datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
lean-verifier-formalizations
Lean Verifier Formalizations
A dataset of Lean 4 theorem-proving tasks for evaluating agentic coding harnesses. Each row pairs a formal task_statement (with the reference proof body removed) against a real Lean 4 repository, plus the informal_excerpt/informal_source_text describing what the theorem claims, permitted_axioms for the verifier, and provenance fields (repo_url, repo_commit_sha, license) tracing back to the source project.
Sources
Every row is pulled… See the full description on the dataset page: https://huggingface.co/datasets/Pradheep1647/lean-verifier-formalizations.atlas-16-verifier-permission-prompt-ablation
ATLAS report 16: does the orchestrator's "cannot solve" clause suppress candidate verification?
Complete raw products of the ATLAS rl-training report 16 experiment
(GitHub issue #36). Two system-prompt arms of the same model over the
same 78 fixed states, greedy decoding, one shared vLLM server.
What the experiment did
The ATLAS orchestrator's frozen system prompt contains the clause
You cannot solve the problem yourself; you decide when to explore
further and… See the full description on the dataset page: https://huggingface.co/datasets/t2ance/atlas-16-verifier-permission-prompt-ablation.verifier-debias-v2
verifier-debias-v2 — de-biased SFT data for a generative verifier
This dataset was presented in the paper One Token to Fool LLM-as-a-Judge.
GitHub repository: yulaizhao/Master-RM
SFT data to train a generative verifier (GenRM, arXiv:2408.15240)
from Qwen/Qwen2.5-7B-Instruct. Each row is a conversational example
(messages = system + user + gold assistant) plus a verdict (PASS/FAIL) and
the underlying 1–5 score. The assistant target is a critique ending in
Verdict: PASS / Verdict:… See the full description on the dataset page: https://huggingface.co/datasets/narcolepticchicken/verifier-debias-v2.SWE_Lego_real_data_Verifier
SWE-Lego Real Data Trajectories Verifier
This dataset contains merged verifier trajectories from verifier_part0 to verifier_part8.
Total samples: 17,642
Format: parquet
Main columns: instance_id, messages, and verifier-related metadata fields.
mathcompose-verifier
mathcompose — Model V (process verifier) SFT data
First-error-localization critiques for training a small generative math verifier.
Each row: a problem + a step-indexed candidate solution (prompt) and a
paragraph-by-paragraph critique ending in \boxed{{k}} (completion), where k
is the 0-based index of the first erroneous step (-1 = all correct).
Labels are ground truth from PRM800K (OpenAI, MIT).
Genuine-detection recipe: claude-opus-4-8 critiques each solution blind
(never… See the full description on the dataset page: https://huggingface.co/datasets/42e/mathcompose-verifier.numeric-claim-verifier
Numeric Claim Verifier (Science) — Adaption AutoScientist
Programmatically verified prompt/completion pairs for scientific and statistical numeric claim verification.
Labels
correct — claim matches ground-truth tables
wrong_direction — trend/sign reversed
wrong_magnitude — right direction, wrong size (25–70% offset)
unverifiable — no matching source row (real entity + absent metric)
Sources
Our World in Data CO₂ / Energy
WHO GHO life expectancy… See the full description on the dataset page: https://huggingface.co/datasets/mishface123/numeric-claim-verifier.m9-verifier-38k-aligned
M9 Verifier 38K Aligned
This dataset contains 38,564 prompts with verifier-compatible gold answers for
an M9 RLVR-GRPO experiment in a unified post-training study with Qwen3-1.7B.
It is an independent research artifact, not an official release from the model
or paper authors.
The bank was reconstructed from the frozen
YangyiH/openreasoning_mixed_100k
prompt mixture. Every recovered row was matched to the frozen base row by
domain, source shard, and prompt SHA-256 before verifier… See the full description on the dataset page: https://huggingface.co/datasets/YangyiH/m9-verifier-38k-aligned.verifiers-tasks-v1
verifiers-tasks-v1
Task dataset for a verifiers RL / eval environment, in the shape used by the
Prime Intellect Environments Hub.
22 tasks about the verifiers library itself — the library every environment in that program
is written against. The skill tested is the one a contributor actually needs: parsing a model's
output, writing reward functions, and composing them into a Rubric whose weighting does what
you meant.
Category
Tasks
Covers
parsers
7
Parser, XMLParser… See the full description on the dataset page: https://huggingface.co/datasets/eltociear/verifiers-tasks-v1.algocean-verifier
algocean-verifier
주장과 근거를 받아 주장이 근거로 뒷받침되는지 판정하도록 가르치는 LoRA SFT 데이터셋입니다.어긋나면 실패 유형까지 말합니다.
RAG 파이프라인의 검증 노드용입니다. 근거를 인용하지 않고 판정만 합니다.
규모
파일
행 수
verifier.jsonl
150,000
verifier.eval.jsonl
2,000
형식: JSONL, messages 3턴
언어: 한국어 약 70% · 영어 약 30%
think 비율이 높아 대조 추론에 적합
어디에 쓰나요
추출된 근거가 주장을 실제로 지지하는지 검사
숫자·날짜 왜곡, 근거 부재 등 실패 유형 분류
경량 모델로 프론티어 답변 검증 필터
어떤 모델에 LoRA 하나요
베이스: 1B ~ 9B 인스트럭트 모델
rank 816, alpha 1632, lr… See the full description on the dataset page: https://huggingface.co/datasets/Algocean/algocean-verifier.