OpenBallast/ballast-evalsets

Ballast eval sets

The probe sets behind the Ballast measurements. 50,147 factual questions, each with a gold answer, alias set, 7 type-matched distractors, and — where resolvable — a Wikidata subject Q-id linking it to the Ballast T0 corpus (90.5% linked at full corpus).

source	origin	notes
popqa	PopQA	native subject Q-ids + popularity; the primary instrument
wikidata	self-generated from T0	rank-stratified, uncontaminated by construction, regenerable
simpleqa	SimpleQA	typed distractors from answer_type × topic
nq_open	Natural Questions (open)	entity-linked via normalized n-gram match
triviaqa	TriviaQA (rc.nocontext)	linked via wiki-title hints + spans

Schema

question_id, source, question, prop, subj, subj_qid, gold, gold_aliases[], distractors[], s_pop, pop_decile

Intended use

Logprob choice scoring: score gold + distractors under the same prompt, prediction = argmax length-normalized answer logprob, confidence = softmax mass on argmax, abstain below 0.5. Grounded condition prepends the subject's rendered evidence block from T0 at a chosen level. Two passes (no corpus / full corpus) compose exactly to every corpus level via the subject's rank bucket — see the methodology.

Contamination stance

Models have seen the upstream benchmarks' sources in pretraining. The wikidata config is uncontaminated by construction (generated from the corpus, not the web). The headline quantity (grounded − ungrounded delta) is contamination-conservative: pretraining exposure inflates the ungrounded floor, shrinking the measured gain.

Licenses

Upstream sets keep their upstream licenses (PopQA/TriviaQA/NQ: research-standard open licenses; SimpleQA: MIT). The wikidata config is CC0. Check upstream terms before commercial use of those subsets.

A second file, halluc_probes.parquet (2-hop composition, unanswerable/false-premise, 2WikiMultiHopQA, TruthfulQA control families), lands when that phase of the matrix completes.