openai/prm800k
prm800k
800,000 step-level correctness labels on LLM solutions to MATH problems
Builder
OpenAI
openai • ai-lab
Stars
2,111
Using upstream star count
Forks
125
Using upstream fork count
Open Issues
0
Activity Score
0/100
0 commits in 30d
Created
Apr 13, 2023
Project creation date
README Summary
PRM800K is a dataset containing 800,000 step-level correctness labels for LLM-generated solutions to mathematical problems from the MATH dataset. The dataset provides human annotations indicating whether each reasoning step in multi-step mathematical solutions is correct or incorrect, enabling training of process reward models.
AI Dev Skills
Unmapped
Process Reward ModelingMathematical Reasoning EvaluationStep-level SupervisionReinforcement Learning from Human FeedbackLarge Language Model TrainingAI Safety and AlignmentDataset Curation and Labeling
Tags
Process Reward ModelingMathematical Reasoning EvaluationStep-level SupervisionReinforcement Learning from Human FeedbackLarge Language Model TrainingAI Safety and AlignmentDataset Curation and LabelingMathematical Tutoring System DevelopmentProcess Reward Model TrainingTextModel Training InfrastructureResearch EnvironmentProcess SupervisionMathematical Problem Solving EvaluationAI Reasoning Step VerificationMathematical ReasoningEducationHuman Feedback LearningAI SafetyPython
Taxonomy
Deployment Context
Industries
Modalities
Skill Areas
Recent Activity
Updated 2 years ago
7 Days
0
30 Days
0
90 Days
0
Quality
research- Quality
- high
- Maturity
- research
Categories
Learning ResourcesPrimarySearch & KnowledgeOther AI / MLEvals & BenchmarkingML Platform & InfrastructureSafety & AlignmentFoundation ModelsModel TrainingComputer Vision
PM Skills
Product Discovery
Languages
Python100.0%
Timeline
- Project created
- Apr 13, 2023
- Forked
- Mar 14, 2026
- Your last push
- 2 years ago
- Upstream last push
- 2 years ago
- Tracked since
- Jun 1, 2023
Similar Repos
pgvector cosine similarity · $0
Loading…