All assets available: VP datasets, eval windows, expert videos, goal-transfer experts; full SHA256SUMS
Browse files- README.md +4 -4
- SHA256SUMS +0 -0
README.md
CHANGED
|
@@ -15,10 +15,10 @@ manifest row), `goal_transfer_experts.json` and `SHA256SUMS` (verify a download
|
|
| 15 |
| `experts/dmc/<env>/` | BRO policy at 1M steps: `policy.msgpack`, `policy.json` | `--config.expert_checkpoint=<dir>` | | available |
|
| 16 |
| `experts/ogbench/<env>/` | FQL agent: `flags.json`, `params_2000000.pkl` | `--config.expert_checkpoint=<dir>` | | available |
|
| 17 |
| `experts/atari/<game>/` | DQN (atari-agents, 50M steps): `manifest.json`, `weights.npz` | `--config.expert_manifest=<manifest.json>` | 0.3 GB (all experts) | available |
|
| 18 |
-
| `experts/ogbench_goals/<env>/task<k>/` | FQL experts of the other OGBench goals (`goal_transfer_experts.json`) | `eval_goal_transfer.py --expert_checkpoint=<dir>` |
|
| 19 |
-
| `eval_datasets/<env>/` | Labeled expert windows for the offline IDM audit (`labeled_eval_dataset`) | `train_mirror.py`, `train_random_idm.py`, `train_bco.py` | 7.6 GB |
|
| 20 |
-
| `expert_videos/<env>/{bco,laifo}.npz` | Expert videos (`expert_dataset`) | `train_bco.py`, `train_laifo.py` | 6.6 GB |
|
| 21 |
-
| `vp_datasets/<env>[_episodes]/` | Video-planner training datasets (`mirror/vp_data.py`) | `train_video_planner.py`, `train_supervised_idm.py --config.dataset_path=...` | ~3.3 TB |
|
| 22 |
|
| 23 |
There is one asset of each kind per task (24 tasks: 4 OGBench, 13 DMC, 7 Atari).
|
| 24 |
|
|
|
|
| 15 |
| `experts/dmc/<env>/` | BRO policy at 1M steps: `policy.msgpack`, `policy.json` | `--config.expert_checkpoint=<dir>` | | available |
|
| 16 |
| `experts/ogbench/<env>/` | FQL agent: `flags.json`, `params_2000000.pkl` | `--config.expert_checkpoint=<dir>` | | available |
|
| 17 |
| `experts/atari/<game>/` | DQN (atari-agents, 50M steps): `manifest.json`, `weights.npz` | `--config.expert_manifest=<manifest.json>` | 0.3 GB (all experts) | available |
|
| 18 |
+
| `experts/ogbench_goals/<env>/task<k>/` | FQL experts of the other OGBench goals (`goal_transfer_experts.json`) | `eval_goal_transfer.py --expert_checkpoint=<dir>` | 1.1 GB | available |
|
| 19 |
+
| `eval_datasets/<env>/` | Labeled expert windows for the offline IDM audit (`labeled_eval_dataset`) | `train_mirror.py`, `train_random_idm.py`, `train_bco.py` | 7.6 GB | available |
|
| 20 |
+
| `expert_videos/<env>/{bco,laifo}.npz` | Expert videos (`expert_dataset`) | `train_bco.py`, `train_laifo.py` | 6.6 GB | available |
|
| 21 |
+
| `vp_datasets/<env>[_episodes]/` | Video-planner training datasets (`mirror/vp_data.py`) | `train_video_planner.py`, `train_supervised_idm.py --config.dataset_path=...` | ~3.3 TB | available |
|
| 22 |
|
| 23 |
There is one asset of each kind per task (24 tasks: 4 OGBench, 13 DMC, 7 Atari).
|
| 24 |
|
SHA256SUMS
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|