96 lines
3.8 KiB
YAML
96 lines
3.8 KiB
YAML
defaults:
|
|
- _self_
|
|
- env: pusht
|
|
- policy: diffusion
|
|
|
|
hydra:
|
|
run:
|
|
# Set `dir` to where you would like to save all of the run outputs. If you run another training session
|
|
# with the same value for `dir` its contents will be overwritten unless you set `resume` to true.
|
|
dir: outputs/train/${now:%Y-%m-%d}/${now:%H-%M-%S}_${env.name}_${policy.name}_${hydra.job.name}
|
|
job:
|
|
name: default
|
|
|
|
# Set `resume` to true to resume a previous run. In order for this to work, you will need to make sure
|
|
# `hydra.run.dir` is the directory of an existing run with at least one checkpoint in it.
|
|
# Note that when resuming a run, the default behavior is to use the configuration from the checkpoint,
|
|
# regardless of what's provided with the training command at the time of resumption.
|
|
resume: false
|
|
device: cuda # cpu
|
|
# `use_amp` determines whether to use Automatic Mixed Precision (AMP) for training and evaluation. With AMP,
|
|
# automatic gradient scaling is used.
|
|
use_amp: false
|
|
# `seed` is used for training (eg: model initialization, dataset shuffling)
|
|
# AND for the evaluation environments.
|
|
seed: ???
|
|
# You may provide a list of datasets here. `train.py` creates them all and concatenates them. Note: only data
|
|
# keys common between the datasets are kept. Each dataset gets and additional transform that inserts the
|
|
# "dataset_index" into the returned item. The index mapping is made according to the order in which the
|
|
# datsets are provided.
|
|
dataset_repo_id: lerobot/pusht
|
|
video_backend: pyav
|
|
|
|
training:
|
|
offline_steps: ???
|
|
# NOTE: `online_steps` is not implemented yet. It's here as a placeholder.
|
|
online_steps: ???
|
|
online_steps_between_rollouts: ???
|
|
online_sampling_ratio: 0.5
|
|
# `online_env_seed` is used for environments for online training data rollouts.
|
|
online_env_seed: ???
|
|
eval_freq: ???
|
|
log_freq: 200
|
|
save_checkpoint: true
|
|
# Checkpoint is saved every `save_freq` training iterations and after the last training step.
|
|
save_freq: ???
|
|
num_workers: 4
|
|
batch_size: ???
|
|
image_transforms:
|
|
# These transforms are all using standard torchvision.transforms.v2
|
|
# You can find out how these transformations affect images here:
|
|
# https://pytorch.org/vision/0.18/auto_examples/transforms/plot_transforms_illustrations.html
|
|
# We use a custom RandomSubsetApply container to sample them.
|
|
# For each transform, the following parameters are available:
|
|
# weight: This represents the multinomial probability (with no replacement)
|
|
# used for sampling the transform. If the sum of the weights is not 1,
|
|
# they will be normalized.
|
|
# min_max: Lower & upper bound respectively used for sampling the transform's parameter
|
|
# (following uniform distribution) when it's applied.
|
|
# Set this flag to `true` to enable transforms during training
|
|
enable: false
|
|
# This is the maximum number of transforms (sampled from these below) that will be applied to each frame.
|
|
# It's an integer in the interval [1, number of available transforms].
|
|
max_num_transforms: 3
|
|
# By default, transforms are applied in Torchvision's suggested order (shown below).
|
|
# Set this to True to apply them in a random order.
|
|
random_order: false
|
|
brightness:
|
|
weight: 1
|
|
min_max: [0.8, 1.2]
|
|
contrast:
|
|
weight: 1
|
|
min_max: [0.8, 1.2]
|
|
saturation:
|
|
weight: 1
|
|
min_max: [0.5, 1.5]
|
|
hue:
|
|
weight: 1
|
|
min_max: [-0.05, 0.05]
|
|
sharpness:
|
|
weight: 1
|
|
min_max: [0.8, 1.2]
|
|
|
|
eval:
|
|
n_episodes: 1
|
|
# `batch_size` specifies the number of environments to use in a gym.vector.VectorEnv.
|
|
batch_size: 1
|
|
# `use_async_envs` specifies whether to use asynchronous environments (multiprocessing).
|
|
use_async_envs: false
|
|
|
|
wandb:
|
|
enable: false
|
|
# Set to true to disable saving an artifact despite save_checkpoint == True
|
|
disable_artifact: false
|
|
project: lerobot
|
|
notes: ""
|