Low-Confidence Gold: Refining Low-Confidence Samples for Efficient Instruction Tuning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917390279770112 |
|---|---|
| author | Cai, Hongyi Li, Jie Rahman, Mohammad Mahdinur Dong, Wenzhen |
| author_facet | Cai, Hongyi Li, Jie Rahman, Mohammad Mahdinur Dong, Wenzhen |
| contents | The effectiveness of instruction fine-tuning for Large Language Models is fundamentally constrained by the quality and efficiency of training datasets. This work introduces Low-Confidence Gold (LCG), a novel filtering framework that employs centroid-based clustering and confidence-guided selection for identifying valuable instruction pairs. Through a semi-supervised approach using a lightweight classifier trained on representative samples, LCG curates high-quality subsets while preserving data diversity. Experimental evaluation demonstrates that models fine-tuned on LCG-filtered subsets of 6K samples achieve superior performance compared to existing methods, with substantial improvements on MT-bench and consistent gains across comprehensive evaluation metrics. The framework's efficacy while maintaining model performance establishes a promising direction for efficient instruction tuning.All open-source assets are publicly available at https://github.com/Lizruletheworld/Low-Confidence_Gold. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_18978 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Low-Confidence Gold: Refining Low-Confidence Samples for Efficient Instruction Tuning Cai, Hongyi Li, Jie Rahman, Mohammad Mahdinur Dong, Wenzhen Computation and Language Artificial Intelligence The effectiveness of instruction fine-tuning for Large Language Models is fundamentally constrained by the quality and efficiency of training datasets. This work introduces Low-Confidence Gold (LCG), a novel filtering framework that employs centroid-based clustering and confidence-guided selection for identifying valuable instruction pairs. Through a semi-supervised approach using a lightweight classifier trained on representative samples, LCG curates high-quality subsets while preserving data diversity. Experimental evaluation demonstrates that models fine-tuned on LCG-filtered subsets of 6K samples achieve superior performance compared to existing methods, with substantial improvements on MT-bench and consistent gains across comprehensive evaluation metrics. The framework's efficacy while maintaining model performance establishes a promising direction for efficient instruction tuning.All open-source assets are publicly available at https://github.com/Lizruletheworld/Low-Confidence_Gold. |
| title | Low-Confidence Gold: Refining Low-Confidence Samples for Efficient Instruction Tuning |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2502.18978 |