Skip to content
This repository has been archived by the owner on Dec 21, 2023. It is now read-only.

Trying to create a model on a larger dataset - Loss stuck at the same number and not moving, resulting model predictions detect nothing #3465

Open
az-oolloow opened this issue Jan 3, 2023 · 0 comments

Comments

@az-oolloow
Copy link

I tried turicreate on a small-ish dataset (a few hundred images and some 20 classes) and was quite happy with the result.
I gathered my full training set to get some 7200 images, 273 classes and some tens of thousands of annotations.
When I feed it into turicreate the model always shows me 0.0311279 loss and it is never moving. No matter how many iterations I go through, resulting model doesn't detect anything when trying to get predictions.

Is there any way to understand what is going on?

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant