Unverified60% confidenceFactExact time
The facts absorbed by large language models during the pre-training phase are encoded in a distributed manner across billions of parameters, with a single fact diffused across the activation patterns of hundreds or even thousands of neurons
2
Sources
60%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
反向微调实验：AI如何被训练相信东京不是日本首都
hackernewshackernews7/5/2026
Related Claims
VerifiedLarge language models do not store raw text in the traditional sense during training; instead, they encode statistical patterns into billions of parameters through the backpropagation algorithm.82% similarUnverifiedParametric memory is knowledge solidified into model weights during the training phase, encoded across billions or even trillions of parameters.78% similarUnverified大语言模型的参数量从数十亿到数千亿不等，本质上是对训练数据中统计规律的压缩存储76% similarUnverified主流大语言模型基于Transformer架构，通过预训练将知识压缩进数十亿乃至数千亿的模型权重中76% similarVerifiedThe inference process of large language models consumes as much computational resources as the training process.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/177037API
curl https://kongchang.com/api/v1/knowledge/claims/177037MCP
get_claim(id=177037)