huggingface / huggingface/diffusers

UNet3DConditionModel does not use class conditioning

Offen
#7,053 7 Kommentare 0 Reaktionen 0 zugewiesene Personen Auf GitHub ansehen
bug stale
Vorherrschende Sprache
Python
Sterne
34.5k
Forks
7.3k
Ø Merge
3 T. 3 Std.
Gemergte PRs (30 T.)
91

Beschreibung

### Describe the bug

Hello guys,

I found a small inconsistency. In `UNet3DConditionModel` forward method, `class_labels` is [defined](https://github.com/huggingface/diffusers/blob/6fedbd850a282ef21009f99a394f619ed7420137/src/diffusers/models/unets/unet_3d_condition.py#L540C13-L540C25) but is never used after that.

This is different from its twin model `UNet2DConditionModel`, where it is [used](https://github.com/huggingface/diffusers/blob/6fedbd850a282ef21009f99a394f619ed7420137/src/diffusers/models/unets/unet_2d_condition.py#L992) to get an embedding which is later used by the model.

Cheers.

### Reproduction

Does not apply.

### Logs

_No response_

### System Info

Does not apply.

### Who can help?

_No response_

Beitragsleitfaden

Beitragsleitfaden öffnen

Rechercherichtung

Start in src/diffusers/models/unets/unet_3d_condition.py at the UNet3DConditionModel forward method and compare its class_labels handling with the linked UNet2DConditionModel implementation. Trace how the class embedding is obtained and consumed in the 2D model. Done means class conditioning is handled consistently in the 3D model and the relevant behavior is covered by the project's existing model tests.

Vom Indexierungsmodell aus dem Issue-Text verfasst.

Bewertung

Tech-Stack
python, pytorch
Bereich
machine-learning
Issue-Typ
Bug
Schwierigkeit
3/5
Geschätzter Aufwand
1-2 Tage
Aktivitätsstatus
Veraltet
Klarheit
Größtenteils klar
Anfängerfreundlichkeit
45/100

Neue Issues direkt in Ihr Postfach

Eine kurze Übersicht über anfängerfreundliche GitHub-Issues.