The Mosaic Memory of Large Language Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909610664787968 |
|---|---|
| author | Shilov, Igor Meeus, Matthieu de Montjoye, Yves-Alexandre |
| author_facet | Shilov, Igor Meeus, Matthieu de Montjoye, Yves-Alexandre |
| contents | As Large Language Models (LLMs) become widely adopted, understanding how they learn from, and memorize, training data becomes crucial. Memorization in LLMs is widely assumed to only occur as a result of sequences being repeated in the training data. Instead, we show that LLMs memorize by assembling information from similar sequences, a phenomena we call mosaic memory. We show major LLMs to exhibit mosaic memory, with fuzzy duplicates contributing to memorization as much as 0.8 of an exact duplicate and even heavily modified sequences contributing substantially to memorization. Despite models display reasoning capabilities, we somewhat surprisingly show memorization to be predominantly syntactic rather than semantic. We finally show fuzzy duplicates to be ubiquitous in real-world data, untouched by deduplication techniques. Taken together, our results challenge widely held beliefs and show memorization to be a more complex, mosaic process, with real-world implications for privacy, confidentiality, model utility and evaluation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_15523 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | The Mosaic Memory of Large Language Models Shilov, Igor Meeus, Matthieu de Montjoye, Yves-Alexandre Computation and Language Machine Learning As Large Language Models (LLMs) become widely adopted, understanding how they learn from, and memorize, training data becomes crucial. Memorization in LLMs is widely assumed to only occur as a result of sequences being repeated in the training data. Instead, we show that LLMs memorize by assembling information from similar sequences, a phenomena we call mosaic memory. We show major LLMs to exhibit mosaic memory, with fuzzy duplicates contributing to memorization as much as 0.8 of an exact duplicate and even heavily modified sequences contributing substantially to memorization. Despite models display reasoning capabilities, we somewhat surprisingly show memorization to be predominantly syntactic rather than semantic. We finally show fuzzy duplicates to be ubiquitous in real-world data, untouched by deduplication techniques. Taken together, our results challenge widely held beliefs and show memorization to be a more complex, mosaic process, with real-world implications for privacy, confidentiality, model utility and evaluation. |
| title | The Mosaic Memory of Large Language Models |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2405.15523 |