Code Monkey home page Code Monkey logo

Comments (4)

dutran avatar dutran commented on July 22, 2024

Try to reduce lr by 3 (as you reduce batch size by a factor of 3). This looks good. Random loss is ln(101) ~ 4.62.

from c3d.

whjxnyzh123 avatar whjxnyzh123 commented on July 22, 2024

Thank you very much @dutran I will try it out

from c3d.

whjxnyzh123 avatar whjxnyzh123 commented on July 22, 2024

Great, the loss is very small now

I0529 08:51:22.795929  7108 solver.cpp:237] Iteration 20020, lr = 0.0001
I0529 08:51:22.796392  7108 solver.cpp:87] Iteration 20020, loss = 0.00696523
I0529 08:52:00.631763  7108 solver.cpp:237] Iteration 20040, lr = 0.0001
I0529 08:52:00.632230  7108 solver.cpp:87] Iteration 20040, loss = 3.70235e-05
I0529 08:52:38.457860  7108 solver.cpp:237] Iteration 20060, lr = 0.0001
I0529 08:52:38.458324  7108 solver.cpp:87] Iteration 20060, loss = 0.0188635
I0529 08:53:16.247462  7108 solver.cpp:237] Iteration 20080, lr = 0.0001
I0529 08:53:16.247931  7108 solver.cpp:87] Iteration 20080, loss = 0.0098978
I0529 08:53:54.076977  7108 solver.cpp:237] Iteration 20100, lr = 0.0001
I0529 08:53:54.077445  7108 solver.cpp:87] Iteration 20100, loss = 0.039086
I0529 08:54:31.882534  7108 solver.cpp:237] Iteration 20120, lr = 0.0001
I0529 08:54:31.883000  7108 solver.cpp:87] Iteration 20120, loss = 0.0021823
I0529 08:55:09.673733  7108 solver.cpp:237] Iteration 20140, lr = 0.0001
I0529 08:55:09.674206  7108 solver.cpp:87] Iteration 20140, loss = 0.0851032
I0529 08:55:47.512164  7108 solver.cpp:237] Iteration 20160, lr = 0.0001
I0529 08:55:47.512662  7108 solver.cpp:87] Iteration 20160, loss = 0.00216188
I0529 08:56:25.341374  7108 solver.cpp:237] Iteration 20180, lr = 0.0001
I0529 08:56:25.341886  7108 solver.cpp:87] Iteration 20180, loss = 0.00296394
I0529 08:57:03.134181  7108 solver.cpp:237] Iteration 20200, lr = 0.0001
I0529 08:57:03.134677  7108 solver.cpp:87] Iteration 20200, loss = 0.0216006
I0529 08:57:40.966392  7108 solver.cpp:237] Iteration 20220, lr = 0.0001
I0529 08:57:40.966887  7108 solver.cpp:87] Iteration 20220, loss = 0.00295561
I0529 08:58:18.788378  7108 solver.cpp:237] Iteration 20240, lr = 0.0001
I0529 08:58:18.788894  7108 solver.cpp:87] Iteration 20240, loss = 0.0404173
I0529 08:58:56.575695  7108 solver.cpp:237] Iteration 20260, lr = 0.0001
I0529 08:58:56.576161  7108 solver.cpp:87] Iteration 20260, loss = 0.0247827
I0529 08:59:34.381561  7108 solver.cpp:237] Iteration 20280, lr = 0.0001
I0529 08:59:34.382026  7108 solver.cpp:87] Iteration 20280, loss = 0.00695413
I0529 09:00:12.204666  7108 solver.cpp:237] Iteration 20300, lr = 0.0001
I0529 09:00:12.205132  7108 solver.cpp:87] Iteration 20300, loss = 0.03263
I0529 09:00:49.999734  7108 solver.cpp:237] Iteration 20320, lr = 0.0001
I0529 09:00:50.000200  7108 solver.cpp:87] Iteration 20320, loss = 0.0037585
I0529 09:01:27.809442  7108 solver.cpp:237] Iteration 20340, lr = 0.0001
I0529 09:01:27.809955  7108 solver.cpp:87] Iteration 20340, loss = 0.00100278
I0529 09:02:05.637639  7108 solver.cpp:237] Iteration 20360, lr = 0.0001
I0529 09:02:05.638101  7108 solver.cpp:87] Iteration 20360, loss = 0.0324224

from c3d.

dutran avatar dutran commented on July 22, 2024

I'm glad to hear, good luck with your experiments.

from c3d.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo 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.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.