sambal shikhar

Training on custom dataset

3

What are some things to consider when training on my own dataset ?

How to get the labels from the data

1

How to get the labels for the image , which one is defective and which is not.How to read the split file.

As mentioned in the paper as training is being done on random patches of size 256x256,512x512 , that is achieved using RandomCropping.However for validation what is image size that comes...

How to retrieve the adjacency matrix

2

I want to interpret the results by looking at the learned graph adjacency matrix .What is the right way to do it ?

While calculating AUC , the abnormal pixel are filtered with mask value of 0 , while in the agrivision data the anomaly pixels are masked with value of 1, am I missing something

1

https://github.com/Jingtao-Li-CVer/ASD/blob/a00d10fff3bd1ad5b4a1dca7e5745bee1c098b3e/metrics/auroc.py#L38

Natural language tasks

1

Hi , i am trying to evaluate the natural language task on mednli and summerisation. However the inference gives out results that do not make sense.I guess the checkpoints for...

Hello , this is a great compilation . Can you add LLMVoX to this compilation - https://mbzuai-oryx.github.io/LLMVoX/