Questions related to model weights and configuration.

[ZQpengyu]

This is a very excellent piece of work, but I encountered the following issues:

1. In run_llm.py at line 361, it prompts for input for "Experiment name:". What does this usually refer to?

2. At line 365, there is an if configs['peft']: condition, but 'peft' is not present in configs.

3. In encoder.py, on line 776, tactile_vificlip.load_state_dict(state_dict, strict=True) throws an error: Error(s) in loading state_dict for ViFiCLIP: Missing key(s) in state_dict: 'clip_model.text_model.VPT', 'clip_model.vision_model.encoder.layers.1.VPT_shallow_sensors.physiclear', ...

I would appreciate it if you could kindly respond. Thank you.

[SamsonYuBaiJian]

Hi, thank you for your interest in our work and catching these errors!

1. This refers to any custom name you want to use for the current run, to make it more identifiable for yourself. Feel free to leave it blank. There will be automatically added datetime and experiment detail identifiers too.
2. The configs is outdated, we will push the right one and let you know.
3. The model weights are incorrect and outdated, we will push the right one and let you know.

Thanks again!

[SamsonYuBaiJian]

I just pushed an update, can you try again?

2. The config file run.yaml has been updated.
3. The model weights are updated in the README.

Thank you!