Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911510108831744 |
|---|---|
| author | Zhuang, Xialie Jia, Zhikai Li, Jianjin Zhang, Zhenyu Shen, Li Cao, Zheng Liu, Shiwei |
| author_facet | Zhuang, Xialie Jia, Zhikai Li, Jianjin Zhang, Zhenyu Shen, Li Cao, Zheng Liu, Shiwei |
| contents | Large Language Models (LLMs) are discovered to suffer from accurately retrieving key information. To address this, we propose Mask-Enhanced Autoregressive Prediction (MEAP), a simple yet effective training paradigm that seamlessly integrates Masked Language Modeling (MLM) into Next-Token Prediction (NTP) to enhance the latter's in-context retrieval capabilities. Specifically, MEAP first randomly masks a small fraction of input tokens and then directly performs the standard next-token prediction autoregressive using a decoder-only Transformer. MEAP eliminates the need for bidirectional attention or encoder-decoder architectures for MLM, incurring no additional computational overhead during pre-training or inference. Intensive experiments demonstrate that MEAP substantially outperforms NTP on key information retrieval and long-context reasoning tasks, while performing on par or better on commonsense reasoning tasks. The benefits of MEAP also extend to supervised fine-tuning, where it shows remarkable advantages in lost-in-the-middle scenarios, outperforming NTP by 11.77 percentage points. Our analysis indicates that MEAP's effectiveness arises from its ability to promote more distinguishable attention scores by concentrating on a reduced set of non-masked tokens. This mechanism improves the model's focus on task-relevant signals while mitigating the influence of peripheral context. These findings position MEAP as a promising training paradigm for large language models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_07490 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More Zhuang, Xialie Jia, Zhikai Li, Jianjin Zhang, Zhenyu Shen, Li Cao, Zheng Liu, Shiwei Computation and Language Machine Learning Large Language Models (LLMs) are discovered to suffer from accurately retrieving key information. To address this, we propose Mask-Enhanced Autoregressive Prediction (MEAP), a simple yet effective training paradigm that seamlessly integrates Masked Language Modeling (MLM) into Next-Token Prediction (NTP) to enhance the latter's in-context retrieval capabilities. Specifically, MEAP first randomly masks a small fraction of input tokens and then directly performs the standard next-token prediction autoregressive using a decoder-only Transformer. MEAP eliminates the need for bidirectional attention or encoder-decoder architectures for MLM, incurring no additional computational overhead during pre-training or inference. Intensive experiments demonstrate that MEAP substantially outperforms NTP on key information retrieval and long-context reasoning tasks, while performing on par or better on commonsense reasoning tasks. The benefits of MEAP also extend to supervised fine-tuning, where it shows remarkable advantages in lost-in-the-middle scenarios, outperforming NTP by 11.77 percentage points. Our analysis indicates that MEAP's effectiveness arises from its ability to promote more distinguishable attention scores by concentrating on a reduced set of non-masked tokens. This mechanism improves the model's focus on task-relevant signals while mitigating the influence of peripheral context. These findings position MEAP as a promising training paradigm for large language models. |
| title | Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2502.07490 |