LoRA: Low-Rank Adaptation of Large Language Models
by Microsoft Research · free · Last verified 2026-03-17
Introduces LoRA, which freezes pretrained model weights and injects trainable low-rank decomposition matrices into Transformer layers. Reduces trainable parameters by 10,000× and GPU memory by 3× with no inference latency overhead, enabling efficient LLM fine-tuning.
https://arxiv.org/abs/2106.09685 ↗B+
B+—Good
Adoption: A+Quality: A+Freshness: BCitations: AEngagement: F
Specifications
- License
- MIT
- Pricing
- free
- Capabilities
- parameter-efficient-fine-tuning, low-rank-adaptation, memory-efficient-training
- Integrations
- huggingface-peft, huggingface-transformers, axolotl
- Use Cases
- fine-tuning, domain-adaptation, instruction-tuning
- API Available
- No
- Tags
- lora, fine-tuning, low-rank, parameter-efficient, peft, adaptation
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
78.8Adoption
95
Quality
94
Freshness
62
Citations
88
Engagement
0
Put AI to work for your business
Deploy this paper alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.