add mask

app.py

@@ -101,7 +101,7 @@ def gr_caption(x):
         return np.clip(img, a_min=0., a_max=1.)
 
     masked, masked_recon, recon, latent = visual_recon(x, model)
-    caption_from_model = caption(10, latent, model, tokenizer, )
+    caption_from_model = caption(8, latent, model, tokenizer, )
     
     masked, masked_recon, recon = map(unnorm_pix, (masked, masked_recon, recon))