offline_dataset_protocol
Validate fixed offline reinforcement learning transition datasets and summarize D4RL-style provenance before recovery experiments.
SKILL.mdbasicagent-skill/skill_library/paperbench/fre/skill/d4rl_offline_rl/offline_dataset_protocol/SKILL.md
GitHubOffline Dataset Protocol
Use this skill when a recovery or benchmark task needs to verify that training data is a fixed offline RL transition dataset. Do not use it to collect online rollouts, run simulators, or claim that synthetic proxy data is a real D4RL dataset.
Inputs
- A JSON file containing either a list of transition records or an object with a
transitionslist. - Each transition must contain
observation,action,reward,next_observation,terminal, andtimeout. - Optional metadata fields such as
source,domain,variant, andis_resource_derived.
Outputs
- A JSON validation report with
ok, counts,quality_tags,errors, andmetadata. - The report preserves whether the dataset is synthetic proxy data, resource-derived data, or a real benchmark item.
Workflow
- Load the dataset JSON without mutating it.
- Confirm that every transition has the required keys.
- Coerce rewards to floats and terminal/timeout markers to booleans.
- Count transitions, terminal endings, timeout endings, and combined episode endings.
- Add the
fixed_datasettag and optional tags from metadata. - Reject empty datasets or missing fields.
Validation
Run python scripts/validate_offline_dataset.py <dataset.json> for CLI validation. Run python -m pytest tests or the Distiller validate_skill_tree.py --run-tests command for deterministic tests.
Limitations
This skill validates dataset structure and provenance; it does not implement a full D4RL environment, simulator, or benchmark download.