Compară metode
Examinează metodele selectate una lângă alta; rândurile care diferă sunt evidențiate.
| QLoRA× | Optimizarea Directă a Preferințelor× | |
|---|---|---|
| Domeniu | Învățare profundă | Învățare profundă |
| Familie | Machine learning | Machine learning |
| Anul apariției | 2023 | 2023 |
| Autorul original≠ | Tim Dettmers | Rafael Rafailov |
| Tip | Training methodology | Training methodology |
| Sursa seminală≠ | Dettmers, T., Pagnoni, A., Holtzman, A., & Contrastive, L. (2023). QLoRA: Efficient finetuning of quantized LLMs. arXiv preprint arXiv:2305.14314. link ↗ | Rafailov, R., Sharma, A., Mitchell, E., Manning, C. D., Ermon, S., & Finn, C. (2023). Direct preference optimization: Your language model is secretly a reward model. arXiv preprint arXiv:2305.18290. link ↗ |
| Denumiri alternative | QLoRA, Quantized LoRA | DPO, Direct preference |
| Înrudite | 4 | 4 |
| Rezumat≠ | QLoRA is an efficient fine-tuning method introduced by Dettmers et al. in 2023 that enables fine-tuning large language models using quantization and low-rank adaptation. By combining 4-bit quantization with LoRA, QLoRA reduces memory requirements by 75%, enabling fine-tuning of 65B-parameter models on single GPUs. | Direct Preference Optimization (DPO) is a training method introduced by Rafailov et al. in 2023 that aligns language models with human preferences without requiring an explicit reward model. By directly optimizing for preference pairs (better response vs worse response), DPO simplifies the training pipeline compared to reinforcement learning from human feedback (RLHF). |
| ScholarGateSet de date ↗ |
|
|