PulseAugur
中
实时 08:56:03
English(EN) A chat that keeps its whole history: 949,591-token prompts, calls 3.5x slower, state wrong from turn 287

Gemini 3.8 Flash聊天历史记录会降低性能和准确性

一项比较标准聊天历史记录方法与AXI架构的测试揭示了显著的性能和状态跟踪问题。在使用Gemini 3.8 Flash时,保留整个聊天历史记录导致调用速度慢了3.5倍,并且从第287轮开始变得不准确。相比之下,AXI架构通过从主机持有的状态重建提示,保持了准确性,并且随着时间的推移,提示大小的增加要小得多。这表明,尽管长上下文窗口很有吸引力,但高效管理聊天历史记录对于保持性能和可靠性至关重要。 AI

影响 突出了朴素的长上下文聊天管理在性能和准确性方面的局限性,表明需要AXI等架构解决方案来实现高效的LLM交互。

排序理由 研究比较了两种管理LLM聊天历史记录的方法。[lever_c_demoted from research: ic=1 ai=1.0]

在 dev.to — LLM tag 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

Gemini 3.8 Flash聊天历史记录会降低性能和准确性

本文如何被排名

Signal score
50 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
研究比较了两种管理LLM聊天历史记录的方法。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
infra, product
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Shalendra gaur ·

    能保存完整历史的聊天:949,591个token的提示,速度慢3.5倍,从第287轮开始状态错误

    <p><em>This article was created with the help of AI. The AXI architecture is by Rajnish Choubey. The tests were built and run by Shalendra Gaur.</em></p> <p>Long context windows make it tempting to keep the whole history and let the model sort it out. I ran that on a real job and…