dodox/dppo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

typo

Browse Source
This commit is contained in:
allenzren 2024-09-11 21:50:02 -04:00
parent f13eb203e1
commit f5a8da5719
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@ -159,7 +159,7 @@ To use DDIM fine-tuning, set `denoising_steps=100` in pre-training and set `mode
## Adding your own dataset/environment ## Adding your own dataset/environment
### Pre-training data ### Pre-training data
Pre-training script is at [`agent/pretrain/train_diffusion_agent.py`](agent/pretrain/train_diffusion_agent.py). The pre-training dataset [loader](agent/dataset/sequence.py) assumes a npz file containing numpy arrays `states`, `actions`, `images` (if using pixel) and `traj_length`, where `states` and `actions` have the shape of num_total_steps x obs_dim/act_dim, `images` num_total_steps x C (concatenated if multiple images) x H x W, and `traj_length` is a 1-D array for indexing across num_total_steps. Pre-training script is at [`agent/pretrain/train_diffusion_agent.py`](agent/pretrain/train_diffusion_agent.py). The pre-training dataset [loader](agent/dataset/sequence.py) assumes a npz file containing numpy arrays `states`, `actions`, `images` (if using pixel) and `traj_lengths`, where `states` and `actions` have the shape of num_total_steps x obs_dim/act_dim, `images` num_total_steps x C (concatenated if multiple images) x H x W, and `traj_lengths` is a 1-D array for indexing across num_total_steps.
<!-- One pre-processing example can be found at [`script/process_robomimic_dataset.py`](script/process_robomimic_dataset.py). --> <!-- One pre-processing example can be found at [`script/process_robomimic_dataset.py`](script/process_robomimic_dataset.py). -->
<!-- **Note:** The current implementation does not support loading history observations (only using observation at the current timestep). If needed, you can modify [here](agent/dataset/sequence.py#L130-L131). --> <!-- **Note:** The current implementation does not support loading history observations (only using observation at the current timestep). If needed, you can modify [here](agent/dataset/sequence.py#L130-L131). -->
@ -168,7 +168,7 @@ Pre-training script is at [`agent/pretrain/train_diffusion_agent.py`](agent/pret
In our experiments we did not use any observation from previous timesteps (state or pixel), but it is implemented. You can set `cond_steps=<num_state_obs_step>` (and `img_cond_steps=<num_img_obs_step>`, no larger than `cond_steps`) in pre-training, and set the same when fine-tuning the newly pre-trained policy. In our experiments we did not use any observation from previous timesteps (state or pixel), but it is implemented. You can set `cond_steps=<num_state_obs_step>` (and `img_cond_steps=<num_img_obs_step>`, no larger than `cond_steps`) in pre-training, and set the same when fine-tuning the newly pre-trained policy.
### Fine-tuning environment ### Fine-tuning environment
We follow the Gym format for interacting with the environments. The vectorized environments are initialized at [make_async](env/gym_utils/__init__.py#L10) (called in the parent fine-tuning agent class [here](agent/finetune/train_agent.py#L38-L39)). The current implementation is not the cleanest as we tried to make it compatible with Gym, Robomimic, Furniture-Bench, and D3IL environments, but it should be easy to modify and allow using other environments. We use [multi_step](env/gym_utils/wrapper/multi_step.py) wrapper for history observations (not used currently) and multi-environment-step action execution. We also use environment-specific wrappers such as [robomimic_lowdim](env/gym_utils/wrapper/robomimic_lowdim.py) and [furniture](env/gym_utils/wrapper/furniture.py) for observation/action normalization, etc. You can implement a new environment wrapper if needed. We follow the Gym format for interacting with the environments. The vectorized environments are initialized at [make_async](env/gym_utils/__init__.py#L10) (called in the parent fine-tuning agent class [here](agent/finetune/train_agent.py#L38-L39)). The current implementation is not the cleanest as we tried to make it compatible with Gym, Robomimic, Furniture-Bench, and D3IL environments, but it should be easy to modify and allow using other environments. We use [multi_step](env/gym_utils/wrapper/multi_step.py) wrapper for history observations and multi-environment-step action execution. We also use environment-specific wrappers such as [robomimic_lowdim](env/gym_utils/wrapper/robomimic_lowdim.py) and [furniture](env/gym_utils/wrapper/furniture.py) for observation/action normalization, etc. You can implement a new environment wrapper if needed.
## Known issues ## Known issues
* IsaacGym simulation can become unstable at times and lead to NaN observations in Furniture-Bench. The current env wrapper does not handle NaN observations. * IsaacGym simulation can become unstable at times and lead to NaN observations in Furniture-Bench. The current env wrapper does not handle NaN observations.