About the pth file

[Kyouka77]

Hello, I would like to know:
1.How to run the pth file trained by train.py. Do I need to write a new running script or use the existing ones? If it's the existing ones, which one should I use?
2.Are the files "chip_rewarded_ppo", "chip_rewarded_sac", and "transformer" in the "chip" folder respectively corresponding to the vlm-rl, sac, and resolve methods? Because I saw a comparison of three methods on the website.
3. My memory is insufficient. Can I use the smaller open_chip_pytorch_model instead? Is that okay?
Thank you!

[zihaosheng]

1.How to run the pth file trained by train.py. Do I need to write a new running script or use the existing ones? If it's the existing ones, which one should I use?

The eval script is provided:

VLM-RL/README.md

Line 136 in 26575ed

## 📊 Evaluation <a name="evaluation"></a>

2.Are the files "chip_rewarded_ppo", "chip_rewarded_sac", and "transformer" in the "chip" folder respectively corresponding to the vlm-rl, sac, and resolve methods? Because I saw a comparison of three methods on the website.

clip_rewarded_ppo is the variant version of VLM-RL with PPO algorithm.

You can find the config of baseline models in config.py

3. My memory is insufficient. Can I use the smaller open_chip_pytorch_model instead? Is that okay?
   Thank you!

Sure, you can do this for debugging or comparison, but the performance could degrade.