlanski/miniconda3/envs/syncode/lib/python3.10/site-packages/requests/models.py\", line 1021, in raise_for_status\r\n raise HTTPError(http_error_msg, response=self)\r\nrequests.exceptions.HTTPError: 502 Server Error: Bad Gateway for url: https://huggingface.co./api/models/microsoft/phi-1_5/discussions?page_index=11923\r\n```\r\n\r\nI set logging to debug level, and it appears to be going through all of the discussion pages:\r\n```shell\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] Starting new HTTPS connection (2): huggingface.co:443\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"HEAD /microsoft/phi-1_5/resolve/main/config.json HTTP/1.1\" 200 0\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5 HTTP/1.1\" 200 3080\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5/commits/main HTTP/1.1\" 200 16853\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5/commits/main?p=1 HTTP/1.1\" 200 1602\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5/discussions?page_index=0 HTTP/1.1\" 200 20155\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5/discussions?page_index=1 HTTP/1.1\" 200 20155\r\n[2024-01-10 10:00:25][urllib3.connectionpool][DEBUG] https://huggingface.co.:443 \"GET /api/models/microsoft/phi-1_5/discussions?page_index=2 HTTP/1.1\" 200 20155\r\n```\r\n\r\n", "Can you please try to load from - `susnato/phi-1_5_dev` ? @gabeorlanski ", "That one seems to work fine. It does still get the discussions (I assume that is intended)\r\n\r\nWhen looking at the `microsoft/phi-1_5` there appears to have been some recent updates, could that be the cause?", "Yes @gabeorlanski, the recent updates should make the repo compatible with the library code...tagging @gugarosa to have a look at the [error message](https://github.com/huggingface/transformers/issues/28416#issuecomment-1885140212). \r\n\r\nUntil it is fixed you can use phi 1.5 from `susnato/phi-1_5_dev`", "The download might be probably hanging because it is trying to find safetensors-based files, whereas Phi-1.5 was not saved with safetensors.\r\n\r\nRegarding the loading issues, using:\r\n```\r\nfrom transformers import AutoModelForCausalLM\r\nmodel = AutoModelForCausalLM.from_pretrained(\"microsoft/phi-1_5\", trust_remote_code=True)\r\n```\r\n\r\nShould be working now regardless of you transformers version.", "@gugarosa that fixed it. Thanks so much! Closing the issue" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28416/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "zhangmiaosen2000", "id": 59921236, "node_id": "MDQ6VXNlcjU5OTIxMjM2", "avatar_url": "https://avatars.githubusercontent.com/u/59921236?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zhangmiaosen2000", "html_url": "https://github.com/zhangmiaosen2000", "followers_url": "https://api.github.com/users/zhangmiaosen2000/followers", "following_url": "https://api.github.com/users/zhangmiaosen2000/following{/other_user}", "gists_url": "https://api.github.com/users/zhangmiaosen2000/gists{/gist_id}", "starred_url": "https://api.github.com/users/zhangmiaosen2000/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zhangmiaosen2000/subscriptions", "organizations_url": "https://api.github.com/users/zhangmiaosen2000/orgs", "repos_url": "https://api.github.com/users/zhangmiaosen2000/repos", "events_url": "https://api.github.com/users/zhangmiaosen2000/events{/privacy}", "received_events_url": "https://api.github.com/users/zhangmiaosen2000/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Running into the exact same issue after fine-tuning with embedding reshaped ", "> But when I try to use this code to load finetuned model:\r\n\r\nI assume that the error appears when you try to merge the adapter and the base model. The problem is that phi-2 expects the vocab_size to be a multiple of 64 (see [configuration_phi.py](https://huggingface.co./microsoft/phi-2/blob/834565c23f9b28b96ccbeabe614dd906b6db551a/configuration_phi.py#L44)). \r\n\r\nWhenever you add new tokens, you have to pad the embedding dimension to a multiple of 64. This has to be done after adding tokens when training **and** before merging the trained adapter. Example:\r\n\r\n```python\r\ntokenizer.add_tokens([\"<|im_start|>\", \"<PAD>\"])\r\ntokenizer.pad_token = \"<PAD>\"\r\ntokenizer.add_special_tokens(dict(eos_token=\"<|im_end|>\"))\r\nmodel.resize_token_embeddings(\r\n new_num_tokens=len(tokenizer),\r\n pad_to_multiple_of=64) # phi2 default is 64, see configuration_phi.py\r\n```\r\n\r\n[Notebook](https://github.com/geronimi73/phi2-finetune/blob/main/nb_qlora.ipynb) with a complete example for a QLoRA finetune of phi2.", "Thanks @geronimi73 for the answer 🤗 ", "> > But when I try to use this code to load finetuned model:\r\n> Whenever you add new tokens, you have to pad the embedding dimension to a multiple of 64. This has to be done after adding tokens when training **and** before merging the trained adapter. Example:\r\n\r\nsorry, forget everything I wrote please. It works but it's not necessary. The model's embedding layer is big enough to accomodate additional tokens, see this: https://huggingface.co./microsoft/phi-2/discussions/22#659d8ba950c1bbee5be6f179\r\n\r\nin other words, when you add tokens to the tokenizer, no need to resize the embeddings. in code:\r\n\r\n```python\r\nif tokenizer.pad_token is None:\r\n tokenizer.add_special_tokens(dict(pad_token=DEFAULT_PAD_TOKEN))\r\n # not needed:\r\n # smart_tokenizer_and_embedding_resize(\r\n # special_tokens_dict=dict(pad_token=DEFAULT_PAD_TOKEN),\r\n # tokenizer=tokenizer,\r\n # model=model,\r\n #)\r\n\r\n```", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28415/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "fxmarty", "id": 9808326, "node_id": "MDQ6VXNlcjk4MDgzMjY=", "avatar_url": "https://avatars.githubusercontent.com/u/9808326?v=4", "gravatar_id": "", "url": "https://api.github.com/users/fxmarty", "html_url": "https://github.com/fxmarty", "followers_url": "https://api.github.com/users/fxmarty/followers", "following_url": "https://api.github.com/users/fxmarty/following{/other_user}", "gists_url": "https://api.github.com/users/fxmarty/gists{/gist_id}", "starred_url": "https://api.github.com/users/fxmarty/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/fxmarty/subscriptions", "organizations_url": "https://api.github.com/users/fxmarty/orgs", "repos_url": "https://api.github.com/users/fxmarty/repos", "events_url": "https://api.github.com/users/fxmarty/events{/privacy}", "received_events_url": "https://api.github.com/users/fxmarty/received_events", "type": "User", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28414). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "@fxmarty Could you add a test that passes with these changes? ", "@amyeroberts sure, where would be a good place for it?", "@fxmarty I'd suggest putting a test in `modeling_utils` which does effectively the same thing as in the example in PR description with a small model. Happy for you to put it somewhere else if you think it's better ", "There seem to be unrelated issues in the CI\r\n\r\n```\r\nBuilding wheels for collected packages: natten\r\n Building wheel for natten (setup.py) ... error\r\n error: subprocess-exited-with-error\r\n \r\n × python setup.py bdist_wheel did not run successfully.\r\n │ exit code: 1\r\n ╰─> [80 lines of output]\r\n Building NATTEN for CPU ONLY.\r\n Number of workers: 9\r\n running bdist_wheel\r\n running build\r\n running build_py\r\n creating build\r\n creating build/lib.linux-x86_64-cpython-38\r\n creating build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/__init__.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/flops.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/functional.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/natten1d.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/natten2d.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/natten3d.py -> build/lib.linux-x86_64-cpython-38/natten\r\n copying src/natten/nested.py -> build/lib.linux-x86_64-cpython-38/natten\r\n creating build/lib.linux-x86_64-cpython-38/natten/utils\r\n copying src/natten/utils/__init__.py -> build/lib.linux-x86_64-cpython-38/natten/utils\r\n copying src/natten/utils/tensor.py -> build/lib.linux-x86_64-cpython-38/natten/utils\r\n copying src/natten/utils/testing.py -> build/lib.linux-x86_64-cpython-38/natten/utils\r\n copying src/natten/utils/typing.py -> build/lib.linux-x86_64-cpython-38/natten/utils\r\n running build_ext\r\n Traceback (most recent call last):\r\n File \"/tmp/pip-install-musj9gr5/natten_ea055ed283804ee0b0ef252dd4ad0d6c/setup.py\", line 147, in build_extension\r\n subprocess.check_output([\"cmake\", \"--version\"])\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/subprocess.py\", line 415, in check_output\r\n return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/subprocess.py\", line 493, in run\r\n with Popen(*popenargs, **kwargs) as process:\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/subprocess.py\", line 858, in __init__\r\n self._execute_child(args, executable, preexec_fn, close_fds,\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/subprocess.py\", line 1704, in _execute_child\r\n raise child_exception_type(errno_num, err_msg, err_filename)\r\n FileNotFoundError: [Errno 2] No such file or directory: 'cmake'\r\n \r\n During handling of the above exception, another exception occurred:\r\n \r\n Traceback (most recent call last):\r\n File \"<string>\", line 2, in <module>\r\n File \"<pip-setuptools-caller>\", line 34, in <module>\r\n File \"/tmp/pip-install-musj9gr5/natten_ea055ed283804ee0b0ef252dd4ad0d6c/setup.py\", line 210, in <module>\r\n setup(\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/__init__.py\", line 103, in setup\r\n return distutils.core.setup(**attrs)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/core.py\", line 185, in setup\r\n return run_commands(dist)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/core.py\", line 201, in run_commands\r\n dist.run_commands()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/dist.py\", line 969, in run_commands\r\n self.run_command(cmd)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/dist.py\", line 963, in run_command\r\n super().run_command(command)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/dist.py\", line 988, in run_command\r\n cmd_obj.run()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/wheel/bdist_wheel.py\", line 368, in run\r\n self.run_command(\"build\")\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/cmd.py\", line 318, in run_command\r\n self.distribution.run_command(command)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/dist.py\", line 963, in run_command\r\n super().run_command(command)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/dist.py\", line 988, in run_command\r\n cmd_obj.run()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/command/build.py\", line 131, in run\r\n self.run_command(cmd_name)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/cmd.py\", line 318, in run_command\r\n self.distribution.run_command(command)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/dist.py\", line 963, in run_command\r\n super().run_command(command)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/dist.py\", line 988, in run_command\r\n cmd_obj.run()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/command/build_ext.py\", line 88, in run\r\n _build_ext.run(self)\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/command/build_ext.py\", line 345, in run\r\n self.build_extensions()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/command/build_ext.py\", line 467, in build_extensions\r\n self._build_extensions_serial()\r\n File \"/home/circleci/.pyenv/versions/3.8.12/lib/python3.8/site-packages/setuptools/_distutils/command/build_ext.py\", line 493, in _build_extensions_serial\r\n self.build_extension(ext)\r\n File \"/tmp/pip-install-musj9gr5/natten_ea055ed283804ee0b0ef252dd4ad0d6c/setup.py\", line 149, in build_extension\r\n raise RuntimeError(\"Cannot find CMake executable\")\r\n RuntimeError: Cannot find CMake executable\r\n [end of output]\r\n \r\n note: This error originates from a subprocess, and is likely not a problem with pip.\r\n ERROR: Failed building wheel for natten\r\n Running setup.py clean for natten\r\nFailed to build natten\r\nERROR: Could not build wheels for natten, which is required to install pyproject.toml-based projects\r\n```\r\n\r\n@ydshieh do you know if something was changed in the workflows?", "Rebase on main should work - https://huggingface.slack.com/archives/C01NE71C4F7/p1704883154443469", "Also cc @SunMarc ", "@amyeroberts Thank you for the review. I modified the test to use a small model (1.6 MB)." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28414/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28414", "html_url": "https://github.com/huggingface/transformers/pull/28414", "diff_url": "https://github.com/huggingface/transformers/pull/28414.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28414.patch", "merged_at": 1705411791000 }

{ "login": "Tiziano41", "id": 156085316, "node_id": "U_kgDOCU2sRA", "avatar_url": "https://avatars.githubusercontent.com/u/156085316?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Tiziano41", "html_url": "https://github.com/Tiziano41", "followers_url": "https://api.github.com/users/Tiziano41/followers", "following_url": "https://api.github.com/users/Tiziano41/following{/other_user}", "gists_url": "https://api.github.com/users/Tiziano41/gists{/gist_id}", "starred_url": "https://api.github.com/users/Tiziano41/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Tiziano41/subscriptions", "organizations_url": "https://api.github.com/users/Tiziano41/orgs", "repos_url": "https://api.github.com/users/Tiziano41/repos", "events_url": "https://api.github.com/users/Tiziano41/events{/privacy}", "received_events_url": "https://api.github.com/users/Tiziano41/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @Tiziano41 \r\nThanks for the issue! It seems you are using the `trust_remote_code=True` model, therefore the bug should be on their end. Would you mind opening a similar issue on the original repository on 🤗 Hub? https://huggingface.co./microsoft/phi-2 ", "Alternatively you can use the HF version of Phi-2 (i.e. without `trust_remote_code=True`) which should be here https://huggingface.co./susnato/phi-2 if I am not mistaken cc @ArthurZucker @susnato ", "Yes please use it from `susnato/phi-2` until the codebase is changed and in proper order at `microsoft/phi-2` 😃 ", "@younesbelkada @susnato OK, Thank you both for the fast reply! I will open the issue on their repository.", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28413/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "RonanKMcGovern", "id": 78278410, "node_id": "MDQ6VXNlcjc4Mjc4NDEw", "avatar_url": "https://avatars.githubusercontent.com/u/78278410?v=4", "gravatar_id": "", "url": "https://api.github.com/users/RonanKMcGovern", "html_url": "https://github.com/RonanKMcGovern", "followers_url": "https://api.github.com/users/RonanKMcGovern/followers", "following_url": "https://api.github.com/users/RonanKMcGovern/following{/other_user}", "gists_url": "https://api.github.com/users/RonanKMcGovern/gists{/gist_id}", "starred_url": "https://api.github.com/users/RonanKMcGovern/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/RonanKMcGovern/subscriptions", "organizations_url": "https://api.github.com/users/RonanKMcGovern/orgs", "repos_url": "https://api.github.com/users/RonanKMcGovern/repos", "events_url": "https://api.github.com/users/RonanKMcGovern/events{/privacy}", "received_events_url": "https://api.github.com/users/RonanKMcGovern/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Sorry, should have been posted in text-generation inference.\r\n\r\nAlso, TGI does work with AWQ. It's just that it doesn't work - at the time of writing with TheBloke's quant. See details [here](https://huggingface.co./TheBloke/Mixtral-8x7B-Instruct-v0.1-AWQ/discussions/1)." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28412/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "nogifeet", "id": 72322393, "node_id": "MDQ6VXNlcjcyMzIyMzkz", "avatar_url": "https://avatars.githubusercontent.com/u/72322393?v=4", "gravatar_id": "", "url": "https://api.github.com/users/nogifeet", "html_url": "https://github.com/nogifeet", "followers_url": "https://api.github.com/users/nogifeet/followers", "following_url": "https://api.github.com/users/nogifeet/following{/other_user}", "gists_url": "https://api.github.com/users/nogifeet/gists{/gist_id}", "starred_url": "https://api.github.com/users/nogifeet/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/nogifeet/subscriptions", "organizations_url": "https://api.github.com/users/nogifeet/orgs", "repos_url": "https://api.github.com/users/nogifeet/repos", "events_url": "https://api.github.com/users/nogifeet/events{/privacy}", "received_events_url": "https://api.github.com/users/nogifeet/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey! THanks for opening an issue, but without an isolated reproducible snippet + a traceback that shows this is an issue from transfromers I am not sure how we can help! 🤗 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28411/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Hannibal046", "id": 38466901, "node_id": "MDQ6VXNlcjM4NDY2OTAx", "avatar_url": "https://avatars.githubusercontent.com/u/38466901?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Hannibal046", "html_url": "https://github.com/Hannibal046", "followers_url": "https://api.github.com/users/Hannibal046/followers", "following_url": "https://api.github.com/users/Hannibal046/following{/other_user}", "gists_url": "https://api.github.com/users/Hannibal046/gists{/gist_id}", "starred_url": "https://api.github.com/users/Hannibal046/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Hannibal046/subscriptions", "organizations_url": "https://api.github.com/users/Hannibal046/orgs", "repos_url": "https://api.github.com/users/Hannibal046/repos", "events_url": "https://api.github.com/users/Hannibal046/events{/privacy}", "received_events_url": "https://api.github.com/users/Hannibal046/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This behavior can be changed by the `add_bos_token` attribute. something like `tokenizer = LlamaTokenizer.from_pretrained(\"meta-llama/Llama-2-7b\", add_bos_token = False)` to de-activate the addition of the token. \r\nALso `tokenizer(..., add_special_tokens = False)` de activates this as well\r\n", "Hi, thanks for response! It seems that the problem is not in the special token. Even if I set `add_special_tokens=False`, it would still output the same length.\r\n\r\n\r\nI check the last token of `prompt`, it is an empty string rather than a space in the prompt.\r\n\r\n\r\nSo I guess this is the default behavior of Tokenizer?", "You should use `tokenize` to check the tokens or `encode`:\r\n```python\r\nIn [2]: tokenizer.tokenize(prompt)\r\nOut[2]: \r\n[...,\r\n 'The',\r\n '▁answer',\r\n '▁is',\r\n ':',\r\n '▁']\r\n\r\nIn [3]: tokenizer.tokenize(label)\r\nOut[3]: \r\n[..., \r\n 'The',\r\n '▁answer',\r\n '▁is',\r\n ':',\r\n '▁C']\r\n```\r\nas you can see, `_` or `▁C` is a single token. This is expected. It's just not the same token. ", "Thanks! Got it!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28410/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "DanieleVeri", "id": 20779433, "node_id": "MDQ6VXNlcjIwNzc5NDMz", "avatar_url": "https://avatars.githubusercontent.com/u/20779433?v=4", "gravatar_id": "", "url": "https://api.github.com/users/DanieleVeri", "html_url": "https://github.com/DanieleVeri", "followers_url": "https://api.github.com/users/DanieleVeri/followers", "following_url": "https://api.github.com/users/DanieleVeri/following{/other_user}", "gists_url": "https://api.github.com/users/DanieleVeri/gists{/gist_id}", "starred_url": "https://api.github.com/users/DanieleVeri/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/DanieleVeri/subscriptions", "organizations_url": "https://api.github.com/users/DanieleVeri/orgs", "repos_url": "https://api.github.com/users/DanieleVeri/repos", "events_url": "https://api.github.com/users/DanieleVeri/events{/privacy}", "received_events_url": "https://api.github.com/users/DanieleVeri/received_events", "type": "User", "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" } ]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28409/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "regisss", "id": 15324346, "node_id": "MDQ6VXNlcjE1MzI0MzQ2", "avatar_url": "https://avatars.githubusercontent.com/u/15324346?v=4", "gravatar_id": "", "url": "https://api.github.com/users/regisss", "html_url": "https://github.com/regisss", "followers_url": "https://api.github.com/users/regisss/followers", "following_url": "https://api.github.com/users/regisss/following{/other_user}", "gists_url": "https://api.github.com/users/regisss/gists{/gist_id}", "starred_url": "https://api.github.com/users/regisss/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/regisss/subscriptions", "organizations_url": "https://api.github.com/users/regisss/orgs", "repos_url": "https://api.github.com/users/regisss/repos", "events_url": "https://api.github.com/users/regisss/events{/privacy}", "received_events_url": "https://api.github.com/users/regisss/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Related to this PR, I opened an issue in `datasets` to improve the user experience when using `DatasetDict.column_names`:\r\n- https://github.com/huggingface/datasets/issues/6571", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28408). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "I think we should also add `image_column_name` (default `image`) and `label_column_name` (default `label`) to the example's `DataTrainingArguments` to be consistent with [`audio-classification`](https://github.com/huggingface/transformers/blob/357971ec367fecb9951ae3218feafece5f61416a/examples/pytorch/audio-classification/run_audio_classification.py#L95-L101) as renaming `img` to `image` is not general enough.", "Thanks @regisss " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28408/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28408", "html_url": "https://github.com/huggingface/transformers/pull/28408", "diff_url": "https://github.com/huggingface/transformers/pull/28408.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28408.patch", "merged_at": 1705388648000 }

{ "login": "patrickvonplaten", "id": 23423619, "node_id": "MDQ6VXNlcjIzNDIzNjE5", "avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4", "gravatar_id": "", "url": "https://api.github.com/users/patrickvonplaten", "html_url": "https://github.com/patrickvonplaten", "followers_url": "https://api.github.com/users/patrickvonplaten/followers", "following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}", "gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}", "starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions", "organizations_url": "https://api.github.com/users/patrickvonplaten/orgs", "repos_url": "https://api.github.com/users/patrickvonplaten/repos", "events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}", "received_events_url": "https://api.github.com/users/patrickvonplaten/received_events", "type": "User", "site_admin": false }

[]

[]

[ "ah, you are too fast. See this comment \r\n\r\nhttps://github.com/huggingface/transformers/pull/28400#discussion_r1445856908\r\n\r\nin short: `mozilla-foundation/common_voice_11_0` doesn't require token if this works for you", "> mozilla-foundation/common_voice_11_0\r\n\r\nThe problem with CV 11.0 is that the expected results will change. 6_1 is exactly the dataset that was there before as well - is it that difficult. Don't we already have access to secret tokens? E.g. couldn't we just use `HUGGINGFACE_PUSH`? ", "I personally don't know if `HUGGINGFACE_PUSH` is the right token (is it a HF token or github token).\r\n\r\nWe have\r\n\r\n```\r\n token: ${{ secrets.HUGGINGFACE_PUSH }}\r\n hf_token: ${{ secrets.HF_DOC_BUILD_PUSH }}\r\n```\r\nSince we have such tokens, I think the infra team is fine with such tokens. We just need to talk to the infra team to make sure we are using the right one. And IMO, for read access, we should not use a push token (and better to have a read only token).\r\n\r\nI will talk to the infra.", "There is a new secret (token) added. I will update this PR accordingly and push", "well, I have to add it to all workflow files 😅 ", "Thanks a lot @ydshieh! ", "This PR should be good to merge! " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28407/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28407", "html_url": "https://github.com/huggingface/transformers/pull/28407", "diff_url": "https://github.com/huggingface/transformers/pull/28407.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28407.patch", "merged_at": 1704922536000 }

{ "login": "SangbumChoi", "id": 34004152, "node_id": "MDQ6VXNlcjM0MDA0MTUy", "avatar_url": "https://avatars.githubusercontent.com/u/34004152?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SangbumChoi", "html_url": "https://github.com/SangbumChoi", "followers_url": "https://api.github.com/users/SangbumChoi/followers", "following_url": "https://api.github.com/users/SangbumChoi/following{/other_user}", "gists_url": "https://api.github.com/users/SangbumChoi/gists{/gist_id}", "starred_url": "https://api.github.com/users/SangbumChoi/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SangbumChoi/subscriptions", "organizations_url": "https://api.github.com/users/SangbumChoi/orgs", "repos_url": "https://api.github.com/users/SangbumChoi/repos", "events_url": "https://api.github.com/users/SangbumChoi/events{/privacy}", "received_events_url": "https://api.github.com/users/SangbumChoi/received_events", "type": "User", "site_admin": false }

[]

[]

[ "```\r\n- conditional_detr\r\n \r\n root@0a2b4fe54761:/mnt/nas2/users/sbchoi/transformers# RUN_SLOW=1 pytest tests/models/conditional_detr/\r\n ================================================================= test session starts ==================================================================\r\n platform linux -- Python 3.10.13, pytest-7.4.4, pluggy-1.0.0\r\n rootdir: /mnt/nas2/users/sbchoi/transformers\r\n configfile: pyproject.toml\r\n plugins: hypothesis-6.92.0, hydra-core-1.3.2\r\n collected 149 items\r\n \r\n tests/models/conditional_detr/test_image_processing_conditional_detr.py .............. [ 9%]\r\n tests/models/conditional_detr/test_modeling_conditional_detr.py .......................ssssss..sssssssss......s..............s......s........s.. [ 63%]\r\n ..sssssssssss.sssssssssssssss.s..s.......s............. [100%]\r\n \r\n =================================================================== warnings summary ===================================================================\r\n ../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n \r\n ```\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\\\n\")\r\n \r\n ```\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#keyword-namespace-packages\r\n declare_namespace(pkg)\r\n \r\n tests/models/conditional_detr/test_modeling_conditional_detr.py::ConditionalDetrModelTest::test_disk_offload_bin\r\n /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.**get**(instance, owner)()\r\n \r\n tests/models/conditional_detr/test_modeling_conditional_detr.py::ConditionalDetrModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:460: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(init_instance.linear.bias, expected_bias, rtol=1e-3, atol=1e-4)\r\n \r\n tests/models/conditional_detr/test_modeling_conditional_detr.py::ConditionalDetrModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:463: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(\r\n \r\n tests/models/conditional_detr/test_modeling_conditional_detr.py::ConditionalDetrModelTest::test_pipeline_object_detection\r\n /opt/conda/lib/python3.10/site-packages/huggingface_hub/repocard.py:105: UserWarning: Repo card metadata block was not found. Setting CardData to empty.\r\n warnings.warn(\"Repo card metadata block was not found. Setting CardData to empty.\")\r\n \r\n - - Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html\r\n ===================================================== 101 passed, 48 skipped, 9 warnings in 54.71s =====================================================\r\n- deformable_detr\r\n \r\n root@0a2b4fe54761:/mnt/nas2/users/sbchoi/transformers# RUN_SLOW=1 pytest tests/models/deformable_detr/\r\n ================================================================= test session starts ==================================================================\r\n platform linux -- Python 3.10.13, pytest-7.4.4, pluggy-1.0.0\r\n rootdir: /mnt/nas2/users/sbchoi/transformers\r\n configfile: pyproject.toml\r\n plugins: hypothesis-6.92.0, hydra-core-1.3.2\r\n collected 151 items\r\n \r\n tests/models/deformable_detr/test_image_processing_deformable_detr.py .............. [ 9%]\r\n tests/models/deformable_detr/test_modeling_deformable_detr.py .......................ssssss.ssssssssss......s..............s......s............. [ 63%]\r\n sssssssssss.sssssssssssssss.s..s.......s............... [100%]\r\n \r\n =================================================================== warnings summary ===================================================================\r\n ../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n \r\n ```\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\\\n\")\r\n \r\n ```\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#keyword-namespace-packages\r\n declare_namespace(pkg)\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_disk_offload_bin\r\n /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.**get**(instance, owner)()\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:460: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(init_instance.linear.bias, expected_bias, rtol=1e-3, atol=1e-4)\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:463: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_model_outputs_equivalence\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/deformable_detr/test_modeling_deformable_detr.py:372: UserWarning: Use of index_put_ on expanded tensors is deprecated. Please clone() the tensor before performing this operation. This also applies to advanced indexing e.g. tensor[indices] = tensor (Triggered internally at /opt/conda/conda-bld/pytorch_1702400410390/work/aten/src/ATen/native/TensorAdvancedIndexing.cpp:708.)\r\n t[t != t] = 0\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_model_outputs_equivalence\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/deformable_detr/test_modeling_deformable_detr.py:372: UserWarning: Use of masked_fill_ on expanded tensors is deprecated. Please clone() the tensor before performing this operation. This also applies to advanced indexing e.g. tensor[mask] = scalar (Triggered internally at /opt/conda/conda-bld/pytorch_1702400410390/work/aten/src/ATen/native/cuda/Indexing.cu:1564.)\r\n t[t != t] = 0\r\n \r\n tests/models/deformable_detr/test_modeling_deformable_detr.py::DeformableDetrModelTest::test_pipeline_object_detection\r\n /opt/conda/lib/python3.10/site-packages/huggingface_hub/repocard.py:105: UserWarning: Repo card metadata block was not found. Setting CardData to empty.\r\n warnings.warn(\"Repo card metadata block was not found. Setting CardData to empty.\")\r\n \r\n - - Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html\r\n =============================================== 103 passed, 48 skipped, 11 warnings in 163.84s (0:02:43) ===============================================\r\n- deta\r\n \r\n root@0a2b4fe54761:/mnt/nas2/users/sbchoi/transformers# RUN_SLOW=1 pytest tests/models/deta\r\n ================================================================= test session starts ==================================================================\r\n platform linux -- Python 3.10.13, pytest-7.4.4, pluggy-1.0.0\r\n rootdir: /mnt/nas2/users/sbchoi/transformers\r\n configfile: pyproject.toml\r\n plugins: hypothesis-6.92.0, hydra-core-1.3.2\r\n collected 151 items\r\n \r\n tests/models/deta/test_image_processing_deta.py .............. [ 9%]\r\n tests/models/deta/test_modeling_deta.py ........s...............ssssss.ssssssssss......s..............s......s.............sssssssssss.sssssssss [ 78%]\r\n ssssss.s..s.......s.s............ [100%]\r\n \r\n =================================================================== warnings summary ===================================================================\r\n ../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n \r\n ```\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\\\n\")\r\n \r\n ```\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#keyword-namespace-packages\r\n declare_namespace(pkg)\r\n \r\n tests/models/deta/test_modeling_deta.py::DetaModelTest::test_disk_offload_bin\r\n /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.**get**(instance, owner)()\r\n \r\n tests/models/deta/test_modeling_deta.py::DetaModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:460: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(init_instance.linear.bias, expected_bias, rtol=1e-3, atol=1e-4)\r\n \r\n tests/models/deta/test_modeling_deta.py::DetaModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:463: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(\r\n \r\n tests/models/deta/test_modeling_deta.py::DetaModelTest::test_model_outputs_equivalence\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:1894: UserWarning: Use of index_put_ on expanded tensors is deprecated. Please clone() the tensor before performing this operation. This also applies to advanced indexing e.g. tensor[indices] = tensor (Triggered internally at /opt/conda/conda-bld/pytorch_1702400410390/work/aten/src/ATen/native/TensorAdvancedIndexing.cpp:708.)\r\n t[t != t] = 0\r\n \r\n tests/models/deta/test_modeling_deta.py::DetaModelTest::test_model_outputs_equivalence\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:1894: UserWarning: Use of masked_fill_ on expanded tensors is deprecated. Please clone() the tensor before performing this operation. This also applies to advanced indexing e.g. tensor[mask] = scalar (Triggered internally at /opt/conda/conda-bld/pytorch_1702400410390/work/aten/src/ATen/native/cuda/Indexing.cu:1564.)\r\n t[t != t] = 0\r\n \r\n - - Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html\r\n ==================================================== 101 passed, 50 skipped, 10 warnings in 53.04s =====================================================\r\n- table transformer\r\n \r\n root@0a2b4fe54761:/mnt/nas2/users/sbchoi/transformers# RUN_SLOW=1 pytest tests/models/table_transformer/\r\n ================================================================= test session starts ==================================================================\r\n platform linux -- Python 3.10.13, pytest-7.4.4, pluggy-1.0.0\r\n rootdir: /mnt/nas2/users/sbchoi/transformers\r\n configfile: pyproject.toml\r\n plugins: hypothesis-6.92.0, hydra-core-1.3.2\r\n collected 136 items\r\n \r\n tests/models/table_transformer/test_modeling_table_transformer.py .....................ssssss..sssssssss......s...............s......s.......... [ 57%]\r\n ...sssssssssss.sssssssssssssss.s..s..........s............ [100%]\r\n \r\n tests/models/deta/test_image_processing_deta.py .............. [ 9%]\r\n tests/models/deta/test_modeling_deta.py ........s...............ssssss.ssssssssss.....s..............s......s.............ssssssssss [ 70%]\r\n s.sssssssssssssss.s..s.......s.s............\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\n\")\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n [100%]\r\n \r\n ============================================================= warnings summary =============================================================\r\n ../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n \r\n ```\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\\\n\")\r\n \r\n ```\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n \r\n ```\r\n declare_namespace(pkg)\r\n \r\n ```\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n \r\n =================================================================== warnings summary ===================================================================\r\n ../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/**init**.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n \r\n ```\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\\\n\")\r\n \r\n ```\r\n \r\n src/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n \r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n ../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/**init**.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#keyword-namespace-packages\r\n declare_namespace(pkg)\r\n \r\n tests/models/table_transformer/test_modeling_table_transformer.py::TableTransformerModelTest::test_disk_offload_bin\r\n /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.**get**(instance, owner)()\r\n \r\n tests/models/table_transformer/test_modeling_table_transformer.py::TableTransformerModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:460: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(init_instance.linear.bias, expected_bias, rtol=1e-3, atol=1e-4)\r\n \r\n tests/models/table_transformer/test_modeling_table_transformer.py::TableTransformerModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:463: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(\r\n \r\n tests/models/table_transformer/test_modeling_table_transformer.py::TableTransformerModelTest::test_pipeline_object_detection\r\n /opt/conda/lib/python3.10/site-packages/huggingface_hub/repocard.py:105: UserWarning: Repo card metadata block was not found. Setting CardData to empty.\r\n warnings.warn(\"Repo card metadata block was not found. Setting CardData to empty.\")\r\n \r\n - - Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html\r\n ================================================ 89 passed, 47 skipped, 9 warnings in 103.04s (0:01:43) ================================================\r\n```\r\n\r\nCurrently I'm out of GPU so unavailable to test YOLOS :(", "@SangbumChoi Thanks for iterating! For MaskFormer, we still need to keep `use_auxiliary_loss`. Rather confusingly, this flag is set in two places: within the `deocder_config` and within the model's config. \r\n\r\nIn the maskformer modeling file, it's the [maskformer's config](https://github.com/huggingface/transformers/blob/07bdbebb48a9fe1e748348e4e14ae0b4659e54c4/src/transformers/models/maskformer/modeling_maskformer.py#L1755) - `self.config.use_auxiliary_loss` - which is used, rather than the deocder's - `self.config.decoder_config.auxiliary_loss`. ", "I can test YOLOs on my machine", "@SangbumChoi It seems that the aux loss tests haven't been added for YOLOs yet (all tests pass - but that's expected :)) ", "@amyeroberts For the case of YOLOs auxiliary head doesn't exist at the first place so the test cannot be established!", "> @SangbumChoi Thanks for iterating! For MaskFormer, we still need to keep `use_auxiliary_loss`. Rather confusingly, this flag is set in two places: within the `deocder_config` and within the model's config.\r\n> \r\n> In the maskformer modeling file, it's the [maskformer's config](https://github.com/huggingface/transformers/blob/07bdbebb48a9fe1e748348e4e14ae0b4659e54c4/src/transformers/models/maskformer/modeling_maskformer.py#L1755) - `self.config.use_auxiliary_loss` - which is used, rather than the deocder's - `self.config.decoder_config.auxiliary_loss`.\r\n\r\nAh I thought config.decoder.auxiliary_loss and config.use_auxiliary_loss was the same variable but it should be remained seperately. Let me revert this and fix again!", "Voila! Let me know if there are other things to change!\r\n\r\nBelow is the modified version of maskformer pytest\r\n\r\n```\r\nroot@bbc12aa272c6:/mnt/nas2/users/sbchoi/transformers# RUN_SLOW=1 pytest tests/models/maskformer/\r\n================================================================================================== test session starts ==================================================================================================\r\nplatform linux -- Python 3.10.13, pytest-7.4.4, pluggy-1.0.0\r\nrootdir: /mnt/nas2/users/sbchoi/transformers\r\nconfigfile: pyproject.toml\r\nplugins: hypothesis-6.92.0, hydra-core-1.3.2\r\ncollected 247 items\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py ...................... [ 8%]\r\ntests/models/maskformer/test_modeling_maskformer.py ........ssssss..sssssssss...s.........s.......s.........sssssssss.ssssssssssssssssss.s..s.....s................. [ 54%]\r\ntests/models/maskformer/test_modeling_maskformer_swin.py s........ssssss.ssssssssss..s........ss.......s.......ssssss.sssssssssssssssssssss.s......s.s.................... [100%]\r\n\r\n=================================================================================================== warnings summary ====================================================================================================\r\n../../../../../opt/conda/lib/python3.10/site-packages/_pytest/config/__init__.py:1373\r\n /opt/conda/lib/python3.10/site-packages/_pytest/config/__init__.py:1373: PytestConfigWarning: Unknown config option: doctest_glob\r\n\r\n self._warn_or_fail_if_strict(f\"Unknown config option: {key}\\n\")\r\n\r\nsrc/transformers/deepspeed.py:23\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\r\n warnings.warn(\r\n\r\n../../../../../opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28\r\n /opt/conda/lib/python3.10/site-packages/torch/utils/cpp_extension.py:28: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html\r\n from pkg_resources import packaging # type: ignore[attr-defined]\r\n\r\n../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/__init__.py:2871\r\n../../../../../opt/conda/lib/python3.10/site-packages/pkg_resources/__init__.py:2871\r\n /opt/conda/lib/python3.10/site-packages/pkg_resources/__init__.py:2871: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('ruamel')`.\r\n Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#keyword-namespace-packages\r\n declare_namespace(pkg)\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_call_with_segmentation_maps\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/image_processing_maskformer.py:699: FutureWarning: The `pad_and_return_pixel_mask` argument is deprecated and will be removed in v4.27\r\n warnings.warn(\r\n\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_feature_extraction\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_image_segmentation\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_fp16\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_instance_segmentation_head\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_instance_segmentation_head_resnet_backbone\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_no_head\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_with_segmentation_maps_and_loss\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/image_processing_maskformer.py:410: FutureWarning: The `size_divisibility` argument is deprecated and will be removed in v4.27. Please use `size_divisor` instead.\r\n warnings.warn(\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_image_processor_from_dict_with_kwargs\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_feature_extraction\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_image_segmentation\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_fp16\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_instance_segmentation_head\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_instance_segmentation_head_resnet_backbone\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_inference_no_head\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelIntegrationTest::test_with_segmentation_maps_and_loss\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/image_processing_maskformer.py:417: FutureWarning: The `max_size` argument is deprecated and will be removed in v4.27. Please use size['longest_edge'] instead.\r\n warnings.warn(\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_instance_segmentation\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_semantic_segmentation\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_feature_extraction\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_pipeline_image_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/image_processing_maskformer.py:428: FutureWarning: The `reduce_labels` argument is deprecated and will be removed in v4.27. Please use `do_reduce_labels` instead.\r\n warnings.warn(\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_instance_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:300: DeprecationWarning: Please use assertEqual instead.\r\n self.assertEquals(inputs[\"mask_labels\"][0].sum().item(), 41527.0)\r\n\r\n\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:395: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\r\n self.assertTrue(torch.allclose(inputs[\"class_labels\"][0], torch.tensor(expected_class_labels)))\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_panoptic_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:403: DeprecationWarning: Please use assertEqual instead.\r\n self.assertEquals(inputs[\"mask_labels\"][0].sum().item(), 315193.0)\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_panoptic_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:404: DeprecationWarning: Please use assertEqual instead.\r\n self.assertEquals(inputs[\"mask_labels\"][1].sum().item(), 350747.0)\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_semantic_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:342: DeprecationWarning: Please use assertEqual instead.\r\n self.assertEquals(inputs[\"mask_labels\"][0].sum().item(), 170200.0)\r\n\r\ntests/models/maskformer/test_image_processing_maskformer.py::MaskFormerImageProcessingTest::test_integration_semantic_segmentation\r\n /mnt/nas2/users/sbchoi/transformers/tests/models/maskformer/test_image_processing_maskformer.py:343: DeprecationWarning: Please use assertEqual instead.\r\n self.assertEquals(inputs[\"mask_labels\"][1].sum().item(), 257036.0)\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_fast_init_context_manager\r\ntests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:460: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(init_instance.linear.bias, expected_bias, rtol=1e-3, atol=1e-4)\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_fast_init_context_manager\r\ntests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_fast_init_context_manager\r\n /mnt/nas2/users/sbchoi/transformers/tests/test_modeling_common.py:463: FutureWarning: `torch.testing.assert_allclose()` is deprecated since 1.12 and will be removed in a future release. Please use `torch.testing.assert_close()` instead. You can find detailed upgrade instructions in https://github.com/pytorch/pytorch/issues/61844.\r\n torch.testing.assert_allclose(\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_fast_init_context_manager\r\n /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.__get__(instance, owner)\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:212: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if num_channels != self.num_channels:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:202: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if width % self.patch_size[1] != 0:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:205: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if height % self.patch_size[0] != 0:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:574: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n was_padded = pad_values[3] > 0 or pad_values[5] > 0\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:575: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if was_padded:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::Mas\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:248: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n should_pad = (height % 2 == 1) or (width % 2 == 1)\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer_swin.py:249: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if should_pad:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /opt/conda/lib/python3.10/site-packages/torch/nn/functional.py:2443: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if size_prods == 1:\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer.py:537: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if attn_weights.size() != (batch_size * self.num_heads, target_len, source_len):\r\n\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_attentions\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_output_hidden_state\r\ntests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_torchscript_simple\r\n /mnt/nas2/users/sbchoi/transformers/src/transformers/models/maskformer/modeling_maskformer.py:568: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!\r\n if attn_output.size() != (batch_size * self.num_heads, target_len, self.head_dim):\r\n\r\n-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html\r\n===================================================================================== 148 passed, 99 skipped, 69 warnings in 55.89s =====================================================================================\r\n\r\n```", "@amyeroberts Changed as you suggested! However, do you know the method rerun the CI? it has time out error.", "@SangbumChoi I can re-run :) ", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28406). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28406/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 1, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28406", "html_url": "https://github.com/huggingface/transformers/pull/28406", "diff_url": "https://github.com/huggingface/transformers/pull/28406.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28406.patch", "merged_at": 1705673522000 }

{ "login": "younesbelkada", "id": 49240599, "node_id": "MDQ6VXNlcjQ5MjQwNTk5", "avatar_url": "https://avatars.githubusercontent.com/u/49240599?v=4", "gravatar_id": "", "url": "https://api.github.com/users/younesbelkada", "html_url": "https://github.com/younesbelkada", "followers_url": "https://api.github.com/users/younesbelkada/followers", "following_url": "https://api.github.com/users/younesbelkada/following{/other_user}", "gists_url": "https://api.github.com/users/younesbelkada/gists{/gist_id}", "starred_url": "https://api.github.com/users/younesbelkada/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/younesbelkada/subscriptions", "organizations_url": "https://api.github.com/users/younesbelkada/orgs", "repos_url": "https://api.github.com/users/younesbelkada/repos", "events_url": "https://api.github.com/users/younesbelkada/events{/privacy}", "received_events_url": "https://api.github.com/users/younesbelkada/received_events", "type": "User", "site_admin": false }

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28405). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "I made some modification after some offline discussion, an exmaple model pushed can be found here: https://huggingface.co./ybelkada/test-tags-model lmk wdyt @amyeroberts @ArthurZucker @Wauplin \r\nI also updated the API on the PR description", "@amyeroberts @Wauplin I think that I have addressed most of your comments now, the API is much more flexible and less \"agressive\" thanks to your suggestions\r\n1- Creates a simple description\r\n2- Moved `transformers` to `library-name`\r\n3- Now Trainer and `model_tags` are compatible together, which was not the case in the first commits. e.g. when combining `SFTTrainer` from trl that automatically pushes `sft` and `trl` tags and this PR, all tags gets successfully pushed. E.g.:\r\n```python\r\nfrom transformers import AutoModelForCausalLM, TrainingArguments\r\nfrom trl import SFTTrainer\r\nfrom datasets import load_dataset\r\n\r\nmodel_name = \"HuggingFaceM4/tiny-random-LlamaForCausalLM\"\r\nmodel = AutoModelForCausalLM.from_pretrained(model_name)\r\n\r\nmodel.add_model_tags([\"tag-test\"])\r\n\r\ntrain_data = load_dataset(\"imdb\", split=\"train\")\r\n\r\ntrainer = SFTTrainer(\r\n model=model,\r\n args=TrainingArguments(output_dir=\"test-tag-already-tagged\"),\r\n train_dataset=train_data,\r\n dataset_text_field=\"text\"\r\n)\r\n\r\ntrainer.push_to_hub(\"ybelkada/test-tag-already-tagged\")\r\n```\r\nThe example repo is here: https://huggingface.co./ybelkada/test-tag-already-tagged\r\n4- If tags already exists it will not overwrite all the tags but add the new tags with existing tags. e.g. if one does `model.add_tags([\"test-tag-2\"])` after already pushing a model with a tag to Hub, as you can see from this repository: https://huggingface.co./ybelkada/test-model-already-tagged the new tag gets appended\r\n\r\ncc @osanseviero as well", "Thanks for handling this @younesbelkada " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28405/reactions", "total_count": 2, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 2, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28405", "html_url": "https://github.com/huggingface/transformers/pull/28405", "diff_url": "https://github.com/huggingface/transformers/pull/28405.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28405.patch", "merged_at": 1705326488000 }

{ "login": "koking0", "id": 45281765, "node_id": "MDQ6VXNlcjQ1MjgxNzY1", "avatar_url": "https://avatars.githubusercontent.com/u/45281765?v=4", "gravatar_id": "", "url": "https://api.github.com/users/koking0", "html_url": "https://github.com/koking0", "followers_url": "https://api.github.com/users/koking0/followers", "following_url": "https://api.github.com/users/koking0/following{/other_user}", "gists_url": "https://api.github.com/users/koking0/gists{/gist_id}", "starred_url": "https://api.github.com/users/koking0/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/koking0/subscriptions", "organizations_url": "https://api.github.com/users/koking0/orgs", "repos_url": "https://api.github.com/users/koking0/repos", "events_url": "https://api.github.com/users/koking0/events{/privacy}", "received_events_url": "https://api.github.com/users/koking0/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi,\r\n\r\nBoth should be equal (cc @younesbelkada).\r\n\r\nWhich difference are you seeing?", "I used the above code, and the initial error was that the preprocessor_config.json file was missing. \r\n\r\nI found that there was indeed no such file in https://huggingface.co./liuhaotian/llava-v1.5-7b/tree/main, so I started from https://huggingface.co./llava-hf/llava-1.5-7b-hf/tree/main Download this file, then put it in the model folder of liuhaotian/llava-v1.5-7b, and then run it, I encountered the following error:\r\n\r\n```\r\nvalueError: The input provided to the model are wrong. The number of image tokens is 0 while the number of image given to the model is 1. This prevents correct indexing and breaks batch generation.\r\n```\r\n\r\n\r\n", "Hi @koking0 \r\nunfortunately the checkpoints are not inter-compatible, if you want to use `liuhaotian/llava-v1.5-7b` you need to use the original author's library and not transformers", "ok, fine.", "Note however `liuhaotian/llava-v1.5-7b` should be the same as `llava-hf/llava-1.5-7b-hf`, same for the 13b model and so on " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28404/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "imoneoi", "id": 26354659, "node_id": "MDQ6VXNlcjI2MzU0NjU5", "avatar_url": "https://avatars.githubusercontent.com/u/26354659?v=4", "gravatar_id": "", "url": "https://api.github.com/users/imoneoi", "html_url": "https://github.com/imoneoi", "followers_url": "https://api.github.com/users/imoneoi/followers", "following_url": "https://api.github.com/users/imoneoi/following{/other_user}", "gists_url": "https://api.github.com/users/imoneoi/gists{/gist_id}", "starred_url": "https://api.github.com/users/imoneoi/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/imoneoi/subscriptions", "organizations_url": "https://api.github.com/users/imoneoi/orgs", "repos_url": "https://api.github.com/users/imoneoi/repos", "events_url": "https://api.github.com/users/imoneoi/events{/privacy}", "received_events_url": "https://api.github.com/users/imoneoi/received_events", "type": "User", "site_admin": false }

[]

[]

[ "It worries me a little bit that Mistral has different implementations for this specific process in the model.\r\n\r\nThey submitted a PR to vLLM which is aligned with the current Huggingface implementation:\r\n\r\nhttps://github.com/mistralai/vllm-release/commit/3e21dacb79471ebf946e72e67a5ca14ebcc598c1#diff-74473eb619768ed055303d29d74c47b1753615e8323e69467506b2b42d3f9898R341-R344", "Yep, we tried to follow this instead of the paper as they shared that code. Thanks for submitting a pr anyways 🤗 ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28403/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28403", "html_url": "https://github.com/huggingface/transformers/pull/28403", "diff_url": "https://github.com/huggingface/transformers/pull/28403.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28403.patch", "merged_at": null }

{ "login": "YJYJLee", "id": 28900943, "node_id": "MDQ6VXNlcjI4OTAwOTQz", "avatar_url": "https://avatars.githubusercontent.com/u/28900943?v=4", "gravatar_id": "", "url": "https://api.github.com/users/YJYJLee", "html_url": "https://github.com/YJYJLee", "followers_url": "https://api.github.com/users/YJYJLee/followers", "following_url": "https://api.github.com/users/YJYJLee/following{/other_user}", "gists_url": "https://api.github.com/users/YJYJLee/gists{/gist_id}", "starred_url": "https://api.github.com/users/YJYJLee/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/YJYJLee/subscriptions", "organizations_url": "https://api.github.com/users/YJYJLee/orgs", "repos_url": "https://api.github.com/users/YJYJLee/repos", "events_url": "https://api.github.com/users/YJYJLee/events{/privacy}", "received_events_url": "https://api.github.com/users/YJYJLee/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi,\r\n\r\nThis is because the inference API uses greedy decoding by default, whereas you are using beam search (as you are providing `num_beams=5` to the generate method). One would need to use the same generation parameters as [the ones used by the inference API](https://huggingface.co./docs/api-inference/detailed_parameters#text-generation-task) to compare apples-to-apples.", "> Hi,\r\n> \r\n> This is because the inference API uses greedy decoding by default, whereas you are using beam search (as you are providing `num_beams=5` to the generate method). One would need to use the same generation parameters as [the ones used by the inference API](https://huggingface.co./docs/api-inference/detailed_parameters#text-generation-task) to compare apples-to-apples.\r\n\r\n@NielsRogge Thanks! I understand the difference in output by using different decoding strategy, but the fact that beam search is not generating output sequence to the end seems like another problem.\r\n\r\nIs it a bug in beam search implementation?", "@NielsRogge \r\n\r\nAlso, I tried the greedy decoding by removing `num_beams=5`, but the output is still not generated to the end\r\n\r\n```\r\nDie Hauptbier der Insel ist \"Number One\", es ist nicht ein\r\n```", "You can increase `max_new_tokens`", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28402/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "weimingzha0", "id": 38259546, "node_id": "MDQ6VXNlcjM4MjU5NTQ2", "avatar_url": "https://avatars.githubusercontent.com/u/38259546?v=4", "gravatar_id": "", "url": "https://api.github.com/users/weimingzha0", "html_url": "https://github.com/weimingzha0", "followers_url": "https://api.github.com/users/weimingzha0/followers", "following_url": "https://api.github.com/users/weimingzha0/following{/other_user}", "gists_url": "https://api.github.com/users/weimingzha0/gists{/gist_id}", "starred_url": "https://api.github.com/users/weimingzha0/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/weimingzha0/subscriptions", "organizations_url": "https://api.github.com/users/weimingzha0/orgs", "repos_url": "https://api.github.com/users/weimingzha0/repos", "events_url": "https://api.github.com/users/weimingzha0/events{/privacy}", "received_events_url": "https://api.github.com/users/weimingzha0/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28401/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28401", "html_url": "https://github.com/huggingface/transformers/pull/28401", "diff_url": "https://github.com/huggingface/transformers/pull/28401.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28401.patch", "merged_at": null }

{ "login": "patrickvonplaten", "id": 23423619, "node_id": "MDQ6VXNlcjIzNDIzNjE5", "avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4", "gravatar_id": "", "url": "https://api.github.com/users/patrickvonplaten", "html_url": "https://github.com/patrickvonplaten", "followers_url": "https://api.github.com/users/patrickvonplaten/followers", "following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}", "gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}", "starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions", "organizations_url": "https://api.github.com/users/patrickvonplaten/orgs", "repos_url": "https://api.github.com/users/patrickvonplaten/repos", "events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}", "received_events_url": "https://api.github.com/users/patrickvonplaten/received_events", "type": "User", "site_admin": false }

[]

[]

[ "@ydshieh Also I don't fully understand why we use the `huggingface/transformers-pytorch-gpu` Docker image for the Pipeline tests here https://github.com/huggingface/transformers/blob/3b742ea84cfc32432d60c0b65c886576ef736833/.github/workflows/self-scheduled.yml#L248 , \r\nbut we use the `huggingface/transformers-all-latest-gpu` image for all other tests. Should we maybe change the image for the pipeline tests? ", "> why we use the huggingface/transformers-pytorch-gpu\r\n\r\nIIRC, this image doesn't have `tensorflow` stuff, but `transformers-all-latest-gpu` has both torch/tf. In theory, we can try to use only `transformers-all-latest-gpu`. But it's not a bad idea if we have environments that has only torch or only tf to test.\r\n\r\n\r\n(there is also a job `run_pipelines_tf_gpu` which uses `transformers-tensorflow-gpu`.)\r\n\r\n", "Moved the fix of Whisper's slow test here: https://github.com/huggingface/transformers/pull/28407 (cc @ydshieh) ", "@patrickvonplaten FYI The image fails to build due to `intel_extension_for_pytorch`, so we are still using the old image (therefore torch 2.1.0)\r\n\r\nI will have to check what happens\r\n\r\n```\r\nERROR: failed to solve: process \"sh -lc python3 -m pip install --no-cache-dir intel_extension_for_pytorch==$INTEL_TORCH_EXT -f https://developer.intel.com/ipex-whl-stable-cpu\" did not complete successfully: exit code: 1\r\nError: buildx failed with: ERROR: failed to solve: process \"sh -lc python3 -m pip install --no-cache-dir intel_extension_for_pytorch==$INTEL_TORCH_EXT -f https://developer.intel.com/ipex-whl-stable-cpu\" did not complete successfully: exit code: 1\r\n```" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28400/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28400", "html_url": "https://github.com/huggingface/transformers/pull/28400", "diff_url": "https://github.com/huggingface/transformers/pull/28400.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28400.patch", "merged_at": 1704794719000 }

{ "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", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28399/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 1, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28399", "html_url": "https://github.com/huggingface/transformers/pull/28399", "diff_url": "https://github.com/huggingface/transformers/pull/28399.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28399.patch", "merged_at": 1704980389000 }

{ "login": "amyeroberts", "id": 22614925, "node_id": "MDQ6VXNlcjIyNjE0OTI1", "avatar_url": "https://avatars.githubusercontent.com/u/22614925?v=4", "gravatar_id": "", "url": "https://api.github.com/users/amyeroberts", "html_url": "https://github.com/amyeroberts", "followers_url": "https://api.github.com/users/amyeroberts/followers", "following_url": "https://api.github.com/users/amyeroberts/following{/other_user}", "gists_url": "https://api.github.com/users/amyeroberts/gists{/gist_id}", "starred_url": "https://api.github.com/users/amyeroberts/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/amyeroberts/subscriptions", "organizations_url": "https://api.github.com/users/amyeroberts/orgs", "repos_url": "https://api.github.com/users/amyeroberts/repos", "events_url": "https://api.github.com/users/amyeroberts/events{/privacy}", "received_events_url": "https://api.github.com/users/amyeroberts/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28398/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28398", "html_url": "https://github.com/huggingface/transformers/pull/28398", "diff_url": "https://github.com/huggingface/transformers/pull/28398.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28398.patch", "merged_at": 1705062931000 }

{ "login": "asusdisciple", "id": 138434950, "node_id": "U_kgDOCEBZhg", "avatar_url": "https://avatars.githubusercontent.com/u/138434950?v=4", "gravatar_id": "", "url": "https://api.github.com/users/asusdisciple", "html_url": "https://github.com/asusdisciple", "followers_url": "https://api.github.com/users/asusdisciple/followers", "following_url": "https://api.github.com/users/asusdisciple/following{/other_user}", "gists_url": "https://api.github.com/users/asusdisciple/gists{/gist_id}", "starred_url": "https://api.github.com/users/asusdisciple/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/asusdisciple/subscriptions", "organizations_url": "https://api.github.com/users/asusdisciple/orgs", "repos_url": "https://api.github.com/users/asusdisciple/repos", "events_url": "https://api.github.com/users/asusdisciple/events{/privacy}", "received_events_url": "https://api.github.com/users/asusdisciple/received_events", "type": "User", "site_admin": false }

[]

[]

[ "cc @ylacombe ", "Hey @asusdisciple, thanks for opening this issue. I'll look into it, but first, could you send a snippet which show how you use the pipeline ?", "Sure. Here is how I initialize the pipeline:\r\n```\r\n\r\nobj = pipeline(task=\"automatic-speech-recognition\",\r\n model=\"facebook/seamless-m4t-v2-large\",\r\n device=device,\r\n chunk_length_s=30\r\n batch_size=64,\r\n generate_kwargs={\"num_beams\": 1,\r\n \"temperature\": 0,\r\n \"task\": \"transcribe\",\r\n \"do_sample\": True\r\n }\r\n )\r\n \r\n```\r\n\r\nAnd this is how I transcribe it with the given language:\r\n\r\n```\r\ntmp = model(audiopath,\r\n generate_kwargs={\"tgt_lang\": lang}\r\n )\r\n```\r\n\r\nJust remember the audiofile should be longer than 10s. I already see this issue with files which are 20s long but 1min probably works better. I am even unsure why it happens with 20s audiofiles because setting chunking to 30s should not affect those right?", "Thanks @asusdisciple, note that the chunk parameter for seq2seq models is highly experimental as pointed out in #20104.\r\n\r\n@Narsil, it could be good to have your opinion on this since you added the algorithm. Could you give an explanation of how it works and how to improve it ? Many thanks", "Could you share an actual file that triggers it, the expected output, and the actual output with chunking ? Having a fully complete reproducer really helps narrow down these kind of things.\r\n\r\nChunking works like the following. Chunk every audio into chunks of length `chunk_length_s` with overlapping (1/6 * chunk_length_s by default). We run the model on each of those, and then stitch the results by finding the maximum overlapping output, and fusing those strings together.\r\n\r\nMany things could be at play, it would be easier to get good hints with the actual issue.\r\n\r\n-> seamless m4t might not be trained on long audio, and having position_ids or something, making it harder for it to run on those extra chunks. Shouldn't affect chunked with chunks < trained length tho.\r\n-> There could be an issue with the chunking and/or stitching code.\r\n-> On whisper, without issue, the fact that the chunking ahppens arbitrarily in audio means that the model is hallucinating more, on both ends of the chunks, meaning the stitching will include the hallucination (since it's a maximally overlapping string). No good solution for that one, either try to precut audio along silences if they exist, or find a way to fine-tune the model to handle those. If it happens for whisper, it can happen here (although the degradation for whisper is a fraction of a percent of WER, as it also sometimes improves baseline whisper). Maybe S4t is more sensitive to it.\r\n", "For whisper I can state that on my FLEURS benchmark, where I concatenated files to longer version (5 and 30min) WER dropped by 10% for all HF implementations. Faster-Whisper (Repo) however performed exactly the same as on small files when I used the parameter `condition_on_previous_text=False`. If I used Faster-Whisper with this parameter on `True` the WER skyrocketed to 200%. So maybe this plays a role since you can not turn it off in HF.\r\n\r\nAs for seemless m4t-v2 here is a minimal reproducer with ground_truth text and the audio file, which was concatenated from a few sentences in FLEURS. WER 66%. English Language. Repetitions were cleaned. The audio file is in the uploads: \r\n[17837018222808398366.wav.zip](https://github.com/huggingface/transformers/files/13920168/17837018222808398366.wav.zip)\r\n\r\n```\r\nfrom transformers import pipeline\r\nimport jiwer\r\n\r\nground_truth = \"Police superintendent Chandra Shekhar Solanki said the accused appeared in court with covered faces. During the struggle for independence organised by the Mau movement, a peaceful gathering in the town resulted in the killing of the paramount chief Tupua Tamasese Lealofi III. Martelly swore in a new Provisional Electoral Council (CEP) of nine members yesterday. Europe is a continent that is relatively small but with many independent countries. Under normal circumstances, travelling through multiple countries would mean having to go through visa applications and passport control multiple times. Their thermal behavior is not as steady as large caves on Earth that often maintain a fairly constant temperature, but it is consistent with these being deep holes in the ground,' said Glen Cushing of the United States Geological Survey (USGS) Astrogeology Team and of Northern Arizona University located in Flagstaff, Arizona. Sleep interruption is the process of purposefully awakening during your normal sleep period and falling asleep a short time later (10–60 minutes). Most of the buildings on the edges of the complex have been rebuilt in order to give tourists a better idea of how they originally appeared. Angel (2006), explains the Continuum approach as a method being used to help organizations reach a higher level of performance. Due to the underwater topology the return flow is concentrated at a few deeper sections, and a fast current to deep water may form there. towards the end of the Middle Ages western Europe began to develop their own style. one of the biggest developments of the time as a result of the crusades people began to use buttons to fasten clothing. NextGen is a system the FAA claims would allow aircraft to fly shorter routes and save millions of gallons of fuel each year and cut carbon emissions. A full 20 percent of the water that pours out of the planet's rivers into the oceans comes from the Amazon. So, it is likely that the notation was added simply as a label. He was greeted by Singapore's Deputy Prime Minister Wong Kan Seng and discussed trade and terrorism issues with the Singapore Prime Minister Lee Hsien Loong. The disease is carried by pigs, which then migrates to humans through mosquitos. Elements like calcium and potassium are considered metals. Of course, there are also metals like silver and gold. Field trips are a large part of any classroom. Quite often a teacher would love to take her students places to which a bus trip is not an option. People now write messages on computer screens, never having to come close to a sharpener. One of the most common methods used to illustrate the importance of socialization is to draw upon the few unfortunate cases of children who were, through neglect, misfortune, or wilful abuse, not socialized by adults while they were growing up. He added that 'they should not, however, be asked to take on obligations that go beyond their development stage, responsibility and capabilities.' Ironing damp clothes can help them dry. Many hotels have an iron and ironing board available for loan, even if one is not present in the room. Apia is the capital of Samoa. The town is on the island of Upolu and has a population of just under 40,000. A curry is a dish based on herbs and spices, together with either meat or vegetables. Four skiers in the women's sitting group failed to finish their runs, and 45 of the 117 total skiers in the Giant Slalom failed to rank in the race. The debate was sparked by controversy over spending on relief and reconstruction in the wake Hurricane Katrina; which some fiscal conservatives have humorously labeled 'Bush's New Orleans Deal.' In just two weeks the Americans and Free French forces had liberated southern France and were turning towards Germany. The Great Pyramid at Giza is the only one of the seven wonders that is still standing today. The two compounds react with one another to form crystals that may block kidney function, researchers at the university said. Most interpretations of technological determinism share two general ideas: that the development of technology itself follows a path largely beyond cultural or political influence, and that technology in turn has 'effects' on societies that are inherent, rather than socially conditioned.\"\r\n\r\n\r\nobj = pipeline(task=\"automatic-speech-recognition\",\r\n model=\"facebook/seamless-m4t-v2-large\",\r\n device=\"cuda:0\",\r\n chunk_length_s=30, # NEVER USE THIS PARAMETER, DEGRADES SCORE FOR LONG AUDIO\r\n batch_size=4,\r\n generate_kwargs={\"num_beams\": 1,\r\n \"temperature\": 0,\r\n \"task\": \"transcribe\",\r\n \"do_sample\": True\r\n }\r\n )\r\n\r\n\r\naudiopath = \"...\"\r\n\r\npred = obj(audiopath,\r\n generate_kwargs={\"tgt_lang\": \"eng\"}\r\n )\r\n\r\n\r\nerr = round(jiwer.process_words(ground_truth, pred[\"text\"]).wer, 2)\r\nprint(err)\r\n\r\n```\r\n\r\n\r\n\r\n\r\n", "First of all, there is a BIG warning that this is experimental:\r\n\r\nhttps://github.com/huggingface/transformers/pull/20104/files#diff-ccc8e98fcaa81fdf6317a652438a309bcede0bbe336774288c2fcf91d9f11082R291-R296\r\n\r\nI played around a bit, I found a bug in ASR pipeline, although it won't fix your issues, since it really seems that it's the model that has a lot of issues with hallucinations.\r\n\r\n[simple.webm](https://github.com/huggingface/transformers/assets/204321/a5f111dc-cfb8-4b62-bf9e-6484cd162135)\r\n\r\nHere is the cut version (30s).\r\n\r\nWithout chunking the output is\r\n```\r\n'Police Superintendent Shandra Shankar Solanki said the accused appeared in court with their faces covered during th\r\ne independence struggle organized by the Maoist movement, which resulted in the killing of the Supreme Leader Tupa T\r\nupa Tupa Tsai.'\r\n```\r\n\r\nWith chunking (it will chunk at 25s and rerun the missing 5 with overlap)\r\n\r\n```\r\n'Police Superintendent Shandra Shankar Solanki said the accused appeared in court with their faces covered during the independence struggle organized by the Maoist movement, which resulted in the killing of the Supreme Leader Tupa Tupa Tupa Tsai. The newly elected president of the United States, Martin Luther King, Jr., was sworn in on Tuesday, July 9, at the age of 93.'\r\n```\r\n\r\nThe second sentence seems like pretty bad hallucination from the actual audio. \r\n\r\n@ylacombe is this model known for hallucinating so much ?\r\n\r\n", "Thanks for investigating @Narsil ! So this is the result after fixing the bug you found ?\r\n\r\n> @ylacombe is this model known for hallucinating so much?\r\n\r\nIt is, the model principal usage is translation. Using it in an ASR settings is likely to produce big hallucinations. Moreover, from my own usage, it is not really good with short audios.\r\n\r\n", "> So this is the result after fixing the bug you found ?\r\n\r\nThe bugfix doesn't really help that much tbh (so no the example doesn't include it, the bug is that we don't reinclude the `stride` parameter when the model is `seq2seq` withing the forward method, meaning we don't use the custom stitching for seq2seq, but given the high hallucination rate, it doesn't really help, it's actually worse on the entire file in this particular instance).", "Are there any benchmarks on hallucinations in m4t and benchmarks in general? In the paper the performance for ASR and M4T-v2 is almost double as good as Whisper-v2. In my own reproduction on Fleurs I was able to reproduce the paper results. \r\n\r\nAlso I know this feature is experimental, just wanted to let you guys now! I really appreciate the work you do and I know more than a few people who use huggingface implementations just for the sake of chunking. So I think its an important feature and even if the behaviour can't be changed, knowledge about things like this is key right? :) \r\n\r\n\r\n", "Hello @asusdisciple and @Narsil, circling back on this issue, I've used jiwer visualisation tools to pinpoint where the transcription fails. (`print(jiwer.visualize_alignment(jiwer.process_words(ground_truth.lower(), pred[\"text\"].lower())))`)\r\n\r\nTurns out M4T doesn't hallucinate that much\r\n\r\n\r\n```python\r\nsentence 1\r\nREF: police superintendent chandra shekhar solanki said the accused appeared in court with covered faces. during the struggle for independence organised by the mau movement, a peaceful gathering in the town resulted in the killing of the paramount chief ***** ******** *** ******** ***** ** * *** *********** ********* ******* tupua tamasese lealofi iii. martelly swore in a new provisional electoral council (cep) of nine members yesterday. europe is a continent that is relatively small but with many independent countries. under normal circumstances, travelling through multiple countries would mean having to go through visa applications and passport ********* ****** ************* ********** ******* ******** ********* ***** **** ****** ** ** ******* **** ************ *** ******** control multiple times. their thermal behavior is not as steady as large caves on earth that often maintain a fairly constant temperature, but it is consistent with these being deep holes in the ground,' said glen cushing of the united states geological survey *** **** ********** **** ***** ***** **** ***** ** *** ******* **** ***** ******* ** *** ****** (usgs) astrogeology team and of northern arizona university located in flagstaff, arizona. sleep interruption is the process of ************ purposefully awakening during your normal sleep period and falling asleep a short time later (10–60 minutes). most of the buildings on the edges of the complex have been rebuilt in order to give tourists a better idea of how they originally appeared. angel (2006), explains the continuum approach as a method being used to help organizations reach a higher level of performance. due to the underwater topology the return flow is concentrated at a few deeper sections, and a fast current to deep water may form there. towards the end of the middle ages western europe began to develop their own style. one of the biggest developments of the time as a result of the crusades people began to use buttons to fasten clothing. nextgen is a system the faa claims would allow aircraft to fly shorter routes and save millions of gallons of fuel each year and cut carbon emissions. a full 20 percent of the water that pours out of the planet's rivers into the oceans comes from the amazon. so, it is likely that the notation was added simply as a label. he was greeted by singapore's deputy prime minister wong kan seng and discussed trade and terrorism issues with the singapore prime minister lee hsien loong. the disease is carried by pigs, which then migrates to humans through mosquitos. elements like calcium and potassium are considered metals. of course, there are also metals like silver and gold. field trips are a large part of any classroom. quite often a teacher would love to take her students places to which a bus trip is not an option. people now write messages on computer screens, never having to come close to a sharpener. one of the most common methods used to illustrate the importance of socialization is to draw upon the few unfortunate cases of children who were, through neglect, misfortune, or wilful abuse, not socialized by ** adults while they were growing up. he added that 'they should not, however, be asked to take on obligations that go beyond their development stage, responsibility and capabilities.' ironing damp clothes can help them dry. many hotels have an iron and ironing board available for loan, even if one is not present in the room. apia is the capital of samoa. the town is on the island of upolu and has a population of just under 40,000. a curry is a dish based on herbs and spices, together with either meat or vegetables. four skiers in the women's sitting group failed to finish their runs, and 45 of the 117 total skiers in the giant slalom failed to rank in the race. the debate was sparked by controversy over spending on relief and reconstruction in the wake ** hurricane katrina; which some fiscal conservatives have humorously labeled 'bush's new orleans deal.' in just two weeks the americans and free french forces had liberated southern france and were turning towards germany. the great pyramid at giza is the only one of the seven wonders that is still standing today. the two compounds react with one another to form crystals that may block kidney function, researchers at the university said. most interpretations of technological determinism share two general ***** **** *** *********** ** ********** ****** ******* * **** ******* ****** ******** ** ********* ********* **** *************** ** ************* *********** ***** *** ******* ideas: that the development of technology itself follows a path largely beyond cultural or political influence, and that technology in turn has 'effects' on societies that are inherent, rather than socially conditioned.\r\nHYP: police superintendent shandra shankar solansky said the accused appeared in court with covered faces during the struggle for independence organized by the mao movement a peaceful gathering in the town resulted in the killing of *** paramount chief tuppa tupaczai iii martelli swore in a new provisional electoral council of nine members yesterday martelli swore in a new provisional electoral council ***** of nine members yesterday. europe is a continent that is relatively small but with many independent countries. under normal circumstances, travelling through multiple countries would mean having to go through visa applications and passport controls. normal circumstances travelling through multiple countries would mean having to go through visa applications and passport control multiple times their thermal behaviour is not as steady as large caves on earth that often maintain a fairly constant temperature but ** is consistent with these being deep holes in the ground said glen cushing of the united states geological survey but it's consistent with these being deep holes in the ground, said glenn cushing of the united states geological survey and the northern arizona university located in flagstaff, arizona. ***** ************ ** the process of deliberately waking up during your normal sleep period and falling asleep a short time later. ****** ********* **** ** the best way to do this is ******* **** **** ******* ** ***** to use ******** a ****** **** ** *** **** ********** ********* ***** ******* ******** *** continuum approach ** * ****** ***** **** to help organizations achieve * higher levels of performance. *** ** the next gen *** ****** **** is ************ ** a system that ********* *** * **** ******* ** **** ***** *** **** ****** ******* the *** ** *** ****** **** ******* ****** ***** ** ******* ***** *** ****** *** ** *** ******* ************ ** *** **** ** * ****** ** *** ******** ****** ***** ** *** ******* ** ****** ********* ******* ** * ****** *** faa claims would allow aircraft to fly shorter routes and save millions of gallons of fuel each year and cut carbon emissions. * **** ** ******* ** the faa claims ***** *** ** *** ******** ****** **** *** ****** ***** **** *** ******* *** ** ** ****** that the system will allow aircraft to fly shorter routes and save millions of gallons of fuel each year **** and cut carbon emissions. ********* ****** **** the ********* ***** ******** *** ***** ****** *** disease is carried by pigs, which then migrates to humans through mosquitoes. the most common methods used to illustrate the importance ******* ***** *** **** ****** **** ****** *** ***** ***** ***** *** * ***** **** of socialization are: ***** ***** * ******* ***** **** ** **** *** ******** ****** ** ***** * *** **** ** *** ** ******* ****** *** ***** ******** ** ******** ******** ***** ****** ** **** ***** ** * ********** one of the most common methods used to illustrate the importance of socialization is to draw on the *** unfortunate cases of children who were through neglect, misfortune or willful abuse not socialized by an adult while they were growing up. the city of ***** ****** **** ******** ** ***** ** **** ** *********** **** ** ****** ***** *********** ****** ************** *** ************** ******* **** ******* *** **** **** **** **** ****** **** ** **** *** ******* ***** ********* *** ***** **** ** *** ** *** ******* ** *** ***** apia is the capital of samoa. the city of upu is home to ***** *** *** a population of just under 40,000 people. ***** ** * **** ***** ** ***** *** ******* ******** **** ****** **** ** *********** **** ****** ** the ******* ******* ***** ****** ** ****** ***** ***** *** ** ** *** *** ***** ****** ** *** ***** ****** ****** ** **** ** *** ***** *** debate was sparked by controversy over spending on relief and reconstruction in the wake of hurricane katrina, which some fiscal conservatives have humorously labeled bush's new orleans deal. its **** *** ***** *** ********* *** **** ****** forces had liberated southern france and were turning towards germany the great pyramid at giza is *** **** one of the seven wonders that is still standing today the two compounds react with one another to form crystals that may block kidney function researchers at the university said most interpretations of technological determinism share two general ideas that the development of technology itself follows a path largely beyond cultural or political influence most interpretations of technological determinism share two general ideas: that the development of technology itself follows a path largely beyond cultural or political influence and that technology in turn has effects on society that are inherent rather than socially conditioned.\r\n S S S S S S S D I I I I I I I I I I I S S S S S D I I I I I I I I I I I I I I I I I S S S D S I I I I I I I I I I I I I I I I I S S S S D D D I S S S D D D D S S S S S S D D D D D D S D D D D D D D D D D D D D D D D D S D S D D S S D D D D D S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S S D D D D D D D D D D D D D D D D D S S S S S S S S S S S S S S S S S D S S S D D D D D D D D D D S S S S S S S S S S D D D D D D D D D D D D D D D S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S D S S S S I S S S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S S S S S S D D D S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D I S S S S D D D D D D D D S D D S S S I I I I I I I I I I I I I I I I I I I I I I I I S S S S \r\n\r\nnumber of sentences: 1\r\nsubstitutions=99 deletions=280 insertions=72 hits=333\r\n\r\nmer=57.53%\r\nwil=69.10%\r\nwip=30.90%\r\nwer=63.34%\r\n```\r\n\r\nIt seems that the transcription works quite well but repeats a lot of sequence -> indicate that it should work better with a low `stride_length_s`.\r\n\r\nI've tried with `stride_length_s=1`, and I already get a much better WER (45%).\r\n\r\n```python\r\nsentence 1\r\nREF: police superintendent chandra shekhar solanki said the accused appeared in court with covered faces. during the struggle for independence organised by the mau movement, a peaceful gathering in the town resulted in the killing of the paramount chief tupua tamasese lealofi iii. martelly swore in a new provisional electoral council (cep) of nine members yesterday. europe is a continent that is relatively small but with many independent countries. under normal circumstances, travelling through multiple countries would mean having to go through visa applications and passport control multiple times. their thermal behavior is not as steady as large caves on ****** *** ***** ** *** ******* earth that often **** ** ***** **** maintain a fairly constant temperature, but it is consistent with these being deep holes in the ground,' said glen cushing of the united states geological survey (usgs) astrogeology team and of northern arizona university located in flagstaff, arizona. sleep interruption is the process of ************ purposefully awakening during your normal sleep period and falling asleep a short time later (10–60 minutes). most of the buildings on the edges of the complex have been rebuilt in order to give tourists a better idea of how they originally appeared. angel (2006), explains the continuum approach as a method being used to help organizations reach a higher level of performance. due to the underwater topology the return flow is concentrated at a few deeper sections, and a fast current to deep water may form there. towards the end of the middle ages western europe began to develop their own style. one of the biggest developments of the time as a result of the crusades ***** people began to use buttons to fasten ***** ******* clothing. nextgen is a system the *** faa claims **** *** ****** would allow aircraft to fly shorter routes and save millions of gallons of fuel each year and cut carbon emissions. a full 20 percent of the water that pours out of the planet's rivers into the oceans comes from the amazon. so, it is likely that the notation was added simply as a label. he was greeted by singapore's deputy prime minister wong kan seng and discussed trade and terrorism issues with the singapore prime minister lee hsien loong. the disease is carried by pigs, which then migrates to humans through mosquitos. elements like calcium and potassium are considered metals. of course, there are also metals like silver and gold. field trips are a large part of any classroom. quite often a teacher would love to take her students places to which a bus trip is not an option. people now write messages on computer screens, never having to come close to a sharpener. one of the most common methods used to illustrate the importance of socialization is to draw upon the few unfortunate cases of children who were, through neglect, misfortune, or wilful abuse, not socialized by ** adults while they were growing up. he added that 'they should not, however, be asked to take on obligations that go beyond their development stage, responsibility and capabilities.' ironing damp clothes can help them dry. many hotels have an iron and ironing board available for loan, even if one is not present in the room. apia is the capital of samoa. the town is on the island of upolu and has a population of just under 40,000. a curry is a dish based on herbs and spices, together with either meat or vegetables. four skiers in the women's sitting group failed to finish their runs, and 45 of the 117 total skiers in the giant slalom failed to rank in the race. the debate was sparked by controversy over spending on relief and reconstruction in the wake ** hurricane katrina; which some fiscal conservatives have humorously labeled 'bush's new orleans deal.' in just two weeks the americans and free french forces had liberated southern france and were turning towards germany. the great pyramid at giza is the only one of the seven wonders that is still standing today. the two compounds react with one another to form crystals that may block kidney function, researchers at the university said. most interpretations of technological determinism share two general ideas: that the development of technology itself follows a path largely beyond cultural or political ********* ******** ** ********* influence, and that technology in turn has 'effects' on societies that are inherent, rather than socially conditioned.\r\nHYP: police superintendent shandra shankar solansky said the accused appeared in court with covered faces during the struggle for independence organized by the mao movement a peaceful gathering in the town resulted in the killing of *** paramount chief tuppa tupaczai iii martelli ******** swore in a new provisional electoral council ***** of nine members yesterday europe is a continent that is relatively small but with many independent countries. under normal circumstances, travelling through multiple countries would mean having to go through visa applications and passport control multiple times. their thermal behaviour is not as steady as large caves on earth. the study of the earth's crust is often done in caves that are relatively stable in temperature, but it is consistent with these being deep holes in the ground, said glenn cushing of the united states geological survey ****** ************ **** and the northern arizona university located in flagstaff, arizona. ***** ************ ** the process of deliberately waking up during your normal sleep period and falling asleep a short time later. ****** ********* **** ** *** ********* ** *** ***** ** *** ******* **** **** ******* ** ***** ** **** ******** * ****** **** ** *** **** ********** ********* ***** ******* ******** *** ********* ******** ** * ****** ***** **** ** **** ************* ***** * ****** ***** ** ************ due to the underwater topology the return flow is concentrated at a few deeper sections and a fast current to deep water may form there towards the end of the middle ages western europe began to develop their own style one of the biggest developments of that time as a result of the crusades began people began to use buttons to fasten their clothes next gen is a system the the faa claims that the system will allow aircraft to fly shorter routes and save millions of gallons of fuel each year and cut carbon emissions. discuss **** ** ******* ** *** ***** **** ***** *** ** *** ******** ****** **** *** ****** ***** **** *** ******* *** ** ** ****** **** *** ******** *** ***** ****** ** * ****** ** *** ******* ** *********** ****** ***** ******** **** *** **** *** ********* trade and terrorism issues with the singapore prime minister lee hsien long the disease is carried by pigs which then migrates to humans through mosquitoes elements like calcium and potassium are considered metals of course there are also metals like silver and gold field trips are a large part of any classroom quite often a teacher would love to take her students places to which a bus **** ** *** ** ******* ****** *** ***** ******** ** ******** ******** ***** ****** ** **** ***** ** * ********** one of the most common methods used to illustrate the importance of socialization is to draw on the *** unfortunate cases of children who were through neglect, misfortune or willful abuse not socialized by an adult while they were growing up. ** ***** **** ***** ****** **** ******** ** ***** ** **** ** *********** **** ** ****** ***** *********** ****** ************** *** ************** ******* **** ******* *** **** **** **** **** ****** **** ** **** *** ******* ***** ********* *** ***** **** ** *** ** *** ******* ** the island **** ** *** ******* of upu *** **** is home to ****** ** ***** *** *** a population of just under 40,000 people. ***** ** * **** ***** ** ***** *** ******* ******** **** ****** **** ** *********** **** ****** ** the ******* ******* ***** ****** ** ****** ***** ***** *** ** ** *** *** ***** ****** ** *** ***** ****** ****** ** **** ** *** ***** *** debate was sparked by controversy over spending on relief and reconstruction in the wake of hurricane katrina, which some fiscal conservatives have humorously labeled bush's new orleans deal. its **** *** ***** *** ********* *** **** ****** forces had liberated southern france and were turning towards germany the great pyramid at giza is *** **** one of the seven wonders that is still standing today the two compounds react with one another to form crystals that may block kidney function researchers at the university said most interpretations of technological determinism share two general ideas that the development of technology itself follows a path largely beyond cultural or political influence cultural or political influence and that technology in turn has effects on society that are inherent rather than socially conditioned.\r\n S S S S S S S D S S S S D D S S I I I I I I S S I I I I S S S S S S D D D S D D D I S S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S S S S I I I S S I I I I S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S S S S S S S D D D D D D D D D D D D D D D D D D D D S D S S S S I S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D S D D D D S D D S S D D D D D S S D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D I S S S S D D D D D D D D S D D S S S S I I I I S S S S \r\n\r\nnumber of sentences: 1\r\nsubstitutions=65 deletions=236 insertions=24 hits=411\r\n\r\nmer=44.16%\r\nwil=52.55%\r\nwip=47.45%\r\nwer=45.65%\r\n```\r\n\r\n~~It also seems that the quality of the transcription degrades the longer the audio is, despite using chunking. I'm indeed under the impression that there are more deletions as we reach the end of the audio.~~\r\n\r\nThis could maybe prevented by modifying a bit how the transcription algorithm `(_find_longest_common_sequence`) work.\r\nAt the moment, for each chunk, it checks the longest common sequence between the already concatened sequence (composed of the concatenation of each previous chunks) and the newest chunk, and add the newest chunk where it doesn't match.\r\n\r\nAs @Narsil indicates, some ways to improve it could maybe be linked to the stride parameters and the fact that we have some heuristics on how to stitch the different chunks\r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28397/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "liechtym", "id": 7433062, "node_id": "MDQ6VXNlcjc0MzMwNjI=", "avatar_url": "https://avatars.githubusercontent.com/u/7433062?v=4", "gravatar_id": "", "url": "https://api.github.com/users/liechtym", "html_url": "https://github.com/liechtym", "followers_url": "https://api.github.com/users/liechtym/followers", "following_url": "https://api.github.com/users/liechtym/following{/other_user}", "gists_url": "https://api.github.com/users/liechtym/gists{/gist_id}", "starred_url": "https://api.github.com/users/liechtym/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/liechtym/subscriptions", "organizations_url": "https://api.github.com/users/liechtym/orgs", "repos_url": "https://api.github.com/users/liechtym/repos", "events_url": "https://api.github.com/users/liechtym/events{/privacy}", "received_events_url": "https://api.github.com/users/liechtym/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Sorry I don't think this is related to `transformers` as there is a wrapper around it. `sdpa` is natively supported in transformers ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28396/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "zhongshsh", "id": 62104945, "node_id": "MDQ6VXNlcjYyMTA0OTQ1", "avatar_url": "https://avatars.githubusercontent.com/u/62104945?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zhongshsh", "html_url": "https://github.com/zhongshsh", "followers_url": "https://api.github.com/users/zhongshsh/followers", "following_url": "https://api.github.com/users/zhongshsh/following{/other_user}", "gists_url": "https://api.github.com/users/zhongshsh/gists{/gist_id}", "starred_url": "https://api.github.com/users/zhongshsh/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zhongshsh/subscriptions", "organizations_url": "https://api.github.com/users/zhongshsh/orgs", "repos_url": "https://api.github.com/users/zhongshsh/repos", "events_url": "https://api.github.com/users/zhongshsh/events{/privacy}", "received_events_url": "https://api.github.com/users/zhongshsh/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey thanks for reporting, can you upgrade to the newest version of transformers 🤗 ", "thx for your reply. After using `pip install -U transformers` or using `conda upgrade transformers`, same error still exists. here is the version info after the upgrade:\r\n```\r\n- `transformers` version: 4.36.2\r\n- Platform: Linux-5.15.0-18-shopee-generic-x86_64-with-glibc2.31\r\n- Python version: 3.10.13\r\n- Huggingface_hub version: 0.20.2\r\n- Safetensors version: 0.4.0\r\n- Accelerate version: 0.22.0\r\n- Accelerate config: not found\r\n- PyTorch version (GPU?): 2.0.1 (True)\r\n- Tensorflow version (GPU?): not installed (NA)\r\n- Flax version (CPU?/GPU?/TPU?): not installed (NA)\r\n- Jax version: not installed\r\n- JaxLib version: not installed\r\n- Using GPU in script?: <fill in>\r\n- Using distributed or parallel set-up in script?: <fill in>\r\n```", "The config class used seems to be `HFDeepSpeedConfig` vs `HfTrainerDeepSpeedConfig` (which inherits from the latter.) ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28395/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "talshaharabany", "id": 50660642, "node_id": "MDQ6VXNlcjUwNjYwNjQy", "avatar_url": "https://avatars.githubusercontent.com/u/50660642?v=4", "gravatar_id": "", "url": "https://api.github.com/users/talshaharabany", "html_url": "https://github.com/talshaharabany", "followers_url": "https://api.github.com/users/talshaharabany/followers", "following_url": "https://api.github.com/users/talshaharabany/following{/other_user}", "gists_url": "https://api.github.com/users/talshaharabany/gists{/gist_id}", "starred_url": "https://api.github.com/users/talshaharabany/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/talshaharabany/subscriptions", "organizations_url": "https://api.github.com/users/talshaharabany/orgs", "repos_url": "https://api.github.com/users/talshaharabany/repos", "events_url": "https://api.github.com/users/talshaharabany/events{/privacy}", "received_events_url": "https://api.github.com/users/talshaharabany/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28394/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28394", "html_url": "https://github.com/huggingface/transformers/pull/28394", "diff_url": "https://github.com/huggingface/transformers/pull/28394.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28394.patch", "merged_at": null }

{ "login": "andysingal", "id": 20493493, "node_id": "MDQ6VXNlcjIwNDkzNDkz", "avatar_url": "https://avatars.githubusercontent.com/u/20493493?v=4", "gravatar_id": "", "url": "https://api.github.com/users/andysingal", "html_url": "https://github.com/andysingal", "followers_url": "https://api.github.com/users/andysingal/followers", "following_url": "https://api.github.com/users/andysingal/following{/other_user}", "gists_url": "https://api.github.com/users/andysingal/gists{/gist_id}", "starred_url": "https://api.github.com/users/andysingal/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/andysingal/subscriptions", "organizations_url": "https://api.github.com/users/andysingal/orgs", "repos_url": "https://api.github.com/users/andysingal/repos", "events_url": "https://api.github.com/users/andysingal/events{/privacy}", "received_events_url": "https://api.github.com/users/andysingal/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This means your inputs are out of bound for the embedding matrix. Do you mind trying to explain your issue instead of just pointing to a collab notebook! 🤗 ", "> This means your inputs are out of bound for the embedding matrix. Do you mind trying to explain your issue instead of just pointing to a collab notebook! 🤗\r\n\r\n```\r\n# Datasets\r\ndataset = load_dataset(\"wikitext\", \"wikitext-2-raw-v1\")\r\ntokenizer = AutoTokenizer.from_pretrained(\"distilgpt2\")\r\ntokenizer.add_special_tokens({'pad_token': '[PAD]'})\r\n\r\n\r\n\r\ndef tokenize_function(examples):\r\n return tokenizer(examples[\"text\"])\r\n\r\ntokenized_datasets = dataset.map(\r\n tokenize_function, batched=True, num_proc=4, remove_columns=[\"text\"]\r\n)\r\n\r\ntokenized_datasets[\"train\"][1]\r\n\r\n```\r\n```\r\n{'input_ids': [796, 569, 18354, 7496, 17740, 6711, 796, 220, 198],\r\n 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1]}\r\n```\r\nProblem comes when i run:\r\n```\r\ndef compute_metrics(eval_pred):\r\n logits, labels = eval_pred\r\n predictions = np.argmax(logits, axis=-1)\r\n return metric.compute(predictions=predictions, references=labels)\r\n\r\ndef model_init():\r\n return AutoModelForSequenceClassification.from_pretrained(\r\n 'distilgpt2', return_dict=True)\r\n\r\nimport torch\r\nfrom transformers import TrainingArguments, Trainer\r\n# model = \"distilgpt2\"\r\ndevice = torch.device(\"cuda\") if torch.cuda.is_available() else torch.device(\"cpu\")\r\ntraining_args = TrainingArguments(\r\n\t\t output_dir=\"test_trainer\",\r\n\t\t evaluation_strategy=\"epoch\",\r\n\t\t report_to=\"wandb\")\r\n\r\ntrainer = Trainer(\r\n model_init=model_init,\r\n tokenizer=tokenizer,\r\n args=training_args,\r\n train_dataset=small_train_dataset,\r\n eval_dataset=small_eval_dataset,\r\n compute_metrics=compute_metrics,\r\n)\r\ntrainer.train()\r\n```\r\nwhich gives error:\r\n```\r\nSome weights of GPT2ForSequenceClassification were not initialized from the model checkpoint at distilgpt2 and are newly initialized: ['score.weight']\r\nYou should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\r\nSome weights of GPT2ForSequenceClassification were not initialized from the model checkpoint at distilgpt2 and are newly initialized: ['score.weight']\r\nYou should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\r\nYou're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding.\r\n---------------------------------------------------------------------------\r\nIndexError Traceback (most recent call last)\r\n<ipython-input-15-856d04b69f6c> in <cell line: 18>()\r\n 16 compute_metrics=compute_metrics,\r\n 17 )\r\n---> 18 trainer.train()\r\n\r\n13 frames\r\n/usr/local/lib/python3.10/dist-packages/torch/nn/functional.py in embedding(input, weight, padding_idx, max_norm, norm_type, scale_grad_by_freq, sparse)\r\n 2231 # remove once script supports set_grad_enabled\r\n 2232 _no_grad_embedding_renorm_(weight, input, max_norm, norm_type)\r\n-> 2233 return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)\r\n 2234 \r\n 2235 \r\n\r\nIndexError: index out of range in self\r\n```\r\nAny help to resolve is appreciated ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28393/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "susnato", "id": 56069179, "node_id": "MDQ6VXNlcjU2MDY5MTc5", "avatar_url": "https://avatars.githubusercontent.com/u/56069179?v=4", "gravatar_id": "", "url": "https://api.github.com/users/susnato", "html_url": "https://github.com/susnato", "followers_url": "https://api.github.com/users/susnato/followers", "following_url": "https://api.github.com/users/susnato/following{/other_user}", "gists_url": "https://api.github.com/users/susnato/gists{/gist_id}", "starred_url": "https://api.github.com/users/susnato/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/susnato/subscriptions", "organizations_url": "https://api.github.com/users/susnato/orgs", "repos_url": "https://api.github.com/users/susnato/repos", "events_url": "https://api.github.com/users/susnato/events{/privacy}", "received_events_url": "https://api.github.com/users/susnato/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Can you make it into a tip / warning for the mean time. ", "Hi @ArthurZucker, I have added the tip/warning. \r\n\r\nLet me know if anything more needs to be added.", "just need to rebase on main 😉 ", "Sorry my bad :sweat_smile: \r\n\r\nDone!", "Thanks! " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28392/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28392", "html_url": "https://github.com/huggingface/transformers/pull/28392", "diff_url": "https://github.com/huggingface/transformers/pull/28392.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28392.patch", "merged_at": 1704899268000 }

{ "login": "pacman100", "id": 13534540, "node_id": "MDQ6VXNlcjEzNTM0NTQw", "avatar_url": "https://avatars.githubusercontent.com/u/13534540?v=4", "gravatar_id": "", "url": "https://api.github.com/users/pacman100", "html_url": "https://github.com/pacman100", "followers_url": "https://api.github.com/users/pacman100/followers", "following_url": "https://api.github.com/users/pacman100/following{/other_user}", "gists_url": "https://api.github.com/users/pacman100/gists{/gist_id}", "starred_url": "https://api.github.com/users/pacman100/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/pacman100/subscriptions", "organizations_url": "https://api.github.com/users/pacman100/orgs", "repos_url": "https://api.github.com/users/pacman100/repos", "events_url": "https://api.github.com/users/pacman100/events{/privacy}", "received_events_url": "https://api.github.com/users/pacman100/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Closing as #28245 fixed it", "These are brutally expensive bugs for your users. What kind of test cases could be implemented to prevent this from happening?", "We added a new test that you can find here: https://github.com/huggingface/transformers/pull/28245/files#diff-712bbefe24deb3653111486b471859bd0f2b4d61b87a81845189da697edde29d ", "@ArthurZucker thank you :pray: " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28391/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yuanwu2017", "id": 34643241, "node_id": "MDQ6VXNlcjM0NjQzMjQx", "avatar_url": "https://avatars.githubusercontent.com/u/34643241?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yuanwu2017", "html_url": "https://github.com/yuanwu2017", "followers_url": "https://api.github.com/users/yuanwu2017/followers", "following_url": "https://api.github.com/users/yuanwu2017/following{/other_user}", "gists_url": "https://api.github.com/users/yuanwu2017/gists{/gist_id}", "starred_url": "https://api.github.com/users/yuanwu2017/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yuanwu2017/subscriptions", "organizations_url": "https://api.github.com/users/yuanwu2017/orgs", "repos_url": "https://api.github.com/users/yuanwu2017/repos", "events_url": "https://api.github.com/users/yuanwu2017/events{/privacy}", "received_events_url": "https://api.github.com/users/yuanwu2017/received_events", "type": "User", "site_admin": false }

[]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28390/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28390", "html_url": "https://github.com/huggingface/transformers/pull/28390", "diff_url": "https://github.com/huggingface/transformers/pull/28390.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28390.patch", "merged_at": null }

{ "login": "andysingal", "id": 20493493, "node_id": "MDQ6VXNlcjIwNDkzNDkz", "avatar_url": "https://avatars.githubusercontent.com/u/20493493?v=4", "gravatar_id": "", "url": "https://api.github.com/users/andysingal", "html_url": "https://github.com/andysingal", "followers_url": "https://api.github.com/users/andysingal/followers", "following_url": "https://api.github.com/users/andysingal/following{/other_user}", "gists_url": "https://api.github.com/users/andysingal/gists{/gist_id}", "starred_url": "https://api.github.com/users/andysingal/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/andysingal/subscriptions", "organizations_url": "https://api.github.com/users/andysingal/orgs", "repos_url": "https://api.github.com/users/andysingal/repos", "events_url": "https://api.github.com/users/andysingal/events{/privacy}", "received_events_url": "https://api.github.com/users/andysingal/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey, according to the documentation: \r\n> model ([`PreTrainedModel`] or `torch.nn.Module`, *optional*):\r\n The model to train, evaluate or use for predictions. If not provided, a `model_init` must be passed.\r\n\r\nyou are passing a string. ", "> Hey, according to the documentation:\r\n> \r\n> > ```\r\n> > model ([`PreTrainedModel`] or `torch.nn.Module`, *optional*):\r\n> > ```\r\n> \r\n> ```\r\n> The model to train, evaluate or use for predictions. If not provided, a `model_init` must be passed.\r\n> ```\r\n> \r\n> you are passing a string.\r\n\r\nThanks, where can i add: https://colab.research.google.com/drive/10JDBNsLlYrQdnI2FWfDK3F5M8wvVUDXG?usp=sharing ", "> > Hey, according to the documentation:\r\n> > > ```\r\n> > > model ([`PreTrainedModel`] or `torch.nn.Module`, *optional*):\r\n> > > ```\r\n> > \r\n> > \r\n> > ```\r\n> > The model to train, evaluate or use for predictions. If not provided, a `model_init` must be passed.\r\n> > ```\r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > \r\n> > you are passing a string.\r\n> \r\n> Thanks, where can i add: https://colab.research.google.com/drive/10JDBNsLlYrQdnI2FWfDK3F5M8wvVUDXG?usp=sharing\r\n\r\ni added: ```\r\ndef model_init():\r\n return AutoModelForSequenceClassification.from_pretrained( 'distilgpt2', return_dict=True)```\r\n\r\nbut got error:\r\n```\r\n---------------------------------------------------------------------------\r\nAssertionError Traceback (most recent call last)\r\n[<ipython-input-10-856d04b69f6c>](https://localhost:8080/#) in <cell line: 18>()\r\n 16 compute_metrics=compute_metrics,\r\n 17 )\r\n---> 18 trainer.train()\r\n\r\n6 frames\r\n[/usr/local/lib/python3.10/dist-packages/transformers/models/gpt2/modeling_gpt2.py](https://localhost:8080/#) in forward(self, input_ids, past_key_values, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, labels, use_cache, output_attentions, output_hidden_states, return_dict)\r\n 1446 \r\n 1447 assert (\r\n-> 1448 self.config.pad_token_id is not None or batch_size == 1\r\n 1449 ), \"Cannot handle batch sizes > 1 if no padding token is defined.\"\r\n 1450 if self.config.pad_token_id is None:\r\n\r\nAssertionError: Cannot handle batch sizes > 1 if no padding token is defined.\r\n```\r\nColab: https://colab.research.google.com/drive/10JDBNsLlYrQdnI2FWfDK3F5M8wvVUDXG?usp=sharing \r\n", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28389/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Bachstelze", "id": 19904888, "node_id": "MDQ6VXNlcjE5OTA0ODg4", "avatar_url": "https://avatars.githubusercontent.com/u/19904888?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Bachstelze", "html_url": "https://github.com/Bachstelze", "followers_url": "https://api.github.com/users/Bachstelze/followers", "following_url": "https://api.github.com/users/Bachstelze/following{/other_user}", "gists_url": "https://api.github.com/users/Bachstelze/gists{/gist_id}", "starred_url": "https://api.github.com/users/Bachstelze/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Bachstelze/subscriptions", "organizations_url": "https://api.github.com/users/Bachstelze/orgs", "repos_url": "https://api.github.com/users/Bachstelze/repos", "events_url": "https://api.github.com/users/Bachstelze/events{/privacy}", "received_events_url": "https://api.github.com/users/Bachstelze/received_events", "type": "User", "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" } ]

[]

[ "Deberta as shared encoder and decoder would be also good: https://github.com/huggingface/transformers/issues/12436" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28388/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "faaany", "id": 24477841, "node_id": "MDQ6VXNlcjI0NDc3ODQx", "avatar_url": "https://avatars.githubusercontent.com/u/24477841?v=4", "gravatar_id": "", "url": "https://api.github.com/users/faaany", "html_url": "https://github.com/faaany", "followers_url": "https://api.github.com/users/faaany/followers", "following_url": "https://api.github.com/users/faaany/following{/other_user}", "gists_url": "https://api.github.com/users/faaany/gists{/gist_id}", "starred_url": "https://api.github.com/users/faaany/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/faaany/subscriptions", "organizations_url": "https://api.github.com/users/faaany/orgs", "repos_url": "https://api.github.com/users/faaany/repos", "events_url": "https://api.github.com/users/faaany/events{/privacy}", "received_events_url": "https://api.github.com/users/faaany/received_events", "type": "User", "site_admin": false }

[]

[]

[ "@ArthurZucker would you take a look at this PR as well? Not quite sure who is the best reviewer for this case. Many thanks!", "Hi folks, any concern or feedback for this PR? ", "cc @abhilash1910", "Thanks @muellerzr for highlighting. As per our previous issue : #27716 (cc @ArthurZucker ) , this was causing an issue in ipex . @faaany I will sync with you offline to find a proper solution. Thanks", "#28442 might give a solution as well", "> #28442 might give a solution as well\r\n\r\nThanks for the review! I will have an offline meeting with @abhilash1910 and @zhuhong61 for an alignment and get back to you soon.\r\n", "@ArthurZucker we talked with each other offline today and decided to move forward with the solution proposed in this PR. \r\nPls give your comments as well, thanks a lot! @abhilash1910 @zhuhong61", "Hi @ArthurZucker We can close this PR https://github.com/huggingface/transformers/pull/28442, and use #28386 as the final solution. Thanks", "Thanks, does this need a specific version of torch / accelerate to work (which I would suspect as the patch was introduced )", "> Thanks, does this need a specific version of torch / accelerate to work (which I would suspect as the patch was introduced )\r\n\r\nyes, a patched version for pytorch is still needed for XPU currently.", "> Thanks, does this need a specific version of torch / accelerate to work (which I would suspect as the patch was introduced )\r\n\r\n@ArthurZucker, yes, for making XPU work, people need to follow this link https://intel.github.io/intel-extension-for-pytorch/index.html#installation?platform=gpu&version=v2.1.10%2Bxpu to set up env, it includes the installation intel oneCCL, intel patched PyTorch and Intel Extensions for PyTorch, these 3 support xpu.\r\n\r\nAnd Intel PyTorch team is also working with Meta to make xpu a built-in device of stock PyTorch. By that time, we can use XPU with stock PyTorch.", "Hi @muellerzr @ArthurZucker, this PR has been there for a while and we are also internally aligned with this solution. Do you guys still have any concerns? ", "Yes for sure! ", "> LGTM my only concern is that this can only work starting latest version of torch, thus we would need to keep the old solution for older versions. WDYT?\r\n\r\nHi @ArthurZucker , the old solution is no longer needed as the changes discussed here fall perfectly in line with the expectations from the xpu . Conclusively this implies that the code changes can be used as is ( as it was being used before with accelerate and others) . \r\nFor any breaking changes related to core pytorch which would require change of apis on xpu side, we would definitely raise a significant PR to address those. But as of now, this change is perfectly fine with the existing software stack. Thanks", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_28386). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "> LGTM my only concern is that this can only work starting latest version of torch, thus we would need to keep the old solution for older versions. WDYT?\r\n\r\nHmmm, I don't think people using older versions of torch would have problems with our current fix. In contrast, they would get the correct device set once our PR is merged. But anyway, thanks so much for the review and approval! ", "Hi @ArthurZucker, could you help merge this PR? Many thanks!", "Thanks for the explanations! " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28386/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28386", "html_url": "https://github.com/huggingface/transformers/pull/28386", "diff_url": "https://github.com/huggingface/transformers/pull/28386.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28386.patch", "merged_at": 1705667334000 }

{ "login": "zhentaocc", "id": 90437536, "node_id": "MDQ6VXNlcjkwNDM3NTM2", "avatar_url": "https://avatars.githubusercontent.com/u/90437536?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zhentaocc", "html_url": "https://github.com/zhentaocc", "followers_url": "https://api.github.com/users/zhentaocc/followers", "following_url": "https://api.github.com/users/zhentaocc/following{/other_user}", "gists_url": "https://api.github.com/users/zhentaocc/gists{/gist_id}", "starred_url": "https://api.github.com/users/zhentaocc/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zhentaocc/subscriptions", "organizations_url": "https://api.github.com/users/zhentaocc/orgs", "repos_url": "https://api.github.com/users/zhentaocc/repos", "events_url": "https://api.github.com/users/zhentaocc/events{/privacy}", "received_events_url": "https://api.github.com/users/zhentaocc/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Feel free to check https://github.com/huggingface/transformers/issues/25420#issuecomment-1775317535 for a better explanation ", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28385/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "denizyuret", "id": 1822118, "node_id": "MDQ6VXNlcjE4MjIxMTg=", "avatar_url": "https://avatars.githubusercontent.com/u/1822118?v=4", "gravatar_id": "", "url": "https://api.github.com/users/denizyuret", "html_url": "https://github.com/denizyuret", "followers_url": "https://api.github.com/users/denizyuret/followers", "following_url": "https://api.github.com/users/denizyuret/following{/other_user}", "gists_url": "https://api.github.com/users/denizyuret/gists{/gist_id}", "starred_url": "https://api.github.com/users/denizyuret/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/denizyuret/subscriptions", "organizations_url": "https://api.github.com/users/denizyuret/orgs", "repos_url": "https://api.github.com/users/denizyuret/repos", "events_url": "https://api.github.com/users/denizyuret/events{/privacy}", "received_events_url": "https://api.github.com/users/denizyuret/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Few things can come into play. By default the token will be normalized, this means that the normalizer will be adding a prefix space to that token. When decoding, that space is removed. You should add the token using `tokenizer.add_tokens(AddedToken(\"Deniz\", normalized = False, special=False))`", "This time there is an extra space that appears after the token:\r\n```\r\n>>> from transformers import AddedToken, AutoTokenizer\r\n>>> mtok = AutoTokenizer.from_pretrained('mistralai/Mistral-7B-v0.1')\r\n>>> mtok.add_tokens(AddedToken('Deniz', normalized=False, special=False))\r\n1\r\n>>> str = 'Arthur met Deniz today.'\r\n>>> mtok.decode(mtok.encode(str))\r\n'<s> Arthur met Deniz today.'\r\n```\r\n\r\nHere are the individual tokens if helps:\r\n```\r\n>>> mtok.convert_ids_to_tokens(mtok.encode(str))\r\n['<s>', '▁Arthur', '▁met', '▁', 'Deniz', '▁', '▁today', '.']\r\n```", "That is also expected, you should either use a slow tokenizer (`use_fast = False`) or follow the fix that is here: #26678. It's a known issue of the normalizer. ", "`use_fast=False` has an interesting behavior:\r\n```\r\n>>> mtok = AutoTokenizer.from_pretrained('mistralai/Mistral-7B-v0.1', use_fast=False)\r\n>>> mtok.add_tokens([AddedToken('Deniz', normalized=False, special=False)])\r\n>>> mtok.decode(mtok.encode('Arthur met Deniz today.'))\r\n'<s>Arthur met Deniz today.'\r\n>>> mtok.convert_ids_to_tokens(mtok.encode('Arthur met Deniz today.'))\r\n['<s>', '▁Arthur', '▁met', '▁', 'Deniz', '▁', '▁today', '.']\r\n```\r\ni.e. `decode` puts extra space before the added token, however `convert_ids_to_tokens` shows an extra space after the added token.\r\n\r\nI tried adding ' Deniz' or '▁Deniz' thinking it would better match the behavior of regular tokens like '▁Arthur' but no success.\r\n\r\nSo far I haven't been able to find a combination of options that will preserve spacing. Will check out your fix next.", "This is the combination that works:\r\n```python \r\nIn [44]: mtok = AutoTokenizer.from_pretrained('mistralai/Mistral-7B-v0.1', use_fast=False, legacy = False)\r\n\r\nIn [45]: mtok.tokenize('Arthur met Deniz today.')\r\nOut[45]: ['▁Arthur', '▁met', '▁Den', 'iz', '▁today', '.']\r\n\r\nIn [46]: mtok.add_tokens([AddedToken('Deniz', normalized=False, special=False)])\r\nOut[46]: 1\r\n\r\nIn [47]: mtok.tokenize('Arthur met Deniz today.')\r\nOut[47]: ['▁Arthur', '▁met', '▁', 'Deniz', '▁today', '.']\r\n```\u001b\r\n", "`mtok.decode` still throws in some extra spaces, but this works for model training etc. Thanks!", "This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.\n\nPlease note that issues that do not follow the [contributing guidelines](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md) are likely to be ignored." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28384/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "mahdip72", "id": 42680708, "node_id": "MDQ6VXNlcjQyNjgwNzA4", "avatar_url": "https://avatars.githubusercontent.com/u/42680708?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mahdip72", "html_url": "https://github.com/mahdip72", "followers_url": "https://api.github.com/users/mahdip72/followers", "following_url": "https://api.github.com/users/mahdip72/following{/other_user}", "gists_url": "https://api.github.com/users/mahdip72/gists{/gist_id}", "starred_url": "https://api.github.com/users/mahdip72/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mahdip72/subscriptions", "organizations_url": "https://api.github.com/users/mahdip72/orgs", "repos_url": "https://api.github.com/users/mahdip72/repos", "events_url": "https://api.github.com/users/mahdip72/events{/privacy}", "received_events_url": "https://api.github.com/users/mahdip72/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Seems like a bug with ESMTokenizer, (which doesn't use this library).\r\n\r\n@ArthurZucker for insights or the more relevant people ?", "Hey, I cannot reproduce this: \r\n```python \r\nIn [23]: model_checkpoint = \"facebook/esm2_t6_8M_UR50D\"\r\n ...: tokenizer_2 = AutoTokenizer.from_pretrained(model_checkpoint)\r\nhuggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...\r\nTo disable this warning, you can either:\r\n - Avoid using `tokenizers` before the fork if possible\r\n - Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)\r\ntokenizer_config.json: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 95.0/95.0 [00:00<00:00, 135kB/s]\r\nvocab.txt: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 93.0/93.0 [00:00<00:00, 247kB/s]\r\nspecial_tokens_map.json: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 125/125 [00:00<00:00, 416kB/s]\r\n\r\nIn [24]: tokenizer_2\r\nOut[24]: \r\nEsmTokenizer(name_or_path='facebook/esm2_t6_8M_UR50D', vocab_size=33, model_max_length=1024, is_fast=False, padding_side='right', truncation_side='right', special_tokens={'eos_token': '<eos>', 'unk_token': '<unk>', 'pad_token': '<pad>', 'cls_token': '<cls>', 'mask_token': '<mask>'}, clean_up_tokenization_spaces=True), added_tokens_decoder={\r\n 0: AddedToken(\"<cls>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 1: AddedToken(\"<pad>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 2: AddedToken(\"<eos>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 3: AddedToken(\"<unk>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 32: AddedToken(\"<mask>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n}\r\n```\r\n\r\n```python \r\n>>> tokenizer_2.add_tokens([\"J\"]) \r\nEsmTokenizer(name_or_path='facebook/esm2_t6_8M_UR50D', vocab_size=33, model_max_length=1024, is_fast=False, padding_side='right', truncation_side='right', special_tokens={'eos_token': '<eos>', 'unk_token': '<unk>', 'pad_token': '<pad>', 'cls_token': '<cls>', 'mask_token': '<mask>', 'additional_special_tokens': ['J']}, clean_up_tokenization_spaces=True), added_tokens_decoder={\r\n 0: AddedToken(\"<cls>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 1: AddedToken(\"<pad>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 2: AddedToken(\"<eos>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 3: AddedToken(\"<unk>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 32: AddedToken(\"<mask>\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n 33: AddedToken(\"J\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\r\n}\r\n```\r\n\r\n```python\r\nIn [29]: tokenizer_2.get_vocab()\r\nOut[29]: \r\n{'<cls>': 0,\r\n '<pad>': 1,\r\n '<eos>': 2,\r\n '<unk>': 3,\r\n 'L': 4,\r\n 'A': 5,\r\n 'G': 6,\r\n 'V': 7,\r\n 'S': 8,\r\n 'E': 9,\r\n 'R': 10,\r\n 'T': 11,\r\n 'I': 12,\r\n 'D': 13,\r\n 'P': 14,\r\n 'K': 15,\r\n 'Q': 16,\r\n 'N': 17,\r\n 'F': 18,\r\n 'Y': 19,\r\n 'M': 20,\r\n 'H': 21,\r\n 'W': 22,\r\n 'C': 23,\r\n 'X': 24,\r\n 'B': 25,\r\n 'U': 26,\r\n 'Z': 27,\r\n 'O': 28,\r\n '.': 29,\r\n '-': 30,\r\n '<null_1>': 31,\r\n '<mask>': 32}\r\n```", "> My main problem is that I noticed the length of the tokenizer does not change after adding the new token and therefore the above code does not extend the embeddings layer as expected.\r\n\r\n@ArthurZucker My problem is not with being a special token. When I am adding new tokens the vocab size does not change (33). Could you help me understand how to correctly increase the embedding size of the model?\r\n\r\nDoes it make sense if I define it manually?\r\n```python\r\nmodel_checkpoint = \"facebook/esm2_t6_8M_UR50D\"\r\nmodel = AutoModelForMaskedLM.from_pretrained(model_checkpoint)\r\ntokenizer = AutoTokenizer.from_pretrained(model_checkpoint)\r\nnum_added_toks = tokenizer.add_tokens(['J'])\r\nmodel.resize_token_embeddings(33 + num_added_toks)\r\n```", "If the token is already part of the vocab, it is expected that the vocab size will not change", "@ArthurZucker I am adding completely new tokens. I see them being added to the tokenizer. But the vocab size doesn't changed despite the fact that the new indexes are being set as the additional_special_tokens_ids.\r\nI bypassed the issue using the following line:\r\n\r\n```python\r\nmodel.resize_token_embeddings(max(tokenizer.additional_special_tokens_ids))\r\n```", "The length of the vocab is different from the max if you have holes in the vocab. This ESMTokenizer uses the length as number of tokens rather than the max! \r\nNice fix and not sure we should change it no? ", "@ArthurZucker @Narsil I fixed my problem, but others using ESM models might still have trouble. These models are very important for protein research now. The way the tokenizer counts words can confuse people when they try to make the model learn new tokens. This is different from the usual instruction of extending embedding layer such as llama 2 and could cause errors. Clearer steps in documentation or a fix in the tokenizer might help researchers.", "cc @Rocketknight1 we might want to update that? WDYT? \r\n@mahdip72 would you like to open a pr for doc fixes? ", "Hi all, I investigated the issue. There is indeed [specific code in the ESM tokenizer](https://github.com/huggingface/transformers/blob/main/src/transformers/models/esm/tokenization_esm.py#L161) that causes all new added tokens to be counted as 'special' tokens. I suspect the reason for this was that the authors felt the token list for proteins was constant (since it was just the list of amino acids), and therefore any new token had to be outside the normal vocabulary.\r\n\r\nIn your case @mahdip72, I'm guessing you want to add either nonstandard amino acids or tokens like `J` that represent \"leucine OR isoleucine\", correct? This is a valid use-case for ESM, and I think we should update the tokenizer code to support it. There is the issue of backward compatibility, though, so I see two possible solutions:\r\n\r\n1 (More backward compatible):\r\nUpdate `add_tokens` so that it keeps `special_tokens=True` as the default, but lets users manually specify `special_tokens=False` for cases like this\r\n\r\n2 (Matches workflows for other models):\r\nUpdate `add_tokens` so that `special_tokens=False` is the default, like other models. Users will need to manually specify `special_tokens=True` to add tokens as special tokens. This is probably a better solution, but it may break existing workflows.\r\n\r\nI'll see if I can grab a member of the ESM team to comment on this!", "> In your case @mahdip72, I'm guessing you want to add either nonstandard amino acids or tokens like J that represent \"leucine OR isoleucine\", correct? \r\n\r\nIt is correct. My goal is to add new non-separatable tokens like the ESM vocabs to the ESM tokenizer. Also, I have seen lots of folk are adding non-separable 3Di [fold seek](https://www.nature.com/articles/s41587-023-01773-0) tokens and/or chemical-related tokens such as [SELFIES](https://arxiv.org/abs/1905.13741) to the protein language models. As far as I am understand, these tokens are non-separable and constant, similar to amino acids tokens.\r\n\r\n@Rocketknight1 Are special tokens constant and inseparable? What is the difference between normal tokens and special tokens in the ESM tokenizer?", "Hi @mahdip72, the idea of \"special tokens\" mostly comes from tokenization for language models. In general, special tokens have two main properties:\r\n\r\n- Special tokens can be skipped when decoding using `skip_special_tokens = True`.\r\n- Special tokens are never split by the tokenizer.\r\n\r\nThese traits aren't especially relevant for ESM - in general, people aren't generating sequences with ESM and so tokenizer decoding doesn't apply, and secondly ESM never splits the text it tokenizes because it always converts one character to one token, unlike tokenizers like sentencepiece that are commonly used for natural language.\r\n\r\nI think the most sensible solution is to just update `add_tokens` for ESM so it behaves like other models and adds tokens as \"non-special\" by default, even though this might affect backward compatibility slightly. What do you think?", "@Rocketknight1 I Agree. A general solution similar to other models is more sensible.", "Hi @mahdip72, I've opened a PR at #28535 that should resolve this. Can you try it out and let me know if it resolves your issue? To install the PR branch, run this command: `pip install --upgrade git+https://github.com/huggingface/transformers.git@allow_esm_add_tokens`", "Thanks for looking into this Matt! Agree ignoring special_tokens and force `special_tokens=True` was problematic there.\r\n Makes sense to make the behavior consistent with the rest of huggingface ecosystem, although I'm a bit nervous about silently breaking backward compatibility for any users who were adding special tokens without explicilty specifying the flag special_tokens=True there", "Yeah @tomsercu - it was a bit of a concern for us too! I suspect adding tokens was quite niche, though - hopefully the improvement from standardization outweighs the very small number of people who were depending on the old behaviour.", "Agreed. Thanks for the fix Matt!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28387/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "PolarPeak", "id": 44831329, "node_id": "MDQ6VXNlcjQ0ODMxMzI5", "avatar_url": "https://avatars.githubusercontent.com/u/44831329?v=4", "gravatar_id": "", "url": "https://api.github.com/users/PolarPeak", "html_url": "https://github.com/PolarPeak", "followers_url": "https://api.github.com/users/PolarPeak/followers", "following_url": "https://api.github.com/users/PolarPeak/following{/other_user}", "gists_url": "https://api.github.com/users/PolarPeak/gists{/gist_id}", "starred_url": "https://api.github.com/users/PolarPeak/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/PolarPeak/subscriptions", "organizations_url": "https://api.github.com/users/PolarPeak/orgs", "repos_url": "https://api.github.com/users/PolarPeak/repos", "events_url": "https://api.github.com/users/PolarPeak/events{/privacy}", "received_events_url": "https://api.github.com/users/PolarPeak/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hey, I don't think Qwen was merged into transformers it's a code on the hub, you should open the issue directly on the model discussion page : https://huggingface.co./Qwen/Qwen-72B/discussions" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28383/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Chandler-Bing", "id": 29994840, "node_id": "MDQ6VXNlcjI5OTk0ODQw", "avatar_url": "https://avatars.githubusercontent.com/u/29994840?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Chandler-Bing", "html_url": "https://github.com/Chandler-Bing", "followers_url": "https://api.github.com/users/Chandler-Bing/followers", "following_url": "https://api.github.com/users/Chandler-Bing/following{/other_user}", "gists_url": "https://api.github.com/users/Chandler-Bing/gists{/gist_id}", "starred_url": "https://api.github.com/users/Chandler-Bing/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Chandler-Bing/subscriptions", "organizations_url": "https://api.github.com/users/Chandler-Bing/orgs", "repos_url": "https://api.github.com/users/Chandler-Bing/repos", "events_url": "https://api.github.com/users/Chandler-Bing/events{/privacy}", "received_events_url": "https://api.github.com/users/Chandler-Bing/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hello, this does not seem to be a direct issue, so would recommend you to ask on[ the forum ](https://discuss.huggingface.co/) 🤗 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28382/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "gmittal", "id": 2015126, "node_id": "MDQ6VXNlcjIwMTUxMjY=", "avatar_url": "https://avatars.githubusercontent.com/u/2015126?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gmittal", "html_url": "https://github.com/gmittal", "followers_url": "https://api.github.com/users/gmittal/followers", "following_url": "https://api.github.com/users/gmittal/following{/other_user}", "gists_url": "https://api.github.com/users/gmittal/gists{/gist_id}", "starred_url": "https://api.github.com/users/gmittal/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gmittal/subscriptions", "organizations_url": "https://api.github.com/users/gmittal/orgs", "repos_url": "https://api.github.com/users/gmittal/repos", "events_url": "https://api.github.com/users/gmittal/events{/privacy}", "received_events_url": "https://api.github.com/users/gmittal/received_events", "type": "User", "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" } ]

[]

[ "Hi, I would like to work on this issue", "Support for Phi-2 is still WIP, you can follow the progress here: https://github.com/huggingface/transformers/pull/28163", "Hi @gmittal, Flash Attention is already implemented for Phi, [PR](https://github.com/huggingface/transformers/pull/27661) \r\n\r\nIt seems that you are using the hub version of `phi-2`. Please use it from the library to properly enable Flash Attention.\r\nFor now `microsoft/phi-2`, does not have the correct order of the weights to be used with the library model so please use it from `susnato/phi-2`.\r\n\r\nFirst update to the latest transformers version - \r\n```\r\npip install -U transformers\r\n```\r\n\r\nthen run - \r\n\r\n```python\r\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\r\n\r\nmodel = AutoModelForCausalLM.from_pretrained(\"susnato/phi-2\", \r\n use_flash_attention_2=True, \r\n torch_dtype=torch.float16)\r\n\r\ntokenizer = AutoTokenizer.from_pretrained(\"susnato/phi-2\")\r\n\r\ninputs = tokenizer('''def print_prime(n):\r\n \"\"\"\r\n Print all primes between 1 and n\r\n \"\"\"''', return_tensors=\"pt\", return_attention_mask=False)\r\n\r\noutputs = model.generate(**inputs, max_length=200)\r\ntext = tokenizer.batch_decode(outputs)[0]\r\nprint(text)\r\n```\r\n\r\n\r\nLet me know if this works or not.", "I would like to work on this issue\r\n", "Using HF alignment notebook, DPO script gives me this error regardless of transformers version. (I already force updated with pip). When I remove flash attention from the yaml it works (after a bit of code adjustment). I am able to fine tune with one of my sft scripts using flash attention, which is the strange part.", "Hello everyone!\r\n\r\nThis should be fixed in transformers 4.37.0.dev. If not using that version, please make sure that `trust_remote_code=True` when loading the model and it should work out-of-the-box with flash-attention 2.", "Thanks! Closing as this was fixed in https://github.com/huggingface/transformers/pull/28163", "I installed from source so now i am on transformers 4.37.dev0 and i am still getting the Incompatible error, even with trust remote code set to true.\r\n\r\n`C:\\Users\\PC\\Documents\\Code-Trainer\\FineTune>py FINETUNERphiFP16.py --model_name_or_path C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\dolphin-2_6-phi-2 --data_path MiniCoderW.json --output_dir C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\TrainedPhi --num_train_epochs 3 --model_max_length 1024 --per_device_train_batch_size 1 --evaluation_strategy \"no\" --save_strategy \"steps\" --save_steps 1000 --save_total_limit 10 --learning_rate 2e-5 --warmup_steps 10 --logging_steps 10 --lr_scheduler_type \"cosine\" --report_to \"tensorboard\" --bf16 False --dataloader_num_workers 12 --optim paged_adamw_8bit\r\nWARNING:tensorflow:From C:\\Python311\\Lib\\site-packages\\keras\\src\\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.\r\n\r\n====================================================================================================\r\nTrainingArguments(\r\n_n_gpu=1,\r\nadafactor=False,\r\nadam_beta1=0.9,\r\nadam_beta2=0.999,\r\nadam_epsilon=1e-08,\r\nauto_find_batch_size=False,\r\nbf16=False,\r\nbf16_full_eval=False,\r\ncache_dir=None,\r\ndata_seed=None,\r\ndataloader_drop_last=False,\r\ndataloader_num_workers=12,\r\ndataloader_persistent_workers=False,\r\ndataloader_pin_memory=True,\r\nddp_backend=None,\r\nddp_broadcast_buffers=None,\r\nddp_bucket_cap_mb=None,\r\nddp_find_unused_parameters=None,\r\nddp_timeout=1800,\r\ndebug=[],\r\ndeepspeed=None,\r\ndisable_tqdm=False,\r\ndispatch_batches=None,\r\ndo_eval=False,\r\ndo_predict=False,\r\ndo_train=False,\r\neval_accumulation_steps=None,\r\neval_delay=0,\r\neval_steps=None,\r\nevaluation_strategy=IntervalStrategy.NO,\r\nfp16=False,\r\nfp16_backend=auto,\r\nfp16_full_eval=False,\r\nfp16_opt_level=O1,\r\nfsdp=[],\r\nfsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_grad_ckpt': False},\r\nfsdp_min_num_params=0,\r\nfsdp_transformer_layer_cls_to_wrap=None,\r\nfull_determinism=False,\r\ngradient_accumulation_steps=1,\r\ngradient_checkpointing=False,\r\ngradient_checkpointing_kwargs=None,\r\ngreater_is_better=None,\r\ngroup_by_length=False,\r\nhalf_precision_backend=auto,\r\nhub_always_push=False,\r\nhub_model_id=None,\r\nhub_private_repo=False,\r\nhub_strategy=HubStrategy.EVERY_SAVE,\r\nhub_token=<HUB_TOKEN>,\r\nignore_data_skip=False,\r\ninclude_inputs_for_metrics=False,\r\ninclude_num_input_tokens_seen=False,\r\ninclude_tokens_per_second=False,\r\njit_mode_eval=False,\r\nlabel_names=None,\r\nlabel_smoothing_factor=0.0,\r\nlearning_rate=2e-05,\r\nlength_column_name=length,\r\nload_best_model_at_end=False,\r\nlocal_rank=0,\r\nlog_level=passive,\r\nlog_level_replica=warning,\r\nlog_on_each_node=True,\r\nlogging_dir=C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\TrainedPhi\\runs\\Jan12_23-36-31_Nicolas,\r\nlogging_first_step=False,\r\nlogging_nan_inf_filter=True,\r\nlogging_steps=10,\r\nlogging_strategy=IntervalStrategy.STEPS,\r\nlr_scheduler_kwargs={},\r\nlr_scheduler_type=SchedulerType.COSINE,\r\nmax_grad_norm=1.0,\r\nmax_steps=-1,\r\nmetric_for_best_model=None,\r\nmodel_max_length=1024,\r\nmp_parameters=,\r\nneftune_noise_alpha=None,\r\nno_cuda=False,\r\nnum_train_epochs=3.0,\r\noptim=OptimizerNames.PAGED_ADAMW_8BIT,\r\noptim_args=None,\r\noutput_dir=C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\TrainedPhi,\r\noverwrite_output_dir=False,\r\npast_index=-1,\r\nper_device_eval_batch_size=8,\r\nper_device_train_batch_size=1,\r\nprediction_loss_only=False,\r\npush_to_hub=False,\r\npush_to_hub_model_id=None,\r\npush_to_hub_organization=None,\r\npush_to_hub_token=<PUSH_TO_HUB_TOKEN>,\r\nray_scope=last,\r\nremove_unused_columns=True,\r\nreport_to=['tensorboard'],\r\nresume_from_checkpoint=None,\r\nrun_name=C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\TrainedPhi,\r\nsave_on_each_node=False,\r\nsave_only_model=False,\r\nsave_safetensors=True,\r\nsave_steps=1000,\r\nsave_strategy=IntervalStrategy.STEPS,\r\nsave_total_limit=10,\r\nseed=42,\r\nskip_memory_metrics=True,\r\nsplit_batches=False,\r\ntf32=None,\r\ntorch_compile=False,\r\ntorch_compile_backend=None,\r\ntorch_compile_mode=None,\r\ntorchdynamo=None,\r\ntpu_metrics_debug=False,\r\ntpu_num_cores=None,\r\nuse_cpu=False,\r\nuse_ipex=False,\r\nuse_legacy_prediction_loop=False,\r\nuse_mps_device=False,\r\nwarmup_ratio=0.0,\r\nwarmup_steps=10,\r\nweight_decay=0.0,\r\n)\r\nSpecial tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\r\nPAD Token: <|endoftext|> 50256\r\nBOS Token <|endoftext|> 50256\r\nEOS Token <|im_end|> 50295\r\nLoad tokenizer from C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\dolphin-2_6-phi-2 over.\r\nTraceback (most recent call last):\r\n File \"C:\\Users\\PC\\Documents\\Code-Trainer\\FineTune\\FINETUNERphiFP16.py\", line 192, in <module>\r\n train()\r\n File \"C:\\Users\\PC\\Documents\\Code-Trainer\\FineTune\\FINETUNERphiFP16.py\", line 145, in train\r\n model = transformers.AutoModelForCausalLM.from_pretrained(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\models\\auto\\auto_factory.py\", line 561, in from_pretrained\r\n return model_class.from_pretrained(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 3497, in from_pretrained\r\n config = cls._autoset_attn_implementation(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 1340, in _autoset_attn_implementation\r\n cls._check_and_enable_flash_attn_2(\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 1420, in _check_and_enable_flash_attn_2\r\n raise ValueError(\r\nValueError: PhiForCausalLM does not support Flash Attention 2.0 yet. Please request to add support where the model is hosted, on its model hub page: https://huggingface.co./C:\\Users\\PC\\Documents\\NEWGEN\\text-generation-webui-main\\models\\dolphin-2_6-phi-2/discussions/new or in the Transformers GitHub repo: https://github.com/huggingface/transformers/issues/new\r\n`\r\n\r\n\r\nHere is the script I am using:\r\n\r\n`import copy\r\nimport random\r\nfrom dataclasses import dataclass, field\r\nfrom typing import Optional, Dict, Sequence\r\n\r\nimport torch\r\nimport transformers\r\nfrom transformers import Trainer\r\nfrom datasets import load_dataset\r\n\r\n\r\nIGNORE_INDEX = -100\r\nEOT_TOKEN = \"<|EOT|>\"\r\n\r\ndef build_instruction_prompt(instruction: str):\r\n return '''\r\nYou are an AI programming assistant, utilizing the DeepSeek Coder model, developed by DeepSeek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer.\r\n### Instruction:\r\n{}\r\n### Response:\r\n'''.format(instruction.strip()).lstrip()\r\n\r\n@dataclass\r\nclass ModelArguments:\r\n model_name_or_path: Optional[str] = field(default=\"deepseek-ai/deepseek-coder-6.7b-instruct\")\r\n\r\n@dataclass\r\nclass DataArguments:\r\n data_path: str = field(default=None, metadata={\"help\": \"Path to the training data.\"})\r\n\r\n\r\n@dataclass\r\nclass TrainingArguments(transformers.TrainingArguments):\r\n cache_dir: Optional[str] = field(default=None)\r\n optim: str = field(default=\"adamw_torch\")\r\n model_max_length: int = field(\r\n default=512,\r\n metadata={\"help\": \"Maximum sequence length. Sequences will be right padded (and possibly truncated).\"},\r\n )\r\n\r\ndef safe_save_model_for_hf_trainer(trainer: transformers.Trainer, output_dir: str):\r\n \"\"\"Collects the state dict and dump to disk.\"\"\"\r\n state_dict = trainer.model.state_dict()\r\n if trainer.args.should_save:\r\n cpu_state_dict = {key: value.cpu() for key, value in state_dict.items()}\r\n del state_dict\r\n trainer._save(output_dir, state_dict=cpu_state_dict) # noqa\r\n\r\n\r\ndef _tokenize_fn(strings: Sequence[str], tokenizer: transformers.PreTrainedTokenizer) -> Dict:\r\n \"\"\"Tokenize a list of strings.\"\"\"\r\n tokenized_list = [\r\n tokenizer(\r\n text,\r\n return_tensors=\"pt\",\r\n padding=\"longest\",\r\n max_length=tokenizer.model_max_length,\r\n truncation=True,\r\n )\r\n for text in strings\r\n ]\r\n\r\n input_ids = labels = [tokenized.input_ids[0] for tokenized in tokenized_list]\r\n input_ids_lens = labels_lens = [\r\n tokenized.input_ids.ne(tokenizer.pad_token_id).sum().item() for tokenized in tokenized_list\r\n ]\r\n\r\n return dict(\r\n input_ids=input_ids,\r\n labels=labels,\r\n input_ids_lens=input_ids_lens,\r\n labels_lens=labels_lens,\r\n )\r\n\r\n\r\ndef preprocess(\r\n sources: Sequence[str],\r\n targets: Sequence[str],\r\n tokenizer: transformers.PreTrainedTokenizer,\r\n) -> Dict:\r\n \"\"\"Preprocess the data by tokenizing.\"\"\"\r\n examples = [s + t for s, t in zip(sources, targets)]\r\n examples_tokenized, sources_tokenized = [_tokenize_fn(strings, tokenizer) for strings in (examples, sources)]\r\n input_ids = examples_tokenized[\"input_ids\"]\r\n\r\n labels = copy.deepcopy(input_ids)\r\n for label, source_len in zip(labels, sources_tokenized[\"input_ids_lens\"]):\r\n label[:source_len] = IGNORE_INDEX\r\n return dict(input_ids=input_ids, labels=labels)\r\n\r\n@dataclass\r\nclass DataCollatorForSupervisedDataset(object):\r\n \"\"\"Collate examples for supervised fine-tuning.\"\"\"\r\n tokenizer: transformers.PreTrainedTokenizer\r\n\r\n def __call__(self, instances: Sequence[Dict]) -> Dict[str, torch.Tensor]:\r\n input_ids, labels = tuple([instance[key] for instance in instances] for key in (\"input_ids\", \"labels\"))\r\n input_ids = [torch.tensor(x) for x in input_ids]\r\n input_ids = torch.nn.utils.rnn.pad_sequence(\r\n input_ids, batch_first=True, padding_value=self.tokenizer.pad_token_id\r\n )\r\n labels = [torch.tensor(x) for x in labels]\r\n labels = torch.nn.utils.rnn.pad_sequence(labels, batch_first=True, padding_value=IGNORE_INDEX)\r\n \r\n return dict(\r\n input_ids=input_ids,\r\n labels=labels,\r\n attention_mask=input_ids.ne(self.tokenizer.pad_token_id),\r\n )\r\n\r\ndef train_tokenize_function(examples, tokenizer):\r\n sources = [\r\n build_instruction_prompt(instruction)\r\n for instruction in examples['instruction']\r\n ]\r\n targets = [f\"{output}\\n{EOT_TOKEN}\" for output in examples['output']]\r\n data_dict = preprocess(sources, targets, tokenizer)\r\n return data_dict\r\n\r\ndef train():\r\n parser = transformers.HfArgumentParser((ModelArguments, DataArguments, TrainingArguments))\r\n model_args, data_args, training_args = parser.parse_args_into_dataclasses()\r\n \r\n if training_args.local_rank == 0:\r\n print('='*100)\r\n print(training_args)\r\n \r\n tokenizer = transformers.AutoTokenizer.from_pretrained(\r\n model_args.model_name_or_path,\r\n model_max_length=training_args.model_max_length,\r\n padding_side=\"right\",\r\n use_fast=True,\r\n trust_remote_code=True\r\n )\r\n if tokenizer.pad_token is None:\r\n tokenizer.add_special_tokens({'pad_token': '[PAD]'})\r\n\r\n print(\"PAD Token:\", tokenizer.pad_token, tokenizer.pad_token_id)\r\n print(\"BOS Token\", tokenizer.bos_token, tokenizer.bos_token_id)\r\n print(\"EOS Token\", tokenizer.eos_token, tokenizer.eos_token_id)\r\n\r\n if training_args.local_rank == 0:\r\n print(\"Load tokenizer from {} over.\".format(model_args.model_name_or_path))\r\n\r\n model = transformers.AutoModelForCausalLM.from_pretrained(\r\n model_args.model_name_or_path,\r\n torch_dtype=torch.bfloat16,\r\n trust_remote_code=True,\r\n attn_implementation=\"flash_attention_2\",\r\n )\r\n\r\n if training_args.local_rank == 0:\r\n print(\"Load model from {} over.\".format(model_args.model_name_or_path))\r\n\r\n\r\n raw_train_datasets = load_dataset(\r\n 'json',\r\n data_files=data_args.data_path,\r\n split=\"train\",\r\n cache_dir=training_args.cache_dir\r\n )\r\n \r\n train_dataset = raw_train_datasets.map(\r\n train_tokenize_function,\r\n batched=True,\r\n batch_size=3000,\r\n num_proc=32,\r\n remove_columns=raw_train_datasets.column_names,\r\n load_from_cache_file=True, # not args.overwrite_cache\r\n desc=\"Running Encoding\",\r\n fn_kwargs={ \"tokenizer\": tokenizer }\r\n )\r\n\r\n \r\n if training_args.local_rank == 0:\r\n print(\"Training dataset samples:\", len(train_dataset))\r\n for index in random.sample(range(len(train_dataset)), 3):\r\n print(f\"Sample {index} of the training set: {train_dataset[index]['input_ids']}, {train_dataset[index]['labels']}.\")\r\n print(f\"Sample {index} of the training set: {tokenizer.decode(list(train_dataset[index]['input_ids']))}.\")\r\n\r\n data_collator = DataCollatorForSupervisedDataset(tokenizer=tokenizer)\r\n data_module = dict(train_dataset=train_dataset, eval_dataset=None, data_collator=data_collator)\r\n\r\n trainer = Trainer(model=model, tokenizer=tokenizer, args=training_args, **data_module)\r\n\r\n trainer.train()\r\n trainer.save_state()\r\n safe_save_model_for_hf_trainer(trainer=trainer, output_dir=training_args.output_dir)\r\n\r\n\r\nif __name__ == \"__main__\":\r\n train()\r\n`\r\n", "Hi @NickWithBotronics if you set `trust_remote_code=True`, then the code from the hub is used (in case of microsoft/phi-2 that's defined [here](https://huggingface.co./microsoft/phi-2/blob/main/modeling_phi.py)), rather than [modeling_phi.py](https://github.com/huggingface/transformers/blob/main/src/transformers/models/phi/modeling_phi.py) defined natively in the Transformers library.\r\n\r\nHence it's recommended to convert the weights from the `microsoft/phi-2` repo to a native one, which will work with Flash Attention 2. One can leverage the [conversion script](https://github.com/huggingface/transformers/blob/main/src/transformers/models/phi/convert_phi_weights_to_hf.py) for that.\r\n\r\n@ArthurZucker should we host the converted phi-2 weights as part of the Microsoft organization? Cause currently one will get a lot of mismatched keys when doing the following:\r\n```\r\nimport torch\r\nfrom transformers import AutoModelForCausalLM\r\n\r\nmodel = AutoModelForCausalLM.from_pretrained(\r\n 'microsoft/phi-2',\r\n use_flash_attention_2=True,\r\n torch_dtype=torch.bfloat16,\r\n)\r\n```\r\ndue to the the model in Transformers using a single matrix for queries, keys and values wheras the code on the hub uses separate matrices.", "Thank you <3 !!!! that fixed that error(using the new modeling.py and converted hf format), now onto a new error that's due to my script I think?. :(\r\n'\r\nC:\\Python311\\Lib\\site-packages\\torch\\_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()\r\n return self.fget.__get__(instance, owner)()\r\nYou are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`.\r\nTraceback (most recent call last):\r\n File \"C:\\Users\\PC\\Documents\\Code-Trainer\\FineTune\\FINETUNERphiFP16.py\", line 192, in <module>\r\n train()\r\n File \"C:\\Users\\PC\\Documents\\Code-Trainer\\FineTune\\FINETUNERphiFP16.py\", line 145, in train\r\n model = transformers.AutoModelForCausalLM.from_pretrained(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\models\\auto\\auto_factory.py\", line 561, in from_pretrained\r\n return model_class.from_pretrained(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 3503, in from_pretrained\r\n model = cls(config, *model_args, **model_kwargs)\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 967, in __init__\r\n self.model = PhiModel(config)\r\n ^^^^^^^^^^^^^^^^\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 821, in __init__\r\n [PhiDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 821, in <listcomp>\r\n [PhiDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 629, in __init__\r\n self.self_attn = PHI_ATTENTION_CLASSES[config._attn_implementation](config, layer_idx=layer_idx)\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 412, in __init__\r\n super().__init__(*args, **kwargs)\r\n File \"C:\\Users\\PC\\.cache\\huggingface\\modules\\transformers_modules\\MiniPhi\\modeling_phi.py\", line 245, in __init__\r\n self.attention_dropout = config.attention_dropout\r\n ^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"C:\\Python311\\Lib\\site-packages\\transformers\\configuration_utils.py\", line 265, in __getattribute__\r\n return super().__getattribute__(key)\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\nAttributeError: 'PhiConfig' object has no attribute 'attention_dropout'\r\n\r\nC:\\Users\\PC\\Documents\\Code-Trainer\\FineTune>\r\n\"\r\n\r\n\r\nEdit: fixed it by downloading the latest: Generation_config.json, Config.json, Configuration_phi.py, and Modeling_phi.py", "while I got it working, the training loss was very wack. It started at 6 and went to 2 (after 3 epochs) but when I used the old config with out flash attention it was .6 to ~.29(also 3 epochs) same dataset same set up, same model. Just different config files and flash attention. I saw someone else experience the same thing on twitter.", "Can you open a seperate issue for this? With a reproducible snippet ", "Gotcha, I’ll move to this ticket #28488 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28381/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "abuelnasr0", "id": 64566340, "node_id": "MDQ6VXNlcjY0NTY2MzQw", "avatar_url": "https://avatars.githubusercontent.com/u/64566340?v=4", "gravatar_id": "", "url": "https://api.github.com/users/abuelnasr0", "html_url": "https://github.com/abuelnasr0", "followers_url": "https://api.github.com/users/abuelnasr0/followers", "following_url": "https://api.github.com/users/abuelnasr0/following{/other_user}", "gists_url": "https://api.github.com/users/abuelnasr0/gists{/gist_id}", "starred_url": "https://api.github.com/users/abuelnasr0/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/abuelnasr0/subscriptions", "organizations_url": "https://api.github.com/users/abuelnasr0/orgs", "repos_url": "https://api.github.com/users/abuelnasr0/repos", "events_url": "https://api.github.com/users/abuelnasr0/events{/privacy}", "received_events_url": "https://api.github.com/users/abuelnasr0/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28380/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28380", "html_url": "https://github.com/huggingface/transformers/pull/28380", "diff_url": "https://github.com/huggingface/transformers/pull/28380.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28380.patch", "merged_at": 1704706781000 }

{ "login": "NielsRogge", "id": 48327001, "node_id": "MDQ6VXNlcjQ4MzI3MDAx", "avatar_url": "https://avatars.githubusercontent.com/u/48327001?v=4", "gravatar_id": "", "url": "https://api.github.com/users/NielsRogge", "html_url": "https://github.com/NielsRogge", "followers_url": "https://api.github.com/users/NielsRogge/followers", "following_url": "https://api.github.com/users/NielsRogge/following{/other_user}", "gists_url": "https://api.github.com/users/NielsRogge/gists{/gist_id}", "starred_url": "https://api.github.com/users/NielsRogge/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/NielsRogge/subscriptions", "organizations_url": "https://api.github.com/users/NielsRogge/orgs", "repos_url": "https://api.github.com/users/NielsRogge/repos", "events_url": "https://api.github.com/users/NielsRogge/events{/privacy}", "received_events_url": "https://api.github.com/users/NielsRogge/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/28379/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/28379", "html_url": "https://github.com/huggingface/transformers/pull/28379", "diff_url": "https://github.com/huggingface/transformers/pull/28379.diff", "patch_url": "https://github.com/huggingface/transformers/pull/28379.patch", "merged_at": null }

{ "login": "borisdayma", "id": 715491, "node_id": "MDQ6VXNlcjcxNTQ5MQ==", "avatar_url": "https://avatars.githubusercontent.com/u/715491?v=4", "gravatar_id": "", "url": "https://api.github.com/users/borisdayma", "html_url": "https://github.com/borisdayma", "followers_url": "https://api.github.com/users/borisdayma/followers", "following_url": "https://api.github.com/users/borisdayma/following{/other_user}", "gists_url": "https://api.github.com/users/borisdayma/gists{/gist_id}", "starred_url": "https://api.github.com/users/borisdayma/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/borisdayma/subscriptions", "organizations_url": "https://api.github.com/users/borisdayma/orgs", "repos_url": "https://api.github.com/users/borisdayma/repos", "events_url": "https://api.github.com/users/borisdayma/events{/privacy}", "received_events_url": "https://api.github.com/users/borisdayma/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Running the same example listed in #28377, we get the correct output with EOS token present before the PAD tokens:\r\n`[\"</s><s><s>My friends are cool but they eat too many carbs. I love carbs. The carb-averse among us have a love for dessert on this one. I've got some amazing desserts. I've been a junkie for 10 years. I can't go straight after dinner, I've been eating for 4. I love it. That's why I made the TV show.</s><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad><pad>\"]`" ]