Dataset · nvidia
OpenCodeReasoning: synthetic code reasoning
nvidia/OpenCodeReasoning
735K synthetic reasoning examples across 28K competitive programming tasks.
What's inside
OpenCodeReasoning is the largest synthetic code reasoning dataset to date. It contains 735,255 Python examples across 28,319 unique competitive programming tasks. The dataset is designed for supervised fine-tuning (SFT); modality is text, and size ranges from 100K to 1M.
What you can do with it
- Fine-tuning coding LLMs
- Fine-tuning LoRA for competitive programming
- Evaluating reasoning models on code tasks
- Tasks
- Text generation
- Size
- 100K–1M rows
- Downloads this month
- 30.1K
- Likes
- 577
Models trained on it
- cybertruck32489/LFM3-1.7B-RDText generation 3.9K
- mradermacher/Bala-30B-A3B-i1-GGUF 3.9K
- mradermacher/Qwen3-Code-Reasoning-Instruct-6B-Brainstorm20x-i1-GGUF 3.3K
- mradermacher/Qwen3-0.9B-A0.6B-GGUF 3.1K
- nvidia/OpenCodeReasoning-Nemotron-7BText generation 2.7K
- mradermacher/Atem-8B-i1-GGUF 2.6K
- suayptalha/Qwen3-0.6B-Code-ExpertText generation 2.1K
- cybertruck32489/LFM3-1.7B-RD-v2Text generation 1.8K
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/OpenCodeReasoning")