QLoRA: Efficient Finetuning of Quantized LLMs
by University of Washington · free · Last verified 2026-03-17
Introduces QLoRA, which combines 4-bit quantization with LoRA adapters to fine-tune a 65B LLM on a single 48GB GPU while preserving full 16-bit fine-tuning performance. Introduces NF4 data type and double quantization for extreme memory reduction.
https://arxiv.org/abs/2305.14314 ↗C+
C+—Average
Adoption: A+Quality: A+Freshness: B+Citations: FEngagement: F
Specifications
- License
- MIT
- Pricing
- free
- Capabilities
- quantized-fine-tuning, memory-efficient-training, 4-bit-inference
- Integrations
- huggingface-peft, bitsandbytes, huggingface-transformers
- Use Cases
- fine-tuning-on-consumer-gpus, memory-constrained-training, llm-fine-tuning
- API Available
- No
- Tags
- qlora, quantization, fine-tuning, 4-bit, nf4, peft, memory-efficient
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
54Adoption
90
Quality
92
Freshness
72
Citations
0
Engagement
0