fix(qwen): correct comments for copied functions in controlnet and inpaint pipelines

Author: Sunhill666

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet.py

@@ -101,7 +101,7 @@
 """
 
 
-# Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift
+# `Copied from` diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift
 def calculate_shift(
     image_seq_len,
     base_seq_len: int = 256,
@@ -239,7 +239,7 @@ def __init__(
         self.prompt_template_encode_start_idx = 34
         self.default_sample_size = 128
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.extract_masked_hidden
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline._extract_masked_hidden
     def _extract_masked_hidden(self, hidden_states: torch.Tensor, mask: torch.Tensor):
         bool_mask = mask.bool()
         valid_lengths = bool_mask.sum(dim=1)
@@ -248,7 +248,7 @@ def _extract_masked_hidden(self, hidden_states: torch.Tensor, mask: torch.Tensor
 
         return split_result
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.get_qwen_prompt_embeds
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline._get_qwen_prompt_embeds
     def _get_qwen_prompt_embeds(
         self,
         prompt: str | list[str] = None,
@@ -287,7 +287,7 @@ def _get_qwen_prompt_embeds(
 
         return prompt_embeds, encoder_attention_mask
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.encode_prompt
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline.encode_prompt
     def encode_prompt(
         self,
         prompt: str | list[str],
@@ -318,11 +318,13 @@ def encode_prompt(
         if prompt_embeds is None:
             prompt_embeds, prompt_embeds_mask = self._get_qwen_prompt_embeds(prompt, device)
 
+        prompt_embeds = prompt_embeds[:, :max_sequence_length]
         _, seq_len, _ = prompt_embeds.shape
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
         if prompt_embeds_mask is not None:
+            prompt_embeds_mask = prompt_embeds_mask[:, :max_sequence_length]
             prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_images_per_prompt, 1)
             prompt_embeds_mask = prompt_embeds_mask.view(batch_size * num_images_per_prompt, seq_len)
 

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet_inpaint.py

@@ -74,7 +74,7 @@
 """
 
 
-# Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift
+# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift
 def calculate_shift(
     image_seq_len,
     base_seq_len: int = 256,
@@ -221,7 +221,7 @@ def __init__(
         self.prompt_template_encode_start_idx = 34
         self.default_sample_size = 128
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.extract_masked_hidden
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline._extract_masked_hidden
     def _extract_masked_hidden(self, hidden_states: torch.Tensor, mask: torch.Tensor):
         bool_mask = mask.bool()
         valid_lengths = bool_mask.sum(dim=1)
@@ -230,7 +230,7 @@ def _extract_masked_hidden(self, hidden_states: torch.Tensor, mask: torch.Tensor
 
         return split_result
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.get_qwen_prompt_embeds
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline._get_qwen_prompt_embeds
     def _get_qwen_prompt_embeds(
         self,
         prompt: str | list[str] = None,
@@ -247,7 +247,7 @@ def _get_qwen_prompt_embeds(
         txt = [template.format(e) for e in prompt]
         txt_tokens = self.tokenizer(
             txt, max_length=self.tokenizer_max_length + drop_idx, padding=True, truncation=True, return_tensors="pt"
-        ).to(self.device)
+        ).to(device)
         encoder_hidden_states = self.text_encoder(
             input_ids=txt_tokens.input_ids,
             attention_mask=txt_tokens.attention_mask,
@@ -269,7 +269,7 @@ def _get_qwen_prompt_embeds(
 
         return prompt_embeds, encoder_attention_mask
 
-    # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.encode_prompt
+    # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline.encode_prompt
     def encode_prompt(
         self,
         prompt: str | list[str],
@@ -280,6 +280,7 @@ def encode_prompt(
         max_sequence_length: int = 1024,
     ):
         r"""
+
         Args:
             prompt (`str` or `list[str]`, *optional*):
                 prompt to be encoded
@@ -299,11 +300,13 @@ def encode_prompt(
         if prompt_embeds is None:
             prompt_embeds, prompt_embeds_mask = self._get_qwen_prompt_embeds(prompt, device)
 
+        prompt_embeds = prompt_embeds[:, :max_sequence_length]
         _, seq_len, _ = prompt_embeds.shape
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
         if prompt_embeds_mask is not None:
+            prompt_embeds_mask = prompt_embeds_mask[:, :max_sequence_length]
             prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_images_per_prompt, 1)
             prompt_embeds_mask = prompt_embeds_mask.view(batch_size * num_images_per_prompt, seq_len)