Context Training with Active Information Seeking
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914566992035840 |
|---|---|
| author | Huang, Zeyu Kuncoro, Adhiguna Feng, Qixuan Shen, Jiajun Dery, Lucio Szlam, Arthur Ranzato, Marc'Aurelio |
| author_facet | Huang, Zeyu Kuncoro, Adhiguna Feng, Qixuan Shen, Jiajun Dery, Lucio Szlam, Arthur Ranzato, Marc'Aurelio |
| contents | Most existing large language models (LLMs) are expensive to adapt after deployment, especially when a task requires newly produced information or niche domain knowledge. Recent work has shown that, by manipulating and optimizing their context, LLMs can be tailored to downstream tasks without updating their weights. However, most existing methods remain closed-loop, relying solely on the model's intrinsic knowledge. In this paper, we equip these context optimizers with Wikipedia search and browser tools for active information seeking. We show that naively adding these tools to a standard sequential context optimization pipeline can actually degrade performance compared to baselines. However, when paired with a search-based training procedure that maintains and prunes multiple candidate contexts, active information seeking delivers consistent and substantial gains. We demonstrate these improvements across diverse domains, including low-resource translation (Flores+), health scenarios (HealthBench), and reasoning-heavy tasks (LiveCodeBench and Humanity's Last Exam). Furthermore, our method proves to be data-efficient, robust across different hyperparameters, and capable of generating effective textual contexts that generalize well across different models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_13050 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Context Training with Active Information Seeking Huang, Zeyu Kuncoro, Adhiguna Feng, Qixuan Shen, Jiajun Dery, Lucio Szlam, Arthur Ranzato, Marc'Aurelio Computation and Language Artificial Intelligence Most existing large language models (LLMs) are expensive to adapt after deployment, especially when a task requires newly produced information or niche domain knowledge. Recent work has shown that, by manipulating and optimizing their context, LLMs can be tailored to downstream tasks without updating their weights. However, most existing methods remain closed-loop, relying solely on the model's intrinsic knowledge. In this paper, we equip these context optimizers with Wikipedia search and browser tools for active information seeking. We show that naively adding these tools to a standard sequential context optimization pipeline can actually degrade performance compared to baselines. However, when paired with a search-based training procedure that maintains and prunes multiple candidate contexts, active information seeking delivers consistent and substantial gains. We demonstrate these improvements across diverse domains, including low-resource translation (Flores+), health scenarios (HealthBench), and reasoning-heavy tasks (LiveCodeBench and Humanity's Last Exam). Furthermore, our method proves to be data-efficient, robust across different hyperparameters, and capable of generating effective textual contexts that generalize well across different models. |
| title | Context Training with Active Information Seeking |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.13050 |