Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2412.18497 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909680334274560 |
|---|---|
| author | Huang, Ko-Wei Fu, Yi-Fu Tsai, Ching-Yu Tu, Yu-Chieh Cheng, Tzu-Ling Lin, Cheng-Yu Yang, Yi-Ting Liu, Heng-Yi Liao, Keng-Te Juan, Da-Cheng Lin, Shou-De |
| author_facet | Huang, Ko-Wei Fu, Yi-Fu Tsai, Ching-Yu Tu, Yu-Chieh Cheng, Tzu-Ling Lin, Cheng-Yu Yang, Yi-Ting Liu, Heng-Yi Liao, Keng-Te Juan, Da-Cheng Lin, Shou-De |
| contents | We investigate how Large Language Models (LLMs) distinguish between memorization and generalization at the neuron level. Through carefully designed tasks, we identify distinct neuron subsets responsible for each behavior. Experiments on both a GPT-2 model trained from scratch and a pretrained LLaMA-3.2 model fine-tuned with LoRA show consistent neuron-level specialization. We further demonstrate that inference-time interventions on these neurons can steer the model's behavior toward memorization or generalization. To assess robustness, we evaluate intra-task and inter-task consistency, confirming that these neuron-behavior associations reflect generalizable patterns rather than dataset-specific artifacts. Our findings reveal modular structure in LLMs and enable controlling memorization and generalization behaviors at inference time. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_18497 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Neuron-Level Differentiation of Memorization and Generalization in Large Language Models Huang, Ko-Wei Fu, Yi-Fu Tsai, Ching-Yu Tu, Yu-Chieh Cheng, Tzu-Ling Lin, Cheng-Yu Yang, Yi-Ting Liu, Heng-Yi Liao, Keng-Te Juan, Da-Cheng Lin, Shou-De Computation and Language We investigate how Large Language Models (LLMs) distinguish between memorization and generalization at the neuron level. Through carefully designed tasks, we identify distinct neuron subsets responsible for each behavior. Experiments on both a GPT-2 model trained from scratch and a pretrained LLaMA-3.2 model fine-tuned with LoRA show consistent neuron-level specialization. We further demonstrate that inference-time interventions on these neurons can steer the model's behavior toward memorization or generalization. To assess robustness, we evaluate intra-task and inter-task consistency, confirming that these neuron-behavior associations reflect generalizable patterns rather than dataset-specific artifacts. Our findings reveal modular structure in LLMs and enable controlling memorization and generalization behaviors at inference time. |
| title | Neuron-Level Differentiation of Memorization and Generalization in Large Language Models |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2412.18497 |