Create app.py

app.py

@@ -0,0 +1,59 @@
+# app.py
+import gradio as gr
+import torch
+from transformers import AutoProcessor, AutoModelForVision2Seq
+
+# -----------------------------------------------------------------------------
+# Model initialization
+# -----------------------------------------------------------------------------
+MODEL_ID = "RedHatAI/Qwen2.5-VL-3B-Instruct-quantized.w8a8"
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+dtype = torch.float16 if device == "cuda" else torch.float32
+
+print(f"🚀 Loading model on {device} ...")
+processor = AutoProcessor.from_pretrained(MODEL_ID)
+model = AutoModelForVision2Seq.from_pretrained(
+    MODEL_ID,
+    torch_dtype=dtype,
+    device_map="auto" if device == "cuda" else None
+)
+model.eval()
+print("✅ Model ready!")
+
+# -----------------------------------------------------------------------------
+# Inference function
+# -----------------------------------------------------------------------------
+def ask_about_image(image, prompt):
+    """
+    Take an uploaded image and user question, run vision-language inference,
+    and return model's generated text.
+    """
+    if image is None or prompt.strip() == "":
+        return "Please upload an image and enter a question."
+
+    # Prepare input
+    inputs = processor(images=image, text=prompt, return_tensors="pt").to(device, dtype)
+
+    # Generate answer
+    with torch.no_grad():
+        generated_ids = model.generate(**inputs, max_new_tokens=256)
+    answer = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    return answer.strip()
+
+# -----------------------------------------------------------------------------
+# Gradio UI
+# -----------------------------------------------------------------------------
+with gr.Blocks(title="🧠 Qwen2.5-VL Image Q&A") as demo:
+    gr.Markdown("## 🧩 Qwen 2.5 VL – 3B Instruct (Quantized)\nAsk questions about any uploaded image.")
+    with gr.Row():
+        img = gr.Image(type="pil", label="Upload Image")
+        with gr.Column():
+            prompt = gr.Textbox(label="Enter your question", placeholder="e.g. What’s written on the sign?")
+            ask_btn = gr.Button("Ask")
+    output = gr.Textbox(label="Model Response", lines=5)
+
+    ask_btn.click(ask_about_image, [img, prompt], output)
+    prompt.submit(ask_about_image, [img, prompt], output)
+
+demo.launch()