<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[长上下文模型来了，历史消息还要不要截断]]></title><description><![CDATA[<p dir="auto">现在模型上下文都很长了，我们前端是不是可以不截历史了？反正塞得下。</p>
]]></description><link>https://localaihub.com/topic/205/长上下文模型来了-历史消息还要不要截断</link><generator>RSS for Node</generator><lastBuildDate>Wed, 03 Jun 2026 18:50:38 GMT</lastBuildDate><atom:link href="https://localaihub.com/topic/205.rss" rel="self" type="application/rss+xml"/><pubDate>Fri, 15 May 2026 03:39:00 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 22:44:00 GMT]]></title><description><![CDATA[<p dir="auto">对。这个结论比“能塞多少”重要。</p>
]]></description><link>https://localaihub.com/post/2310</link><guid isPermaLink="true">https://localaihub.com/post/2310</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Fri, 15 May 2026 22:44:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 19:57:00 GMT]]></title><description><![CDATA[<p dir="auto">所以长上下文是缓冲，不是垃圾桶。</p>
]]></description><link>https://localaihub.com/post/2309</link><guid isPermaLink="true">https://localaihub.com/post/2309</guid><dc:creator><![CDATA[小郑]]></dc:creator><pubDate>Fri, 15 May 2026 19:57:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 19:46:00 GMT]]></title><description><![CDATA[<p dir="auto">可以，但摘要也会丢细节和引入偏差。关键任务要保留可追溯原文。</p>
]]></description><link>https://localaihub.com/post/2308</link><guid isPermaLink="true">https://localaihub.com/post/2308</guid><dc:creator><![CDATA[陈一]]></dc:creator><pubDate>Fri, 15 May 2026 19:46:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 17:09:00 GMT]]></title><description><![CDATA[<p dir="auto">摘要历史可以吗？</p>
]]></description><link>https://localaihub.com/post/2307</link><guid isPermaLink="true">https://localaihub.com/post/2307</guid><dc:creator><![CDATA[小周]]></dc:creator><pubDate>Fri, 15 May 2026 17:09:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 16:55:00 GMT]]></title><description><![CDATA[<p dir="auto">如果有敏感信息，历史保留还涉及隐私，不只是成本。</p>
]]></description><link>https://localaihub.com/post/2306</link><guid isPermaLink="true">https://localaihub.com/post/2306</guid><dc:creator><![CDATA[nora]]></dc:creator><pubDate>Fri, 15 May 2026 16:55:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 16:00:00 GMT]]></title><description><![CDATA[<p dir="auto">后端也要有硬限制。不要把上下文管理完全交给前端。</p>
]]></description><link>https://localaihub.com/post/2305</link><guid isPermaLink="true">https://localaihub.com/post/2305</guid><dc:creator><![CDATA[阿航]]></dc:creator><pubDate>Fri, 15 May 2026 16:00:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 13:45:00 GMT]]></title><description><![CDATA[<p dir="auto">产品上可以做“新任务”按钮。用户换话题时，不要让旧上下文继续跟着。</p>
]]></description><link>https://localaihub.com/post/2304</link><guid isPermaLink="true">https://localaihub.com/post/2304</guid><dc:creator><![CDATA[Grace]]></dc:creator><pubDate>Fri, 15 May 2026 13:45:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 12:43:00 GMT]]></title><description><![CDATA[<p dir="auto">留当前任务目标、明确约束、最近关键状态。删寒暄、重复解释、已经完成的步骤。</p>
]]></description><link>https://localaihub.com/post/2303</link><guid isPermaLink="true">https://localaihub.com/post/2303</guid><dc:creator><![CDATA[melo]]></dc:creator><pubDate>Fri, 15 May 2026 12:43:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 10:36:00 GMT]]></title><description><![CDATA[<p dir="auto">那怎么判断该留什么？</p>
]]></description><link>https://localaihub.com/post/2302</link><guid isPermaLink="true">https://localaihub.com/post/2302</guid><dc:creator><![CDATA[普通网友A]]></dc:creator><pubDate>Fri, 15 May 2026 10:36:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 08:40:00 GMT]]></title><description><![CDATA[<p dir="auto">这叫上下文污染。模型不会自动知道哪些历史已经过期。</p>
]]></description><link>https://localaihub.com/post/2301</link><guid isPermaLink="true">https://localaihub.com/post/2301</guid><dc:creator><![CDATA[陈一]]></dc:creator><pubDate>Fri, 15 May 2026 08:40:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 07:58:00 GMT]]></title><description><![CDATA[<p dir="auto">我们之前就是全量历史，聊久了回答开始引用很早之前的废话。</p>
]]></description><link>https://localaihub.com/post/2300</link><guid isPermaLink="true">https://localaihub.com/post/2300</guid><dc:creator><![CDATA[小蓝]]></dc:creator><pubDate>Fri, 15 May 2026 07:58:00 GMT</pubDate></item><item><title><![CDATA[Reply to 长上下文模型来了，历史消息还要不要截断 on Fri, 15 May 2026 06:41:00 GMT]]></title><description><![CDATA[<p dir="auto">塞得下不代表该塞。历史越长，成本、延迟、干扰都上来。</p>
]]></description><link>https://localaihub.com/post/2299</link><guid isPermaLink="true">https://localaihub.com/post/2299</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Fri, 15 May 2026 06:41:00 GMT</pubDate></item></channel></rss>