Fix typo

README.md

@@ -12,7 +12,7 @@ Paella is a novel text-to-image model that uses a compressed quantized latent sp
 
 The models in these repo refer to the "Arroz con Cosas" variation of Paella that provide:
 * A clip2img model, to turn CLIP image embeddings into images.
-* A prioir to transform CLIP text embeddings into CLIP image embeddings.
+* A prior to transform CLIP text embeddings into CLIP image embeddings.
 * A custom VQ-GAN trained on 7K watercolor images.
 * A Paella generator model.