use pipeline instead of endpoint

.gitignore

@@ -2,4 +2,5 @@
 *.pyc
 __pycache__
 .env
-*.env
+*.env
+.idea

inference.py

@@ -4,6 +4,8 @@ import io
 import config
 import random
 
+from diffusers import AutoPipelineForText2Image
+import torch
 
 class DiffusionInference:
     def __init__(self, api_key=None):
@@ -64,7 +66,7 @@ class DiffusionInference:
         
         try:
             # Call the API with all parameters as kwargs
-            image = self.client.text_to_image(**params)
+            image = self.run_text_to_image_pipeline(**params)
             return image
         except Exception as e:
             print(f"Error generating image: {e}")
@@ -159,3 +161,8 @@ class DiffusionInference:
                     os.remove(temp_file)
                 except Exception as e:
                     print(f"Warning: Could not delete temporary file {temp_file}: {e}")
+
+    def run_text_to_image_pipeline(self,model_name, **kwargs):
+        pipeline = AutoPipelineForText2Image.from_pretrained(model=model_name, torch_dtype=torch.float16).to("cuda")
+        image = pipeline(**kwargs).images[0]
+        return image