add Paella (Fast Text-Conditional Discrete Denoising on Vector-Quantized Latent Spaces)

src/diffusers/__init__.py

@@ -65,6 +65,7 @@
         PNDMPipeline,
         RePaintPipeline,
         ScoreSdeVePipeline,
+        PaellaTextToImagePipeline,
     )
     from .schedulers import (
         DDIMScheduler,
@@ -85,6 +86,7 @@
         ScoreSdeVeScheduler,
         UnCLIPScheduler,
         VQDiffusionScheduler,
+        PaellaScheduler,
     )
     from .training_utils import EMAModel
 

src/diffusers/pipelines/__init__.py

@@ -26,6 +26,7 @@
     from .repaint import RePaintPipeline
     from .score_sde_ve import ScoreSdeVePipeline
     from .stochastic_karras_ve import KarrasVePipeline
+    from .paella import PaellaTextToImagePipeline
 
 try:
     if not (is_torch_available() and is_librosa_available()):

src/diffusers/pipelines/paella/__init__.py

@@ -0,0 +1,5 @@
+from ...utils import is_torch_available
+
+
+if is_torch_available():
+    from .pipeline_paella import PaellaTextToImagePipeline