Skip to content

WIP: Torch upgrade. Aux loss. Vision backbone networks. #304

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
ViktorM wants to merge 35 commits into
base: master
Choose a base branch
from
Open

WIP: Torch upgrade. Aux loss. Vision backbone networks. #304

ViktorM wants to merge 35 commits into from

Conversation

@ViktorM
Copy link
Collaborator

@ViktorM ViktorM commented Sep 9, 2024

No description provided.

ViktorM and others added 30 commits May 21, 2024 14:12
@ViktorM
Updated to torch 2. Compile without max-autotune is working.
c41212b
@ViktorM
Fixed minibatch_per_env bug.
6a91bd3
@ViktorM
Merge from master.
46c8c48
@ViktorM
Merge branch 'master' into VM/torch_upgrade
ae043b9
@ViktorM
WIP: adding proprio observations to resnet network.
1e58c1e
@ViktorM
Added pretrained resnet e2e network. Fixes.
80f43f7
@ViktorM
Clean up.
7ea73fa
@ViktorM
Temporary moved vision_actor_critic back to network_builder.
2c285bd
@ViktorM
E2e training is working with Impala network and proprieception.
ec4e4f1
@ViktorM
Fixed input dictinary observation shapes for resnet network. Fixed amp.
04d653a
@ViktorM
Added support for more visual backbones.
637d2bc
@ViktorM
WIP: network with different vision backbones.
0dc5b43
@ViktorM
Vision backbone improvements.
eeae30e
@ViktorM
Resnet18 works!
006fe2c
@ViktorM
Added convnext and vit backbones support. Added preprocessing.
61d1bc4
@ViktorM
Fixed observation dict.
c1eeeba
@ViktorM
Pacman training.
3531fb8
@ViktorM
Refactored vision backbone networks. Maniskill first pass.
d073f4d
@ViktorM
Fixed Maniskill state-based training.
ead882c
@ViktorM
Fixed Maniskill resets. Push and pick configs.
8be57aa
@ViktorM
Maniskill RGB and Depth only observations. Added resnet and impala cu…
d21fd8f 
…be pickup configs.
@ViktorM
Fix.
dc4e279
@ViktorM
WIP: norm layer with Impala for pick cube vision training.
a78caac
@DenSumy
added test aux_loss
2fe6c5f
@ViktorM
Layer norm for vision model. Better maniskill training configs.
8b274d1
@ViktorM
Merge with aux_loss branch.
00cbd3d
@ViktorM
Aux loss.
4d29423
@ViktorM
Merge from master.
b3467f2
@ViktorM
Auxilarry losses visual training WIP. Maniskill env switched to using…
80ee4df 
… uint8 for RGBD envs.
@ViktorM
Aux loss is now optional. Confif fix.
39e0ff8
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@ViktorM @DenSumy