Giter Site home page Giter Site logo

Comments (2)

albertz avatar albertz commented on May 24, 2024

Yes, that happens when the target seq len is longer than the source len (which is the case initially for some cases because of a high reduction factor). If it happens every single step, something is wrong, but if it happens occasionally, this is fine. Can you just continue running the training and see if it works? In later epochs, it should not happen anymore (or much rarely).

Best,
Albert

from returnn-experiments.

akshatdewan avatar akshatdewan commented on May 24, 2024

Hi Albert,

Many thanks for the explanation. It makes sense now. I ran the training and it finished 32 epochs of pretraining and the subject mentioned messages are much more rare now. I am actually training on a dataset that is 2 times the size of librispeech (librispeech + proprietary data). I tested one of the models (after 29 epochs) by running 23_recog.sh and I get meaningful hypotheses. I will let it run for 100 epochs and see how it goes.

Thanks for making this toolkit available and your support.
Akshat

from returnn-experiments.

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.