WizardLM Evol-Instruct
by Microsoft Research · free · Last verified 2026-03-17
WizardLM Evol-Instruct is a synthetic dataset created by Microsoft Research for fine-tuning large language models. It uses an LLM-based evolutionary process to iteratively rewrite and complicate a seed set of instructions, progressively increasing their complexity and diversity. The dataset is designed to enhance a model's ability to follow intricate, multi-step commands across various domains like coding, math, and reasoning.
https://huggingface.co/datasets/WizardLMTeam/WizardLM_evol_instruct_V2_196k ↗B
B—Above Average
Adoption: B+Quality: AFreshness: BCitations: B+Engagement: F
Specifications
- License
- Apache-2.0
- Pricing
- free
- Capabilities
- Supervised fine-tuning (SFT), Instruction complexity scaling, Enhancing LLM reasoning abilities, Improving code generation and debugging, Training models for multi-turn dialogue, Fine-tuning for long-form text generation, Data augmentation for instruction datasets, Benchmarking instruction-following capabilities
- Integrations
- Use Cases
- [object Object], [object Object], [object Object], [object Object]
- API Available
- No
- Tags
- evol-instruct, complexity-evolution, synthetic, wizardlm, instruction-following, microsoft-research, data-augmentation, llm-training, reasoning, code-generation, sft
- Added
- 2026-03-17
- Completeness
- 0.9%
Index Score
67.2Adoption
77
Quality
83
Freshness
68
Citations
79
Engagement
0