Verified65% confidenceFactExact time
原始Transformer由编码器（Encoder）和解码器（Decoder）构成，BERT类模型只用编码器擅长理解和分类，GPT类模型只用解码器专注文本生成
3
Sources
65%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedTransformer的编码器负责将输入序列映射为上下文表示，适用于分类理解类任务；解码器自回归生成输出序列，适用于文本生成任务80% similarUnverifiedThe Transformer spawned two major technical paths: the encoder approach represented by BERT (excelling at understanding) and the decoder approach represented by GPT (excelling at generation), with the decoder approach becoming the dominant paradigm for large language models79% similarUnverifiedTransformer架构由Google于2017年提出，Encoder负责将输入文本编码为高维向量表示，Decoder基于编码表示逐token生成输出文本78% similarUnverified大语言模型通常采用仅解码器（Decoder-only）的Transformer变体，通过因果掩码确保每个位置只能关注其之前的token77% similarUnverifiedTransformer使用位置编码（Positional Encoding）在并行计算的同时保留词的顺序信息73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/926967API
curl https://kongchang.com/api/v1/knowledge/claims/926967MCP
get_claim(id=926967)