search

facebookresearch / d2go

D2Go is a toolkit for efficient deep learning
Apache License 2.0
826 stars 197 forks source link

do not fuse model again for a QAT model #643

Closed stephenyan1231 closed 6 months ago

stephenyan1231 commented 6 months ago

Summary: For a QAT model, it contains observers. After QAT training, those observers already contain updated statistics, such as min_val, max_val.

When we want to export FP32 QAT model for a sanity check, if we call fuse_utils.fuse_model() again (which is often already called when we build the QAT model before QAT training), it will remove statistics in the observers.

Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request was exported from Phabricator. Differential Revision: D52152688

facebook-github-bot commented 6 months ago

This pull request has been merged in facebookresearch/d2go@8f130231ebb783cb1e353bdd78f22efcd9ae1ec7.