Create app.py

app.py

@@ -0,0 +1,44 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
+import torch
+
+MODEL_NAME = "rahul-shrivastav/medical-qna"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_NAME,
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+def generate_response(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+
+    generation_config = GenerationConfig(
+        do_sample=True,
+        top_k=50,
+        temperature=0.7,
+        max_new_tokens=200,
+        pad_token_id=tokenizer.eos_token_id
+    )
+
+    outputs = model.generate(**inputs, generation_config=generation_config)
+    text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return {"response": text}
+
+# API mode only — no UI
+demo = gr.Interface(
+    fn=generate_response,
+    inputs=gr.Textbox(),
+    outputs="json",
+    allow_flagging="never"
+)
+
+if __name__ == "__main__":
+    demo.launch(
+        server_name="0.0.0.0",  # Needed for Spaces to accept incoming requests
+        server_port=7860
+        # enable_api=True,     # allows /gradio_api calls
+        # allow_flagging="never",  # no flag button in UI
+        # share=True           # optional, for public link
+    )