msr-spare-1/spare-gpt55-static-corpus
SPARE GPT-5.5 Grounded Cognitive Multi-Turn Games This public dataset contains 7,872 validated Python game environments for actor-only SPARE training. Six cognitive skills, exactly 1,312 environments per skill Generated with GPT-5.5 and grounded by spice_megascience_15k.jsonl Grounding corpus SHA-256: a36a928b4940b5b5d9e3f4cb5804a94c69462360943adb3be14613c82f0f72c0 Maximum 25 turns and 32K generation context Every environment passes load, reset, step, and replay validation with… See the full description on the dataset page: https://huggingface.co/datasets/msr-spare-1/spare-gpt55-static-corpus.
SPARE GPT-5.5 Grounded Cognitive Multi-Turn Games
This public dataset contains 7,872 validated Python game environments for actor-only SPARE training.
- Six cognitive skills, exactly 1,312 environments per skill
- Generated with GPT-5.5 and grounded by
spice_megascience_15k.jsonl - Grounding corpus SHA-256:
a36a928b4940b5b5d9e3f4cb5804a94c69462360943adb3be14613c82f0f72c0 - Maximum 25 turns and 32K generation context
- Every environment passes load, reset, step, and replay validation with a 2-second reset budget
- No tool-use or terminal environments
Layout
games/*.py: trainer-ready SPARE environment filesdata/train.jsonl: Hugging Facedatasetsrepresentation with embedded source codemanifest.jsonl: generation, skill, grounding, source, and checksum metadatavalidation_manifest.jsonl: per-environment validation resultsvalidation_summary.json: validation totals and parameterschecksums.sha256: SHA-256 checksums for every game file
Training
The pool supports 400 rollout steps with 24 environments per step. SPARE's static no-replacement sampler exhausts and reshuffles the full pool before reuse, so 7,872 environments are seen once before 1,728 controlled repeats.
hf download simonycl/spare-gpt55-cognitive-multiturn-7872 \
--repo-type dataset \
--local-dir /scratch/spare-gpt55-cognitive-multiturn-7872Point STATIC_POOL_DIR at the downloaded games/ directory.
