Thomas Lips
Thomas Lips
So I ran with the transfer cube task now and also made some better plots. Pusht + diffusion:  correlation between eval success and val loss: 0.73 correlation between eval...
> @tlpss insertion is probably not a great task as the success rate is low even in the paper. Owing to that, I think it's quite noisy and susceptible to...
> @tlpss here's what I got for validation loss with ACT on Aloha Transfer Cube : > > > So the validation loss does indeed start high @alexander-soare And very...
@alexander-soare I think I'll close this one? For now my main question was if validation loss can be used as predictor for evaluation performance. My (premature) conlusion is that validation...
@aliberts @Cadene I'm still interested in this feature. If no one else is working on it, I might create a PR. Any suggestions or questionsbefore I start? I would implement...
@aliberts , one more ping? If not I will try to implement when I find some time
I would love to have this feature as well @armanhar
@brimoor (sorry for random tagging) There has been no response on this issue. I would still be very interested in this feature though and would be willing to contribute if...
@kevinzakka sorry to reopen this but I was wondering why this rotation is included in the base body? The base frame of the mujoco model is now not the actual...
@kevinzakka, I can make a PR if you want