datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
bounded-returns-deep-research
Bounded Returns to Orchestration — Deep Research Evaluation Frames
Judge-level evaluation data from a controlled comparison of automated
deep-research architectures over a 90-query manifest. These are the tidy frames
the paper's statistics are computed from: the reported numbers are recomputable
from the tables here using the analysis scripts in the code repository.
The paper compares eleven architectures on a single shared tool layer: a
single-pass baseline, eight orchestration… See the full description on the dataset page: https://huggingface.co/datasets/PeterStrain77/bounded-returns-deep-research.source-bounded-expert-notes-preview
Eight Source-Bounded Expert Notes
This repository contains public-preview, source-bounded expert notes generated from public research metadata and local derived receipt signals. It records bounded readings of research programs, questions that would sharpen the model, and source-bounded test ideas extracted across domains.
This is a public preview batch. Later batches may expand the index and add deeper receipt surfaces.
Data Boundary
Generated UTC:… See the full description on the dataset page: https://huggingface.co/datasets/cjc0013/source-bounded-expert-notes-preview.eh-caution-install-bounded-site-sweep
caution-install-bounded-site-sweep -- aggregate exhaust
Aggregate-only: every file committed under this experiment's analysis-committed/ tree (dose-response tables, direction fits, gate AUROCs, manifests, and any other analysis artifact), copied byte-for-byte. No source question text, aliases, or per-row generation text -- analysis-committed/ never carries those.
HF repo: professorsynapse/eh-caution-install-bounded-site-sweep
Provenance
Experiment:… See the full description on the dataset page: https://huggingface.co/datasets/professorsynapse/eh-caution-install-bounded-site-sweep.foi-process-event-logs-boundedspecsafe-bounded-negative-evidence-v1
SpecSafe Bounded Negative-Evidence Release
Validity marker
CALIBRATION_UNFIT_FOR_ADAPTIVE_POLICY
Release status
release_id=specsafe-bounded-negative-evidence-v1
release_type=bounded_negative_evidence
publication_status=local_pack_only
candidate_not_promoted=true
threshold_promotion_authorized=false
scheduler_promotion_authorized=false
production_claim_authorized=false
This local pack presents a governed negative result. It does not present the… See the full description on the dataset page: https://huggingface.co/datasets/KaboKableMolefe/specsafe-bounded-negative-evidence-v1.ns-boundedtrust-bounded-governance-trace-formatztl-bounded-int-claims
Bounded-integer claims: TRUE, FALSE, or the bounds do not decide
Each row is a small comparison over integer unknowns with stated bounds, for example
x is an integer with 0 <= x <= 1000. Consider the claim: x == 1 - x. Answer TRUE if it holds
for every such value, FALSE if it fails for every such value, or UNDETERMINED if it holds for
some values and fails for others.
Gold: FALSE — no integer equals one minus itself (the real number ½ would, but it is not admissible).
The… See the full description on the dataset page: https://huggingface.co/datasets/Inventor1975/ztl-bounded-int-claims.