Unverified50% confidenceFactExact time
BERT、GPT系列、T5等预训练大模型均以Transformer为基础架构，形成了预训练+微调的主流范式
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
VerifiedGPT stands for 'Generative Pre-trained Transformer'.78% similarUnverifiedGPT 系列、Llama 等模型的核心能力绝大部分来自预训练而非后续微调阶段76% similarUnverifiedThe GPT-5 series adopts a Mixture of Experts (MoE) architecture and reinforcement learning post-training technology.76% similarVerifiedThe Transformer architecture became the core foundation of modern large language models such as BERT (2018), the GPT series, and T5, with citation counts exceeding 100,000.76% similarUnverifiedThe current mainstream approach for training LLMs uses a decoder-only Transformer architecture, the same paradigm used by the GPT series75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/566484API
curl https://kongchang.com/api/v1/knowledge/claims/566484MCP
get_claim(id=566484)