PARSER：并行读取深度推理的长上下文Agent架构
This is a brilliant paper.
长上下文Agent的新范式，用并行读取+RL训练主代理解决了顺序处理的位置敏感和延迟问题，数据扎实，值得研究长文本Agent的同学关注。
This is a brilliant paper.
这是一篇出色的论文。
It's of the cleanest long-context agent designs I have seen in the past couple of months.
这是我过去几个月见过的最简洁的长上下文智能体设计之一。
Sequential memory agents read chunks one after another while maintaining a compact memory state.
顺序记忆智能体在保持紧凑记忆状态的同时，依次读取各个文本块。
This behavior ties reasoning depth to document traversal and makes accuracy sensitive to where the evidence sits. It also makes latency grow linearly with document length.
这种机制将推理深度与文档遍历过程绑定在一起，使得准确率对证据所在的位置非常敏感，同时也导致延迟随文档长度呈线性增长。
PARSER decouples the two.
PARSER 将这两者解耦。
A bank of lightweight subagents, each bound to a single chunk, reads the whole document in parallel.
一组轻量级子智能体（每个子智能体绑定到一个单独的文本块）并行读取整个文档。
A lead agent reasons through iterative scatter-gather rounds, broadcasting a query to all subagents, aggregating the returned evidence, and forming a deeper follow-up query conditioned on what it has found.
一个主智能体通过迭代式的散射-收集（scatter-gather）轮次进行推理：向所有子智能体广播查询，聚合返回的证据，并根据已发现的内容形成更深层次的后续查询。
All the learnable behavior is build into the lead agent, which is trained with RL. The subagents stay frozen off-the-shelf models.
所有可学习的行为都内置于主智能体中，并通过强化学习（RL）进行训练。子智能体则保持为冻结状态的现成模型。
On multi-hop QA from 7K to 896K tokens, a 4B PARSER beats the strongest sequential memory baseline by 5.7 points on average and 12.0 points at 896K. At 9B it passes DeepSeek-V4-Pro by 6.3 points.
在从 7K 到 896K token 的多跳问答任务中，4B 参数的 PARSER 平均比最强的顺序记忆基线高出 5.7 分，在 896K 长度时高出 12.0 分。在 9B 参数规模下，它比 DeepSeek-V4-Pro 高出 6.3 分。
Controlled experiments show it holds up under perturbations to evidence position, order and distance, which cause large accuracy swings in sequential methods, while cutting inference latency by up to 11x.
受控实验表明，该方法在面对证据位置、顺序和距离的扰动时依然稳健，而这些扰动会导致顺序方法出现大幅度的准确率波动，同时该方法可将推理延迟降低多达 11 倍。
Paper: https://academy.dair.ai/papers/parser-read-in-parallel-reason-in-depth-for-long-context-llm-agents-2609.06702
论文：https://academy.dair.ai/papers/parser-read-in-parallel-reason-in-depth-for-long-context-llm-agents-2609.06702
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力