search

octo-models / octo

Octo is a transformer-based robot policy trained on a diverse mix of 800k robot trajectories.
https://octo-models.github.io/
MIT License
885 stars 166 forks source link

Can you provide finetuning ckpt #60

Closed zwbx closed 8 months ago

zwbx commented 8 months ago

Hi thanks your great work. I notice that you provide the finetune code, do you consider to directly provide the ckpt of finetuning example?

kpertsch commented 8 months ago

We don't provide checkpoints for the finetuned models since they are very specific to the setups they were finetuned in, so it would be very hard to reproduce these setups elsewhere. Best to collect a small dataset in your own setup and run the finetuning yourself!