Unsloth
by Unsloth AI · open-source · Last verified 2026-03-17
Library for 2-5x faster LLM fine-tuning with 80% less memory usage through custom CUDA kernels. Provides optimized training for popular model architectures with automatic quantization and model export.
https://unsloth.ai ↗C+
C+—Average
Adoption: BQuality: AFreshness: A+Citations: BEngagement: F
Specifications
- License
- Apache-2.0
- Pricing
- open-source
- Capabilities
- fast-fine-tuning, memory-optimization, custom-cuda-kernels, auto-quantization, model-export
- Integrations
- hugging-face, weights-and-biases
- Use Cases
- efficient-fine-tuning, consumer-gpu-training, model-customization, research
- API Available
- No
- SDK Languages
- python
- Deployment
- self-hosted, google-colab
- Rate Limits
- N/A (self-hosted)
- Data Privacy
- Self-hosted, user-managed; runs locally
- Tags
- fine-tuning, optimization, fast-training, memory-efficient
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
58.7Adoption
65
Quality
86
Freshness
92
Citations
62
Engagement
0