How-Resilient-are-Imitation-Learning-Methods-to-Sub-Optimal-Experts / Experts /expert_MountainCar-v0_criteria160.npz
Download Experts/expert_MountainCar-v0_criteria160.npz from NathanGavenski/How-Resilient-are-Imitation-Learning-Methods-to-Sub-Optimal-Experts: direct link, hf CLI and curl.
- Browser
- Download file 19.6 MB
-
https://huggingface.co/datasets/NathanGavenski/How-Resilient-are-Imitation-Learning-Methods-to-Sub-Optimal-Experts/resolve/main/Experts/expert_MountainCar-v0_criteria160.npz
- Command line
-
hf download hf://datasets/NathanGavenski/How-Resilient-are-Imitation-Learning-Methods-to-Sub-Optimal-Experts/Experts/expert_MountainCar-v0_criteria160.npz
-
curl -L -o expert_MountainCar-v0_criteria160.npz https://huggingface.co/datasets/NathanGavenski/How-Resilient-are-Imitation-Learning-Methods-to-Sub-Optimal-Experts/resolve/main/Experts/expert_MountainCar-v0_criteria160.npz
19.6 MB
- Xet hash:
- d877870ca0b7da46f2b4103577fdcafccd3d0563c5658ded120629a13df71c3c
- Size of remote file:
- 19.6 MB
- SHA256:
- 7ae602d7a13b7f9b53f4a5c86faff22baba1428e4b234c08af3870eea118d939
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.