Instructions to use tuanhqv123/tars-mini with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use tuanhqv123/tars-mini with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="tuanhqv123/tars-mini", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
File size: 893 Bytes
4b7148d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | """Render verification videos: walk_forward and the sleep/stand lifecycle."""
import imageio.v2 as imageio, numpy as np
from model import Params
from sleep_pose import lifecycle, POWER
from run_gaits import evaluate
def save(frames, name, fps=25, step=1):
imageio.mimsave(f"../results/{name}.gif", frames[::step], duration=1 / fps, loop=0)
k = np.linspace(0, len(frames) - 1, 8).astype(int)
sheet = np.vstack([np.hstack([frames[i] for i in k[:4]]), np.hstack([frames[i] for i in k[4:]])])
imageio.imwrite(f"../results/{name}_sheet.png", sheet[::2, ::2])
p = Params(mu=0.8, **POWER)
r, s = lifecycle(p, 60, record=True, cam=dict(azimuth=0, elevation=-5, distance=0.65))
print(r); save(s.frames, "sleep_standup", step=2)
r, s = evaluate(p, "walk_forward", record=True)
print({k: r[k] for k in ("forward_cm", "max_tilt_deg", "yaw_deg")}); save(s.frames, "walk_forward", step=3)
|