Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.08780 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916889185222656 |
|---|---|
| author | Nijasure, Atharva Chowdhury, Tanya Allan, James |
| author_facet | Nijasure, Atharva Chowdhury, Tanya Allan, James |
| contents | We conduct a behavioral exploration of LoRA fine-tuned LLMs for Passage Reranking to understand how relevance signals are learned and deployed by Large Language Models. By fine-tuning Mistral-7B, LLaMA3.1-8B, and Pythia-6.9B on MS MARCO under diverse LoRA configurations, we investigate how relevance modeling evolves across checkpoints, the impact of LoRA rank (1, 2, 8, 32), and the relative importance of updated MHA vs. MLP components. Our ablations reveal which layers and projections within LoRA transformations are most critical for reranking accuracy. These findings offer fresh explanations into LoRA's adaptation mechanisms, setting the stage for deeper mechanistic studies in Information Retrieval. All models used in this study have been shared. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_08780 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | How Relevance Emerges: Interpreting LoRA Fine-Tuning in Reranking LLMs Nijasure, Atharva Chowdhury, Tanya Allan, James Information Retrieval Machine Learning We conduct a behavioral exploration of LoRA fine-tuned LLMs for Passage Reranking to understand how relevance signals are learned and deployed by Large Language Models. By fine-tuning Mistral-7B, LLaMA3.1-8B, and Pythia-6.9B on MS MARCO under diverse LoRA configurations, we investigate how relevance modeling evolves across checkpoints, the impact of LoRA rank (1, 2, 8, 32), and the relative importance of updated MHA vs. MLP components. Our ablations reveal which layers and projections within LoRA transformations are most critical for reranking accuracy. These findings offer fresh explanations into LoRA's adaptation mechanisms, setting the stage for deeper mechanistic studies in Information Retrieval. All models used in this study have been shared. |
| title | How Relevance Emerges: Interpreting LoRA Fine-Tuning in Reranking LLMs |
| topic | Information Retrieval Machine Learning |
| url | https://arxiv.org/abs/2504.08780 |