feature(xyy):add HPT model to implement PolicyStem+DuelingHead #841
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
puyuan1996
reviewed
Nov 29, 2024
ding/model/template/__init__.py
Outdated
| @@ -24,6 +24,8 @@ | |||
| from .vae import VanillaVAE | |||
| from .decision_transformer import DecisionTransformer | |||
| from .procedure_cloning import ProcedureCloningMCTS, ProcedureCloningBFS | |||
| from .hpt import HPT | |||
|
|
|||
ding/model/template/hpt.py
Outdated
|
|
||
|
|
||
| class PolicyStem(nn.Module): | ||
| """policy stem |
There was a problem hiding this comment.
reformat the docstring as the DI-engine style
| create_config = lunarlander_hpt_create_config | ||
|
|
||
| if __name__ == "__main__": | ||
| # or you can enter `ding -m serial -c lunarlander_dqn_config.py -s 0` |
| import_names=['dizoo.box2d.lunarlander.envs.lunarlander_env'], | ||
| ), | ||
| env_manager=dict(type='subprocess'), | ||
| # env_manager=dict(type='base'), |
ding/model/template/hpt.py
Outdated
| @MODEL_REGISTRY.register('hpt') | ||
| class HPT(nn.Module): | ||
|
|
||
| def __init__(self, state_dim, action_dim): |
There was a problem hiding this comment.
add overview and related introduction
There was a problem hiding this comment.
add unittest like other template in DI-engine
| policy=dict( | ||
| # Whether to use cuda for network. | ||
| cuda=True, | ||
| load_path="./lunarlander_hpt_seed0/ckpt/ckpt_best.pth.tar", |
PaParaZz1
requested changes
Dec 6, 2024
| - mask (:obj:`torch.Tensor`, optional): The attention mask tensor. Defaults to None. | ||
|
|
||
| Returns: | ||
| - torch.Tensor: The output tensor after applying attention. |
| from ding.torch_utils import is_differentiable | ||
|
|
||
| T, B = 3, 4 | ||
| obs_shape = [4, (8, ), (4, 64, 64)] |
There was a problem hiding this comment.
Remove the group (4, 64, 64) here, the current HPT model you implemented here can't support the image input like (4, 64, 64). It need a CNN feature extractor, and the corresponding state_dim should be `(4, 64, 64) rather than the current scalar. We will left this part for future, thus you can remove this group now.
PaParaZz1
approved these changes
Dec 8, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Here are some tensorboard plots from the lunarlander_hpt_example.py run.
Related Issue
TODO
Check List