You Liang Tan

Could you also remove the --no-tune_diffusion_model, this will tune the entire expert head too. Also you can also use a larger action horizon too by configuring the data_config.py

Yes! You should use your own dataset for posttraining. The unfortunate situation is that the so101 comes with their own calibration file. Technically if all robots are "calibrated" and "tuned" correctly, with the same camera configurations, it will transfer.