Search results

What's the learning rate of FLatten-DeiT?

Status: Open.

#35 In LeapLabTHU/FLatten-Transformer;

· atti0127 opened on Jan 26, 2026

WindowAttention is faster than FocusedLinearAttention

Status: Open.

#34 In LeapLabTHU/FLatten-Transformer;

· ZZZENG opened on Aug 27, 2025

关于位置编码的疑问

Status: Open.

#33 In LeapLabTHU/FLatten-Transformer;

· WateverOk opened on Dec 30, 2024

将softmax注意力替换为线性注意力后，模型训练出现Nan

Status: Open.

#32 In LeapLabTHU/FLatten-Transformer;

· wzh326 opened on Dec 22, 2024

model diagram

Status: Open.

#28 In LeapLabTHU/FLatten-Transformer;

· jiangsu415 opened on Oct 14, 2024

训练代码的时候，cpu只有一个核心两个线程被调用，这是正常的吗？我的cpu是5950x，显卡是4090。

Status: Open.

#27 In LeapLabTHU/FLatten-Transformer;

· Senqi-Guan opened on Aug 22, 2024

论文中Figure 7. Accuracy-Runtime curve 是怎么画出来的？

Status: Open.

#24 In LeapLabTHU/FLatten-Transformer;

· thunderbolt-fire opened on May 1, 2024

pre-trained model about cswin

Status: Open.

#23 In LeapLabTHU/FLatten-Transformer;

· yychayitu opened on Mar 8, 2024

visualization

Status: Open.

#22 In LeapLabTHU/FLatten-Transformer;

· likepd opened on Mar 4, 2024

Hello，Could you provide more pretrained weights for PVT-V2，such as PVT_V2_b4.

Status: Open.

#20 In LeapLabTHU/FLatten-Transformer;

· Kuzu710 opened on Dec 29, 2023

Hello, I am very interested in your work.

Status: Open.

#9 In LeapLabTHU/FLatten-Transformer;

· Qudk23 opened on Oct 7, 2023

It is hoped that more pretraining parameter models can be provided for PVT-V2

Status: Open.

#8 In LeapLabTHU/FLatten-Transformer;

· bevisxia520 opened on Sep 22, 2023