Unverified60% confidenceFactExact time
In language modeling, the Transformer is essentially learning the conditional probability distribution P(xt|x1,...,xt-1) of token sequences, and autoregressive generation is the process of sequentially sampling from this probability distribution.
2
Sources
60%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTransformer 在预训练阶段通过 Next Token Prediction 任务学习概率分布，每次输出本质是在给定上文条件下预测下一个词最可能是什么79% similarUnverifiedThe hallucination phenomenon in large models is rooted in the autoregressive generation mechanism of Transformers, where the model learns probability distributions through the Next Token Prediction task during the pre-training phase.75% similarVerified当前主流大语言模型本质上是基于Transformer架构的概率预测系统，通过预测下一个最可能的Token来生成内容75% similarVerifiedModern large language models are based on the Transformer architecture and generate text by predicting the probability distribution of the next token, with the training objective of minimizing cross-entropy loss74% similarUnverifiedDecision Transformer 于 2021 年提出，将强化学习问题重新表述为序列建模问题，使用 Transformer 架构对（回报、状态、动作）序列进行自回归预测74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/897039API
curl https://kongchang.com/api/v1/knowledge/claims/897039MCP
get_claim(id=897039)