No project description provided
Project description
Torch Lure
Installations
pip install torchlure
Usage
import torchlure as lure
# Optimizers
lure.SophiaG(lr=1e-3, weight_decay=0.2)
# Functions
lure.tanh_exp(x)
lure.TanhExp()
lure.quantile_loss(y_pred, y_target, quantile=0.5)
lure.QuantileLoss(quantile=0.5)
lure.RMSNrom(dim=256, eps=1e-6)
# Noise Scheduler
lure.LinearNoiseScheduler(beta=1e-4, beta_end=0.02, num_timesteps=1000)
lure.CosineNoiseScheduler(max_beta=0.999, s=0.008, num_timesteps=1000):
Dataset
import gymnasium as gym
import numpy as np
import torch
from torchlure.datasets import MinariEpisodeDataset, MinariTrajectoryDataset
env = gym.make("Hopper-v4")
minari_dataset = MinariEpisodeDataset("Hopper-v4.2407")
minari_dataset.create(env, n_episodes=100, exist_ok=True)
minari_dataset.info()
# Observation space: Box(-inf, inf, (11,), float64)
# Action space: Box(-1.0, 1.0, (3,), float32)
# Total episodes: 100
# Total steps: 2,182
traj_dataset = MinariTrajectoryDataset(minari_dataset, traj_len=20)
traj = traj_dataset[2]
traj = traj_dataset[[3, 8, 15]]
traj = traj_dataset[np.arange(16)]
traj = traj_dataset[torch.arange(16)]
traj = traj_dataset[-16:]
traj["observations"].shape, traj["actions"].shape, traj["rewards"].shape, traj[
"terminated"
].shape, traj["truncated"].shape, traj["timesteps"].shape
# (torch.Size([16, 20, 4, 4, 16]),
# torch.Size([16, 20]),
# torch.Size([16, 20]),
# torch.Size([16, 20]),
# torch.Size([16, 20]),
# torch.Size([16, 20]))
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
torchlure-0.2407.7.tar.gz
(19.1 kB
view hashes)
Built Distribution
Close
Hashes for torchlure-0.2407.7-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | b784eba113765c7202cd9689d24a53c0cae5ce7e8b01e4f8bdfd86c8241d80f9 |
|
MD5 | 5868255c3f0ed4b262a9000d9d2ab12a |
|
BLAKE2b-256 | 9eea3837439da0e774d6410ff41cb5b4fc3b3d155d929492fc48936d8d95d135 |