I am currently working on training a model using a dataset from Roboflow, specifically the Manga Translator Detection dataset available at
To ensure the best possible results and maintain consistency with the original model, I would like to replicate the training conditions as closely as possible. Could anyone provide insights into the following details for the above-mentioned model if possible?
Model Architecture and Parameters: Details regarding the neural network architecture, input size, and other relevant parameters.
Preprocessing Steps: Information on any preprocessing steps applied to the dataset before training, such as image normalization, resizing, etc.
Data Augmentation Techniques: Specific augmentations used during training to enhance model robustness.
Training Hyperparameters: Insights into the learning rate, batch size, number of epochs, and other hyperparameters.
Your assistance in this matter would be greatly appreciated as it would help me ensure the accuracy and effectiveness of my model.
This dataset was a YOLOv8 model, with v6 being YOLOv8m.
In the future, you can see the model type yourself, which is listed under each Universe model page. If you click on the “View Version” link next to the listing for the dataset it was trained on, you will be able to find the dataset version which will have the preprocessing and augmentation steps.
Could you explain more about what you are having trouble with? Some information that might be helpful is: what are you trying to accomplish, what was the expected behavior, as well as any screenshots
