中文
相关论文

相关论文: Memorization Dynamics of Fill-in-the-Middle Pretra…

200 篇论文

We show that autoregressive language models can learn to infill text after we apply a straightforward transformation to the dataset, which simply moves a span of text from the middle of a document to its end. While this data augmentation…

计算与语言 · 计算机科学 2022-07-29 Mohammad Bavarian , Heewoo Jun , Nikolas Tezak , John Schulman , Christine McLeavey , Jerry Tworek , Mark Chen

Fill-in-the-Middle (FIM) is a common pretraining method for code LLMs, where models complete code segments given surrounding context. However, existing LLMs treat code as plain text and mask random character spans. We propose and evaluate…

计算与语言 · 计算机科学 2025-06-03 Linyuan Gong , Alvin Cheung , Mostafa Elhoushi , Sida Wang

Large Language Models (LLMs) have significantly advanced code completion, yet they often fail when the developer's intent is underspecified in the code context. To address this, developers usually add natural language instructions (e.g.,…

软件工程 · 计算机科学 2025-10-14 Zhensu Sun , Chengran Yang , Chao Peng , Pengfei Gao , Xiaoning Du , Li Li , David Lo

We introduce Syntax-Aware Fill-In-the-Middle (SAFIM), a new benchmark for evaluating Large Language Models (LLMs) on the code Fill-in-the-Middle (FIM) task. This benchmark focuses on syntax-aware completions of program structures such as…

计算与语言 · 计算机科学 2024-06-25 Linyuan Gong , Sida Wang , Mostafa Elhoushi , Alvin Cheung

The performance of Large Language Models (LLMs) often degrades when crucial information is in the middle of a long context, a "lost-in-the-middle" phenomenon that mirrors the primacy and recency effects in human memory. We propose that this…

机器学习 · 计算机科学 2025-10-14 Nikolaus Salvatore , Hao Wang , Qiong Zhang

Fill-in-the-Middle (FIM) models play a vital role in code completion tasks, leveraging both prefix and suffix context to provide more accurate and contextually relevant suggestions. This paper presents approaches to improve FIM code…

信息检索 · 计算机科学 2024-12-24 Hitesh Sagtani , Rishabh Mehrotra , Beyang Liu

Large language models (LLMs) are often used for infilling tasks, which involve predicting or generating missing information in a given text. These tasks typically require multiple interactions with similar context. To reduce the computation…

计算与语言 · 计算机科学 2025-05-30 Tianyu Guo , Hande Dong , Yichong Leng , Feng Liu , Cheater Lin , Nong Xiao , Xianwei Zhang

We present PRISM, a comprehensive empirical study of mid-training design choices for large language models. Through controlled experiments across seven base models spanning four families (Granite, LLaMA, Mistral, Nemotron-H), two…

机器学习 · 计算机科学 2026-03-25 Bharat Runwal , Ashish Agrawal , Anurag Roy , Rameswar Panda

Although large language models excel across many tasks, they can memorise training data and thereby expose private or copyrighted text. Most defences target the pre-training stage, leaving memorisation during fine-tuning, especially for…

计算与语言 · 计算机科学 2025-10-14 Dean L. Slack , Noura Al Moubayed

Large Language Models (LLMs) frequently memorize long sequences verbatim, often with serious legal and privacy implications. Much prior work has studied such verbatim memorization using observational data. To complement such work, we…

计算与语言 · 计算机科学 2024-07-26 Jing Huang , Diyi Yang , Christopher Potts

Fill-in-the-Middle (FIM), or infilling, has become integral to code language models, enabling generation of missing code given both left and right contexts. However, the current FIM training paradigm which performs next-token prediction…

机器学习 · 计算机科学 2025-11-20 Yifeng Ding , Hantian Ding , Shiqi Wang , Qing Sun , Varun Kumar , Zijian Wang

Language models have become the backbone of today's AI systems. However, their predominant left-to-right generation limits the use of bidirectional context, which is essential for tasks that involve filling text in the middle. We propose…

计算与语言 · 计算机科学 2023-10-17 Tianxiao Shen , Hao Peng , Ruoqi Shen , Yao Fu , Zaid Harchaoui , Yejin Choi

Memorization, or the tendency of large language models (LLMs) to output entire sequences from their training data verbatim, is a key concern for safely deploying language models. In particular, it is vital to minimize a model's memorization…

The ability of large language models (LLMs) to recall and retrieve information from long contexts is critical for many real-world applications. Prior work (Liu et al., 2023) reported that LLMs suffer significant drops in retrieval accuracy…

信息检索 · 计算机科学 2025-11-11 Max McKinnon

As new knowledge rapidly accumulates, language models (LMs) with pretrained knowledge quickly become obsolete. A common approach to updating LMs is fine-tuning them directly on new knowledge. However, recent studies have shown that…

计算与语言 · 计算机科学 2025-02-28 Howard Chen , Jiayi Geng , Adithya Bhaskar , Dan Friedman , Danqi Chen

Protein language models (pLMs), pre-trained via causal language modeling on protein sequences, have been a promising tool for protein sequence design. In real-world protein engineering, there are many cases where the amino acids in the…

机器学习 · 计算机科学 2023-03-30 Youhan Lee , Hasun Yu

Large Language Models have received significant attention due to their abilities to solve a wide range of complex tasks. However these models memorize a significant proportion of their training data, posing a serious threat when disclosed…

密码学与安全 · 计算机科学 2025-07-16 Jérémie Dentan , Davide Buscaldi , Aymen Shabou , Sonia Vanier

Studying data memorization in neural language models helps us understand the risks (e.g., to privacy or copyright) associated with models regurgitating training data and aids in the development of countermeasures. Many prior works -- and…

Most language models (LMs) are trained and applied in an autoregressive left-to-right fashion, assuming that the next token only depends on the preceding ones. However, this assumption ignores the potential benefits of using the full…

计算与语言 · 计算机科学 2023-03-14 Anh Nguyen , Nikos Karampatziakis , Weizhu Chen

The fine-tuning of large vision-language foundation models remains an underexplored area, particularly regarding its impact on learning gains and catastrophic forgetting. Inspired by the significance of modality gaps in contrastive…

计算机视觉与模式识别 · 计算机科学 2025-08-05 Laura Niss , Kevin Vogt-Lowell , Theodoros Tsiligkaridis
‹ 上一页 1 2 3 10 下一页 ›