Remove past_key_value blocked by transformers cache error huggingface…

…/transformers#34678

node-hub/dora-qwen2-5-vl/dora_qwen2_5_vl/main.py

@@ -32,13 +32,13 @@
 IMAGE_WIDTH = int(
     os.getenv(
         "IMAGE_WIDTH",
-        "320",
+        "0",
     )
 )
 IMAGE_HEIGHT = int(
     os.getenv(
         "IMAGE_HEIGHT",
-        "225",
+        "0",
     )
 )
 HISTORY = os.getenv("HISTORY", "False") in ["True", "true"]
@@ -104,10 +104,12 @@ def generate(frames: dict, question, history, past_key_values=None):
     inputs = inputs.to(model.device)
 
     # Inference: Generation of the output
+    ## TODO: Add past_key_values to the inputs when https://github.com/huggingface/transformers/issues/34678 is fixed.
     outputs = model.generate(
-        **inputs, max_new_tokens=128, past_key_values=past_key_values
+        **inputs,
+        max_new_tokens=128,  # past_key_values=past_key_values
     )
-    past_key_values = outputs.past_key_values
+    # past_key_values = outputs.past_key_values
 
     generated_ids_trimmed = [
         out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, outputs)
@@ -201,7 +203,10 @@ def main():
                 else:
                     raise RuntimeError(f"Unsupported image encoding: {encoding}")
                 image = Image.fromarray(frame)
-                frames[event_id] = image.resize((IMAGE_HEIGHT, IMAGE_WIDTH))
+                if IMAGE_HEIGHT > 0 and IMAGE_WIDTH > 0:
+                    frames[event_id] = image.resize((IMAGE_HEIGHT, IMAGE_WIDTH))
+                else:
+                    frames[event_id] = image
 
             elif "text" in event_id:
                 if len(event["value"]) > 0: