Unverified50% confidenceFactExact time
Swin Transformer introduced a local window attention mechanism.
1
Sources
50%
Confidence
Long-term
Relevance
9/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTransformer架构的注意力机制决定了模型只能处理当前上下文窗口内的信息75% similarUnverified新内核为滑动窗口注意力（SWA）层提供双输入支持，并引入 attention sinks 机制70% similarUnverifiedGPU attention kernels are responsible for executing the self-attention mechanism in the Transformer architecture.70% similarUnverified各向异性现象的根源在于Transformer的自注意力机制和残差连接结构，使高频词的梯度信号主导嵌入空间整体方向68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/932001API
curl https://kongchang.com/api/v1/knowledge/claims/932001MCP
get_claim(id=932001)