Dataset · KodCode
Synthetic coding tasks and tests for SFT
KodCode/KodCode-V1-SFT-4o
Large open-source English synthetic dataset with verifiable solutions and programming tests.
What's inside
Fully synthetic corpus consisting of 12 thematic subsets in English, ranging from basic exercises and library-specific knowledge to competitive programming and interview questions. Size ranges from 100k to 1M records in tabular and text format, with each task provided with a verifiable solution and tests.
What you can do with it
- Fine-tuning language models for code generation and debugging
- Training reinforcement learning (RL) models on tasks with automated test verification
- Tasks
- Question answering
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 24.5K
- Likes
- 10
How to load it
from datasets import load_dataset
ds = load_dataset("KodCode/KodCode-V1-SFT-4o")