Neuroplasticity and Corruption in Model Mechanisms: A Case Study Of Indirect Object Identification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in transformer-based language modelling have garnered attention in widespread applications. |
| Approach: | They investigate the effects of fine-tuning on poisoned data and analyze the changes after retraining a corrupted model on the original dataset and observe neuroplasticity behaviors. |
| Outcome: | The proposed model corruption mechanisms can be generalized to longer epochs and model reforming can be performed on clean datasets. |
Similar Papers
Dissecting Fine-Tuning Unlearning in Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning-based unlearning are ineffective at completely erasing model-embedded knowledge, but their true effectiveness remains unclear. |
| Approach: | They propose to use activation patching and parameter restoration experiments to examine the limitations of fine-tuning-based unlearning methods for erasing harmful, sensitive, or copyrighted information within large language models. |
| Outcome: | The proposed methods alter the model’s knowledge retrieval process rather than genuinely erasing the problematic knowledge embedded in the model parameters. |
Fine-tuning Happens in Tiny Subspaces: Exploring Intrinsic Task-specific Subspaces of Pre-trained Language Models (2023.acl-long)
Copied to clipboard
| Challenge: | Pre-trained language models are overly parameterized and have significant redundancy . recent studies show that PLMs are highly over-parameterized and robust to pruning . |
| Approach: | They propose to re-parameter and fine-tune pre-trained language models from a new perspective: Discovery of intrinsic task-specific subspace. |
| Outcome: | The proposed model can be fine-tuned in the subspace with a small number of free parameters. |
Corrupted but Not Broken: Understanding and Mitigating the Negative Impacts of Corrupted Data in Visual Instruction Tuning (2025.emnlp-main)
Copied to clipboard
Yunhao Gou, Hansi Yang, Zhili Liu, Kai Chen, Yihan Zeng, Lanqing Hong, Zhenguo Li, Qun Liu, Bo Han, James Kwok, Yu Zhang
| Challenge: | Visual Instruction Tuning (VIT) aims to enhance Multimodal Large Language Models (MLLMs), but its effectiveness is often compromised by corrupted datasets with issues such as hallucinated content and poor OCR quality. |
| Approach: | They propose a corruption-robust training paradigm that surpasses existing strategies for mitigating the effects of corrupted data. |
| Outcome: | The proposed training paradigm surpasses existing strategies for mitigating the effects of corrupted data. |
An Empirical Analysis of Memorization in Fine-tuned Autoregressive Language Models (2022.emnlp-main)
Copied to clipboard
| Challenge: | Large language models are shown to present privacy risks through memorization of training data, but little attention has been given to the fine-tuning phase. |
| Approach: | They empirically study memorization of fine-tuning methods using membership inference and extraction attacks and show that fine-timing the head of the model has the highest susceptibility to attacks. |
| Outcome: | The proposed methods have the highest memorization risk, whereas the smaller adapters are less vulnerable to known extraction attacks. |
Safeguard Fine-Tuned LLMs Through Pre- and Post-Tuning Model Merging (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Fine-tuning large language models for downstream tasks often leads to catastrophic forgetting, notably degrading the safety of original alignments. |
| Approach: | They propose to merge the weights of pre- and post-fine-tuned models to improve safety while enhancing performance. |
| Outcome: | Experiments across different downstream tasks and models validate the method’s practicality and effectiveness. |
On the Importance of Data Size in Probing Fine-tuned Models (2022.findings-acl)
Copied to clipboard
| Challenge: | Several studies have investigated the reasons behind the effectiveness of fine-tuning, usually through the lens of probing. |
| Approach: | They propose to investigate the reasons behind the effectiveness of fine-tuning by examining the impact of data size on the extent of encoded linguistic knowledge. |
| Outcome: | The proposed probes show that the size of the training data affects the recoverability of the changes made to the model’s linguistic knowledge. |
Matching Pairs: Attributing Fine-Tuned Models to their Pre-Trained Large Language Models (2023.acl-long)
Copied to clipboard
| Challenge: | generative large language models (LLMs) are widely used but fine-tuned to improve performance on downstream applications leads to violations of model licenses, model theft, and copyright infringement. |
| Approach: | They propose to trace back the origin of a model trained to its pre-trained base model . they use different knowledge levels and attribution strategies to find out how the model was trained . |
| Outcome: | The proposed method can trace back 8 out of 10 fine tuned models with different knowledge levels and attribution strategies. |
Modular and Parameter-Efficient Fine-Tuning for NLP Models (2022.emnlp-tutorials)
Copied to clipboard
| Challenge: | State-of-the-art language models in NLP perform best when fine-tuned even on small datasets. |
| Approach: | They provide an overview of parameter-efficient fine-tuning methods and highlight similarities and differences . they highlight benefits and usage scenarios of a neglected property of parameter efficient models . |
| Outcome: | This paper provides an overview of parameter-efficient fine-tuning methods . it highlights similarities and differences by presenting them in a unified view . |
Uncovering Constraint-Based Behavior in Neural Models via Targeted Fine-Tuning (2021.acl-long)
Copied to clipboard
| Challenge: | Existing work has shown that non-linguistic biases in language models obscure linguistic knowledge. |
| Approach: | They hypothesize competing linguistic processes within a language could obscure linguistic knowledge. |
| Outcome: | The proposed model can learn linguistic constraints in a language and their relative ranking, the authors show . linguistic biases can obscure underlying linguistic knowledge, they show a single phenomenon in four languages. |
How Does Fine-tuning Affect the Geometry of Embedding Space: A Case Study on Isotropy (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning pre-trained language models are ineffective, despite their potential, pre-training models suffer from important weaknesses. |
| Approach: | They analyze the extent to which the isotropy of the embedding space changes after fine-tuning. |
| Outcome: | The proposed model improves the isotropy of embedding space after fine-tuning . the model can encode linguistic properties, but lacks the social bias needed to improve it . |