spacegoing comments

Results 37 comments of


                                            spacegoing

RNN+PPO bug in test/continuous/test_ppo.py

Hi @Caopeng17 , I run into this bug before but forgot the reason. Yet here is my LSTM version adapted from `test_ppo.py`: https://github.com/spacegoing/SwitcherEnv/blob/main/switcher_learner.py. Feel free to check out~

`compute_episodic_return` bug when `v_s=None`

@MischaPanch Good to know! Actually I finally figured it out. This IS NOT a bug but is a design failure. I really hope it could be updated in the document...

`compute_episodic_return` bug when `v_s=None`

Sorry. That one was a misleading comment LoL. I run 10 runs for each config and there wasn't much difference. I guess that was an unlucky v.s. lucky seed case

`compute_episodic_return` bug when `v_s=None`

Much appreciated bro! Many thanks for your effort!

Enabling the semantic layer shows always `invalid` on the top bar.

> @lebensterben Thanks for the hint, and I found this string of whole line belongs to `lsp-headerline-breadcrumb-enable`, just set it to `nil` if anyone wants to hide it, and install...

about inference

same issue here, img vid aud tokens are constantly blank

After finetune the model, inference still get noise.

@wtjiang98 genius

After finetune the model, inference still get noise.

> > @wtjiang98 genius > > Still don't know how to solve the problem. Should we modify the inference code? Ah I see. I thought u solved the issue LoL....

After finetune the model, inference still get noise.

> > > @wtjiang98 genius > > > > > > Still don't know how to solve the problem. Should we modify the inference code? > > Ah I see....

After finetune the model, inference still get noise.

> For fine-tuning, we offer the following suggestions: > > 1. Reduce the learning rate. We recommend a learning rate of 1e-5 to 1e-6 for fine-tuning. > 2. If there...