[KongchangAI]
ModelSumOne 摘要器

SumOne

A plug-and-play memory summarizer for agents, trained via masked reconstruction and GRPO optimization. It compresses the interaction history of long-horizon agents to reduce inference cost without modifying the original agent.

Timeline (last 90 days)

Nov 5

SumOne is a standalone summarizer that preserves information in agent history through masked reconstruction training.

Unverified50%
Nov 5

SumOne uses a frozen reconstructor to recover randomly masked portions of the original trajectory from the summary, and combines reconstruction quality with a length penalty to optimize the summarizer via GRPO.

Unverified50%
Nov 5

On DeepSearchQA, SumOne reduces inference cost by about 78% compared with full context, with an absolute performance drop of only 0.04.

Unverified50%
Nov 5

The same SumOne checkpoint transfers across different downstream models and agent frameworks without additional fine-tuning.

Unverified50%
Nov 5

At inference time, SumOne retains only a lightweight summarizer and does not alter the original agent, achieving plug-and-play context compression.

Unverified50%
Nov 5

SumOne's training requires no reference summaries or downstream task execution, learning self-supervised directly from raw trajectories.

Unverified50%

All Facts (6)

Source Articles