search

huggingface / autotrain-advanced

🤗 AutoTrain Advanced
https://huggingface.co/autotrain
Apache License 2.0
3.84k stars 471 forks source link

Image Classification Training Issue with Hugging Face's AutoTrain #660

Closed emidiosouza closed 4 months ago

emidiosouza commented 4 months ago

I am encountering an issue while trying to train an image classification model using Hugging Face's AutoTrain. I filled out the form with the following settings:

Project Name: train_demo Base Model: google/vit-base-patch16-224 Dataset Source: Hugging Face Hub Hub dataset path: food101 Train split: train Valid split: left blank Column Mapping: image: image, label: label Parameters: Mixed precision, fp16; Optimizer: adamw_torch; Scheduler: linear; Batch size: 8; Epochs: 3; Gradient accumulation: 1; Learning rate: 0.00005 Accelerators: 1 However, when I submit the form, I receive an error indicating that the 'NoneType' object is not subscriptable. I have checked the dataset and file format, and everything seems correct. Could you help me diagnose and resolve this issue? I have provided the relevant details and would appreciate any guidance.

abhishekkrthakur commented 4 months ago

thanks for reporting. there was a small issue after ui redesign and it has been fixed. please factory rebuild your training space and make sure you are on version 0.7.113 or above. please let me know if you are still facing the issue.