Merge branch 'main' into flux-control-lora-training-script

Author: sayakpaul

src/diffusers/loaders/lora_pipeline.py

@@ -2313,7 +2313,7 @@ def _maybe_expand_transformer_param_shape_or_error_(
         for name, module in transformer.named_modules():
             if isinstance(module, torch.nn.Linear):
                 module_weight = module.weight.data
-                module_bias = module.bias.data if hasattr(module, "bias") else None
+                module_bias = module.bias.data if module.bias is not None else None
                 bias = module_bias is not None
 
                 lora_A_weight_name = f"{name}.lora_A.weight"

src/diffusers/pipelines/stable_diffusion_gligen/pipeline_stable_diffusion_gligen_text_image.py

@@ -446,13 +446,14 @@ def prepare_extra_step_kwargs(self, generator, eta):
             extra_step_kwargs["generator"] = generator
         return extra_step_kwargs
 
-    # Copied from diffusers.pipelines.stable_diffusion_k_diffusion.pipeline_stable_diffusion_k_diffusion.StableDiffusionKDiffusionPipeline.check_inputs
     def check_inputs(
         self,
         prompt,
         height,
         width,
         callback_steps,
+        gligen_images,
+        gligen_phrases,
         negative_prompt=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
@@ -499,6 +500,13 @@ def check_inputs(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+        if gligen_images is not None and gligen_phrases is not None:
+            if len(gligen_images) != len(gligen_phrases):
+                raise ValueError(
+                    "`gligen_images` and `gligen_phrases` must have the same length when both are provided, but"
+                    f" got: `gligen_images` with length {len(gligen_images)} != `gligen_phrases` with length {len(gligen_phrases)}."
+                )
+
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
     def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
         shape = (
@@ -814,6 +822,8 @@ def __call__(
             height,
             width,
             callback_steps,
+            gligen_images,
+            gligen_phrases,
             negative_prompt,
             prompt_embeds,
             negative_prompt_embeds,