microsoft/nni

Does ProxylessNas implementation really support optimizing inference latency?

Open

#3.113 geöffnet am 22. Nov. 2020

Auf GitHub ansehen
 (1 Kommentar) (0 Reaktionen) (1 zugewiesene Person)Python (13.504 Stars) (1.830 Forks)batch import
NAShelp wantednew featureuser raised

Beschreibung

Environment:

  • NNI version: v1.9 (including current master)
  • NNI mode (local|remote|pai):
  • Client OS:
  • Server OS (for remote mode only):
  • Python version:
  • PyTorch/TensorFlow version:
  • Is conda/virtualenv/venv used?:
  • Is running in Docker?:

Log message:

  • nnimanager.log:
  • dispatcher.log:
  • nnictl stdout and stderr:

What issue meet, what's expected?:

The most important feature of ProxylessNas is that it can balance deployment latency and accuracy with simple regularization parameters. But this feature is clearly missing in nni. I only found loss = criterion(outputs, labels) and loss.backward() where the criterion is just a cross-entropy loss and only applicable for image classification.

Would nni team consider adding this feature? If not, would you mind I writing this feature and pulling a request?

How to reproduce it?:

Additional information:

Contributor Guide

Tech Stack
pythonpytorch
Domain
machine learningperformance
Issue Type
feature
SchwierigkeitGeschätzte Implementierungsschwierigkeit für neue Contributors, von 1 für sehr kleine Änderungen bis 5 für Expertenarbeit.
4
Geschätzte ZeitEin grober Zeitrahmen, um zu recherchieren, zu implementieren, zu testen und einen Pull Request vorzubereiten.
over 1 week
AktivitätsstatusWie verfügbar das Issue gerade wirkt: frisch, aktiv, stale, blockiert oder wartet auf Maintainer-Input.
stale
KlarheitWie klar das Issue die erwartete Änderung, Akzeptanzkriterien und den nächsten Schritt erklärt.
mostly clear
Voraussetzungen
ProxylessNAS understandingNNI NAS codebasePyTorch
EinsteigerfreundlichkeitEin Score von 1 bis 100, der schätzt, wie zugänglich dieses Issue für First-Time Contributors ist.
35
Research-Richtung
The issue requests adding latency optimization to ProxylessNAS. Currently, only cross entropy loss is used. To implement, study the original ProxylessNAS paper for latency regularization, then modify the loss function in the NNI ProxylessNAS trainer to include a latency penalty. The relevant code is likely under `nni/nas/pytorch/proxylessnas`. Maintainer response is needed to clarify scope.