Katelyn Morrison

icon indicating a website link katelyn98.github.io icon indicating an email [email protected]

icon company Carnegie Mellon University icon location CMU HCII PhD Student working on Human-AI Collaboration, Explainable AI, and Interactive Decision-Making Tools

Results 2 comments of Katelyn Morrison

on eval mode, ground truth labels are concatenated with image_features!

The ids are only used to generate the results dictionary in eval mode, the ids are not being passed into the model.generate_sentence call. See line [here](https://github.com/salaniz/pytorch-gve-lrcn/blob/d1d9d29a96ce803cfb350ae77597434e67fefb57/train/lrcn_trainer.py#L102) in lcrn_trainer.py

on eval mode, ground truth labels are concatenated with image_features!

Sorry, nevermind...you are right. The ground truth labels are appended to the image features and passed into the evaluation....not sure where to go from here then...