Comments (6)
我用2个GPU训练的时候发现score loss的收敛似乎很困难,比较慢,这个正常吗?用的是源代码,没有修改
from qdmn.
不需要调整,我在预训练没有使用衰减,不过你也可以试一下加上衰减。score loss收敛应该是比较快的,有log吗
from qdmn.
不需要调整,我在预训练没有使用衰减,不过你也可以试一下加上衰减。score loss收敛应该是比较快的,有log吗
我迭代了2000多次,前面一排是每个batch中score的值,都很小,所以我感觉它几乎没收敛
from qdmn.
我没有用tps,这应该不会影响收敛的吧
from qdmn.
我刚才跑了一下代码,两张卡的话score loss在500--600ite之间会迅速收敛,500之前收敛的会比较慢。并且score的预测会从1e-7级别慢慢上升到1e-4然后迅速到达正常输出如0.7、0.8。这是由于sigmoid函数的梯度在输出值非常接近于0时很小,逐渐上涨之后梯度变大了收敛就迅速了。2000ite还未收敛不是正常现象,或许与您模型的初始值有关系,或许您改变了部分代码。我刚才跑的是原始代码,您也可以实验一下不加修改的原始代码。目前的代码是没有问题的。
from qdmn.
我刚才跑了一下代码,两张卡的话score loss在500--600ite之间会迅速收敛,500之前收敛的会比较慢。并且score的预测会从1e-7级别慢慢上升到1e-4然后迅速到达正常输出如0.7、0.8。这是由于sigmoid函数的梯度在输出值非常接近于0时很小,逐渐上涨之后梯度变大了收敛就迅速了。2000ite还未收敛不是正常现象,或许与您模型的初始值有关系,或许您改变了部分代码。我刚才跑的是原始代码,您也可以实验一下不加修改的原始代码。目前的代码是没有问题的。
ok,谢谢您的指导
from qdmn.
Related Issues (19)
- Is the input dimension of Score 256 when QAM is used in STCN? HOT 3
- I use QAM for training, the number of iterations is 75K, but the loss only drops to about 0.5, is this normal?And how much does the final loss value of the original code drop? HOT 12
- When the QAM is used in STCN,is the iterations the same as STCN? HOT 1
- 您好,请问论文中为什么不使用L2 distance而是dot product呢,是L2 distance效果比dot product差吗 HOT 2
- 您好,注意到您的代码变了,请问是之前的代码有问题吗? HOT 2
- 您好,当我将QAM加入到STCN当中时,且输入到Score模块的特征是fuse后的,但是训练得到的score值到后面都是0了(Score模块里面全连接层的激活函数一个是relu,一个是sigmoid)我是1GPU训练,batch size为16,学习率为1e-5 HOT 7
- checkpoint file HOT 1
- train.py启动问题 HOT 5
- 关于预训练的问题 HOT 3
- Questions about pre-training HOT 19
- Do you train the whole network when you apply QAM in STCN? HOT 1
- May ask the result when you use the global average pooling layers in the code of QAM module? HOT 1
- youtube2018 HOT 2
- Adding QAM only in STM's inference stage HOT 1
- 为什么eval代码要用到label?
- The code has been reproduced, but the results are inconsistent with the provided model.
- About Score class
- About Score class
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from qdmn.