Fastformer_Initial_Params_Paddle

Name: Fastformer_Initial_Params_Paddle
Creator: rainyBJ
Published: 2025-08-23T11:41:17+08:00

AI_Studio2021-11-03 更新2025-08-23 收录

Transformer优化

注意力机制

数据链接：

https://aistudio.baidu.com/datasetdetail/114900 数据链接链接失效反馈

官方服务：

资源简介：

Fastformer: Additive Attention is All You Need

《Fastformer：加性注意力（Additive Attention）即为你所需的一切》

提供机构：

该数据集用于Flip-Flop Language Modeling任务，旨在正确执行1位寄存器的顺序操作。尽管Transformer架构似乎适合此操作，但它会偶尔出现外推错误（称为注意力故障）。一个开放的挑战是如何在不依赖长尾数据或递归架构的情况下修复这些错误。数据集包含训练集、验证集、密集验证集和稀疏验证集，分别来自不同的FFL配置。

Hugging Face2023-10-19 更新240

Universal Approximation and Optimization Theory for Multi-Head Self-Attention: Theoretical Foundations and Scaling Laws

深度学习

注意力机制

We have established comprehensive theoretical foundations for understanding multi-head self-attention mechanisms through universal approximation theory and convex optimization. Our analysis reveals th

DataCite Commons2025-07-28 更新70

Performance comparison on COCO2017 dataset.

小目标检测

Transformer优化

Oriented small object detection remains a challenging problem in computer vision, largely due to the weak feature representation and high computational cost of existing detection Transformer (DETR)-ba

NIAID Data Ecosystem40

FLUX attention data

注意力机制

机器学习

Attention data st for FLUX pipeline

DataCite Commons2021-09-17 更新80

Testing the Attentional Dwelling Hypothesis of Attentional Capture

感知与认知心理学

注意力机制

Data sets for the two experiments published in "Testing the Attentional Dwelling Hypothesis of Attentional Capture" by Lamy D., Darnell. M., Levy A. and Bublil, C. in `journal Of Cognition (2018)

DataCite Commons2020-08-28 更新70

Fastformer_Initial_Params_Paddle

资源简介：

相关数据集