Kai Zhang

@mmhzheng Thanks!

Hi @nghiemkythu I'm not certain if I've encountered a similar issue before, but I think the problem may not be related to the length of the input because I have...

The code is executable using the modified version of Fairseq provided in this repository, not the vanilla Fairseq. Therefore, simply installing Fairseq will not work. Please first uninstall Fairseq, ensure...

I have converted BiomedGPT-B to be compatible with the Transformers library. You can refer to this Colab notebook for more details: [link](https://colab.research.google.com/drive/1AMG-OwmDpnu24a9ZvCNvZi3BZwb3nSfS?usp=sharing). You can run the example directly.

The current checkpoints are pre-trained or fine-tuned on vision-language tasks and are not suitable for language-specific tasks. I will upload checkpoints trained on open-source datasets, such as summarization tasks based...

I am unable to identify the exact cause of the error log provided, but I suspect a few potential reasons: (1) the dataset used for training may be too large,...

@zh-zhang1984 I am not familiar with running code on Apple’s chips, so I’m unsure how to resolve this issue. Could you try running the code on a single computing node...

The scores variable is used to print the evaluation metrics, while the results variable stores the model’s predictions alongside the ground truth in the test_predict.json file. The scores should not...

**------ what is the 224*224 what you mean?** I understand. I suppose you are using MedMNIST data in the beginning, which has a default resolution of 28x28, it's worth noting...

Hi the changes of 224*224 MedMNIST data is not matter for your custom data set, and will not influence any implementation in current codebase.