# When Fine-Tuning Fails: Lessons from MS MARCO Passage Ranking

arXiv: https://arxiv.org/abs/2506.18535 · HTML: https://mlsanity.com/paper/2506.18535/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 16 references that are in the map.
- First posted on arXiv 23 Jun 2025.

## Builds on (most cited first, 9 of 9)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction](https://mlsanity.com/paper/1802.03426/) (2018)
- [FaceNet: A Unified Embedding for Face Recognition and Clustering](https://mlsanity.com/paper/1503.03832/) (2015)
- [MS MARCO: A Human Generated MAchine Reading COmprehension Dataset](https://mlsanity.com/paper/1611.09268/) (2016)
- [ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT](https://mlsanity.com/paper/2004.12832/) (2020)
- [Exploring Dual Encoder Architectures for Question Answering](https://mlsanity.com/paper/2204.07120/) (2022)
- [Pathway to Relevance: How Cross-Encoders Implement a Semantic Variant of BM25](https://mlsanity.com/paper/2502.04645/) (2025)

## Built on by (most cited first, 1 of 1)

- [Fine-Tuning vs. RAG for Multi-Hop Question Answering with Novel Knowledge](https://mlsanity.com/paper/2601.07054/) (2026)

## Abstract (arXiv)

This paper investigates the counterintuitive phenomenon where fine-tuning pre-trained transformer models degrades performance on the MS MARCO passage ranking task. Through comprehensive experiments involving five model variants-including full parameter fine-tuning and parameter efficient LoRA adaptations-we demonstrate that all fine-tuning approaches underperform the base sentence-transformers/all- MiniLM-L6-v2 model (MRR@10: 0.3026). Our analysis reveals that fine-tuning disrupts the optimal embedding space structure learned during the base model's extensive pre-training on 1 billion sentence pairs, including 9.1 million MS MARCO samples. UMAP visualizations show progressive embedding space flattening, while training dynamics analysis and computational efficiency metrics further support our findings. These results challenge conventional wisdom about transfer learning effectiveness on saturated benchmarks and suggest architectural innovations may be necessary for meaningful improvements.