Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Finish e2e GPU memory fix #9019

Closed
4 tasks
twerkmeister opened this issue Jul 5, 2021 · 1 comment
Closed
4 tasks

Finish e2e GPU memory fix #9019

twerkmeister opened this issue Jul 5, 2021 · 1 comment
Labels
area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning research:end-to-end type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR

Comments

@twerkmeister
Copy link
Contributor

twerkmeister commented Jul 5, 2021

The prototypical GPU memory fix has been tested and solves the problem that the GPU runs out of memory for large action sets in e2e scenarios.

Implementation issue for #8734

To make it production ready, the following things need to be addressed / Definition of done:

  • code quality (comments, minor refactorings / renaming)
  • tests are added for not running into memory problems again
  • label_batch_size should be turned into a boolean variable balanced_sampling_negative_labels to remove unnecessary complexity in the interplay between label_batch_size and number_of_negative_examples
  • Sampling inside DotProductLossLayer should be unique as well using uniform_candidate_sampler
@twerkmeister twerkmeister added type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning research:end-to-end labels Jul 5, 2021
@rasabot-exalate rasabot-exalate added area:rasa-oss and removed type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning labels Mar 15, 2022 — with Exalate Issue Sync
@m-vdb m-vdb added area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning and removed area:rasa-oss labels Mar 16, 2022
@rasabot rasabot added type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR and removed type:enhancement labels Mar 16, 2022
@rasabot-exalate rasabot-exalate added area:rasa-oss area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning and removed type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning area:rasa-oss labels Mar 17, 2022 — with Exalate Issue Sync
@rasabot-exalate rasabot-exalate added the type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR label Mar 21, 2022 — with Exalate Issue Sync
@m-vdb
Copy link
Collaborator

m-vdb commented Dec 8, 2022

closing as stale

@m-vdb m-vdb closed this as not planned Won't fix, can't repro, duplicate, stale Dec 8, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
area:rasa-oss 🎡 Anything related to the open source Rasa framework area:rasa-oss/ml 👁 All issues related to machine learning research:end-to-end type:enhancement ✨ Additions of new features or changes to existing ones, should be doable in a single PR
Projects
None yet
Development

No branches or pull requests

4 participants