Open Issues Need Help
View All on GitHubAI Summary: The user is encountering an `IsADirectoryError` during training with the `transformers` `Trainer` when `tqdm` is enabled. The error message is highly unusual, as it shows `tqdm`'s HTML progress bar output being misinterpreted as a file path, suggesting an unexpected interaction between the `Trainer`'s internal file operations (e.g., logging or checkpointing) and `tqdm`'s rich display output, likely in a notebook environment.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
AI Summary: This GitHub issue proposes improving the documentation for the `EncoderDecoderModel` in Hugging Face Transformers, as the current docs are outdated and cause user confusion. The task involves updating the existing `encoder-decoder.mdx` file to include a "How-to-guide" on creating, saving, and fine-tuning the model, along with a warning about correctly setting configuration values.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
AI Summary: The user requests the addition of DINOv3 to AutoBackbone, noting that DINOv2 is already included. They suggest DINOv3 could directly inherit from DINOv2 for ease of implementation and user convenience.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
AI Summary: The task is to investigate and resolve a performance issue in the Hugging Face Transformers library. The text generation pipeline is significantly slower when a large list of `bad_words_ids` is used. The solution requires profiling the code to identify the bottleneck (inefficient looping, tensor access, or slow regex) and optimizing the relevant section to improve performance.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.