fix for tests

src/diffusers/models/transformers/transformer_chroma.py

@@ -418,6 +418,7 @@ class ChromaTransformer2DModel(
         attention_head_dim: int = 128,
         num_attention_heads: int = 24,
         joint_attention_dim: int = 4096,
+        pooled_projection_dim: int = 768,
         axes_dims_rope: Tuple[int, ...] = (16, 56, 56),
         approximator_num_channels: int = 64,
         approximator_hidden_dim: int = 5120,

src/diffusers/pipelines/chroma/pipeline_chroma.py

@@ -219,25 +219,23 @@ class ChromaPipeline(
 
         text_inputs = self.tokenizer(
             prompt,
-            padding=False,
+            padding=True,
             max_length=max_sequence_length,
             truncation=True,
             return_length=False,
             return_overflowing_tokens=False,
             return_tensors="pt",
         )
-        pad_token_id = self.tokenizer.pad_token_id
-        text_input_ids = torch.cat(
-            [
-                text_inputs.input_ids,
-                torch.full((text_inputs.input_ids.size(0), 1), pad_token_id, dtype=text_inputs.input_ids.dtype),
-            ],
-            dim=1,
-        )
+        text_input_ids = text_inputs.input_ids
+        attention_mask = text_inputs.attention_mask.clone()
+
+        # Chroma requires the attention mask to include one padding token
+        seq_lengths = attention_mask.sum(dim=1)
+        mask_indices = torch.arange(attention_mask.size(1)).unsqueeze(0).expand(batch_size, -1)
+        attention_mask = (mask_indices <= seq_lengths.unsqueeze(1)).long()
 
         prompt_embeds = self.text_encoder(
-            text_input_ids.to(device),
-            output_hidden_states=False,
+            text_input_ids.to(device), output_hidden_states=False, attention_mask=attention_mask.to(device)
         )[0]
 
         dtype = self.text_encoder.dtype

tests/pipelines/chroma/chroma.py

@@ -19,7 +19,6 @@ from ..test_pipelines_common import (
     FasterCacheTesterMixin,
     FluxIPAdapterTesterMixin,
     PipelineTesterMixin,
-    PyramidAttentionBroadcastTesterMixin,
     check_qkv_fusion_matches_attn_procs_length,
     check_qkv_fusion_processors_exist,
 )
@@ -29,11 +28,10 @@ class ChromaPipelineFastTests(
     unittest.TestCase,
     PipelineTesterMixin,
     FluxIPAdapterTesterMixin,
-    PyramidAttentionBroadcastTesterMixin,
     FasterCacheTesterMixin,
 ):
     pipeline_class = ChromaPipeline
-    params = frozenset(["prompt", "height", "width", "guidance_scale", "prompt_embeds", "pooled_prompt_embeds"])
+    params = frozenset(["prompt", "height", "width", "guidance_scale", "prompt_embeds"])
     batch_params = frozenset(["prompt"])
 
     # there is no xformers processor for Flux
@@ -46,7 +44,7 @@ class ChromaPipelineFastTests(
         spatial_attention_timestep_skip_range=(-1, 901),
         unconditional_batch_skip_range=2,
         attention_weight_callback=lambda _: 0.5,
-        is_guidance_distilled=True,
+        is_guidance_distilled=False,
     )
 
     def get_dummy_components(self, num_layers: int = 1, num_single_layers: int = 1):
@@ -182,3 +180,9 @@ class ChromaPipelineFastTests(
             image = pipe(**inputs).images[0]
             output_height, output_width, _ = image.shape
             assert (output_height, output_width) == (expected_height, expected_width)
+
+    @unittest.skip(
+        "Chroma uses Flux encode_prompt but uses CFG. This test is incompatible with the pipeline since the test case does not use a negative prompt embeds"
+    )
+    def test_encode_prompt_works_in_isolation(self):
+        pass

tests/pipelines/test_pipelines_common.py

@@ -307,6 +307,7 @@ class IPAdapterTesterMixin:
 
         # forward pass without ip adapter
         inputs = self._modify_inputs_for_ip_adapter_test(self.get_dummy_inputs(torch_device))
+        __import__("ipdb").set_trace()
         if expected_pipe_slice is None:
             output_without_adapter = pipe(**inputs)[0]
         else:
@@ -521,7 +522,8 @@ class FluxIPAdapterTesterMixin:
 
     def _modify_inputs_for_ip_adapter_test(self, inputs: Dict[str, Any]):
         inputs["negative_prompt"] = ""
-        inputs["true_cfg_scale"] = 4.0
+        if "true_cfg_scale" in inspect.signature(self.pipeline_class.__call__).parameters:
+            inputs["true_cfg_scale"] = 4.0
         inputs["output_type"] = "np"
         inputs["return_dict"] = False
         return inputs