Unverified50% confidenceFactExact time
Each forward pass of the original Transformer has a fixed computational depth that cannot be dynamically extended, making it naturally correspond to System 1.
1
Sources
50%
Confidence
Long-term
Relevance
10/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSet Transformer通过去掉标准Transformer中的位置编码来保持排列不变性73% similarUnverifiedBy reusing the same weight matrices, the Recurrent Transformer can achieve deeper levels of information processing without significantly increasing the number of parameters.72% similarUnverified在Transformer中残差流的写入操作是线性叠加（加法），使不同模块的贡献保持可分离性，是机制可解释性追踪信息来源的数学基础71% similarUnverifiedRecurrent Looped Transformer proposes having part of the model's layer structure be invoked repeatedly in a recurrent manner, providing variable reasoning depth without increasing the number of parameters.71% similarUnverifiedAutoregressive normalizing flows are essentially autoregressive Transformers.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/973486API
curl https://kongchang.com/api/v1/knowledge/claims/973486MCP
get_claim(id=973486)