Unverified50% confidenceFactExact time
Transformer架构最初由Vaswani等人于2017年为NLP设计，后被迁移到视觉（ViT）、音频、蛋白质结构预测等几乎所有AI子领域
1
Sources
50%
Confidence
Long-term
Relevance
8/7/2026
First Seen
Sources
Related Claims
VerifiedTransformer has expanded from NLP into computer vision (ViT), speech recognition (Whisper), multimodal models (CLIP, GPT-4V), and even protein structure prediction (AlphaFold2).72% similarUnverifiedStarting from GLM 5.x, Zhipu AI's architectural design has gradually shifted toward the industry-mainstream Transformer Decoder-Only paradigm.71% similarUnverifiedThe Transformers framework already supports hundreds of model architectures, covering NLP, CV, Audio, Multimodal, and other directions.71% similarUnverifiedDiffusion models based on the Transformer architecture have gradually replaced GANs as the mainstream approach for deepfakes.64% similarUnverifiedSwitch Transformer and the Mixtral series have widely adopted the MoE architecture.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/704244API
curl https://kongchang.com/api/v1/knowledge/claims/704244MCP
get_claim(id=704244)