Unverified50% confidenceFactExact time
CLIP文本编码器的标准上下文窗口长度约为77个Token，超出部分的语义信息会被截断或降权处理
1
Sources
50%
Confidence
Long-term
Relevance
8/6/2026
First Seen
Sources
Midjourney中世纪场景Prompt技巧拆解与实战指南
Related Claims
Unverified文本条件注入通过CLIP的交叉注意力层实现，ViT-L/14版本的CLIP将文本token映射为768维向量序列76% similarUnverified早期Stable Diffusion系列以CLIP为文本编码器，对超过77个Token的提示词处理能力受限72% similarUnverified在Stable Diffusion中，CLIP文本编码器将自然语言切分为最多77个token，每个token映射为768维或1024维嵌入向量70% similarUnverifiedReading code and drawing flowcharts may consume sixty to seventy thousand Tokens, causing the context to be automatically compressed during subsequent code modifications and resulting in information loss.68% similarUnverifiedSplitting documents into chunks by a fixed character count destroys semantic integrity; intelligent semantic chunking or a sliding window strategy can ensure that each document chunk contains complete semantic information.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/699123API
curl https://kongchang.com/api/v1/knowledge/claims/699123MCP
get_claim(id=699123)