metadata
license: other
license_name: ase-sample-license
license_link: >-
https://huggingface.co/datasets/Deterministic-Data/ase-trajectories/blob/main/LICENSE
tags:
- rl
- reinforcement-learning
- trajectory
- synthetic-data
- offline-rl
- imitation-learning
- behavioral-cloning
- deterministic
- sequential-decision
task_categories:
- reinforcement-learning
- tabular-regression
- time-series-forecasting
ASE Syntax Extractions
The machine does not dream. It computes — and in that computation, structure emerges. This is not simulated data; it is an extraction of axiomatic necessity.
Overview
This dataset contains deterministic trajectory extractions from a closed, axiomatic system. Every frame is the output of a syntax engine where (seed, tick, entity) tuples are resolved through fixed transformations.
- Deterministic: The same
(run_id, batch_index)generates identical output, bit-for-bit, regardless of platform. - Pure: No human data, no scraping. Every byte is synthetic, produced by pure operation.
Formats
.jsonl(6 fields): Minimal state transitions.frame_id: Sequence identifier.context_hash_crc32,state_hash_crc32,action_hash_crc32,next_state_hash_crc32: Deterministic fingerprints.reward: Scalar feedback signal.
.csv(90 dimensions): Full expansion (v0–v89) including topology, behavioral indices, and system ecology metrics.
Determinism Logic
The engine seeds each batch from a compound key:
| Property | Value |
|---|---|
| Throughput | 200+ ticks/sec |
| Collision rate | ~2.3×10⁻⁸ % per pair |
| Stability | Closed-loop, zero invalid states |
Sample (JSONL)
{"frame_id":104,"context_hash_crc32":2210934871,"state_hash_crc32":991823410,"action_hash_crc32":4022881193,"next_state_hash_crc32":88213410,"reward":-0.1832}
{"frame_id":105,"context_hash_crc32":2210934871,"state_hash_crc32":88213410,"action_hash_crc32":129384710,"next_state_hash_crc32":340281993,"reward":0.5011}