Back to AI News
arxiv

Seer: Guiding Small Deep-Research Agents with Shared Decision Models

Seer is an inference-time framework that separates a deep-research agent's action proposal from learned consequence prediction and value estimation. It trains a tool-response simulator and a value estimator offline on a shared corpus of real research trajectories, then reuses them across agents without further training. Seer executes only the selected action and adds only real tool responses to the committed history, avoiding extra tool calls for discarded candidates. Across deep-research benchmarks, it improves GAIA accuracy from 13.6% to 41.7% for Qwen3.5-4B and from 22.3% to 50.5% for Qwen3.5-9B.

AI Brief

It offers a concrete, reusable inference-time method for improving small on-device research agents without retraining or extra real tool executions.

arXiv:2611.26423v1 Announce Type: new Abstract: Deep-research agents repeatedly decide what to search, read, and verify, often using large cloud-hosted language models for action proposal. However, privacy and local control motivate running agents on users' devices, where limited memory and compute favor smaller models and complicate on-device training. Yet inference-time guidance faces three obstacles: exploring alternatives through real execution adds tool calls, prompted self-evaluation is bounded by the agent's judgment, and model-specific guidance risks repeated adaptation costs. We introduce Seer, an inference-time framework providing local guidance that separates the agent's action proposal from learned consequence prediction and value estimation. From a shared corpus of real research trajectories, we train two models: a tool-response simulator on observed call--response pairs and a value estimator on histories labeled with discounted terminal correctness. The simulator and estimator are trained offline and reused across agents without further training. Moreover, Seer executes only the selected action and adds only real tool responses to the committed history, avoiding tool calls on discarded candidates at additional inference cost. Across multiple deep-research benchmarks, Seer substantially improves agent performance for different language models. For instance, GAIA accuracy rises from 13.6% to 41.7% for Qwen3.5-4B and from 22.3% to 50.5% for Qwen3.5-9B. Furthermore, our analyses suggest that broader candidate coverage can compensate for imperfect simulation, while prediction and scoring errors limit the benefits of wider imagined search. These findings highlight the potential of reusable models for prediction and evaluation to strengthen small research agents through extra local computation without the real execution of the discarded candidates.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!