Synthetic regex dataset for reinforcement learning post-training.
Size: 1,158 examples
Format: JSONL
Use Case: GRPO/RL training for regex generation
{
"prompt": "Write a Python regex pattern that matches:
",
"solution": "<regex_pattern>",
"test_cases": {
"positive": ["match1", "match2", "match3", "match4", "match5"],
"negative": ["no_match1", "no_match2", "no_match3", "no_match4"… See the full description on the dataset page: https://huggingface.co/datasets/zenzen9/regex-rl-dataset.
