demo

app.py

@@ -1,8 +1,6 @@
 import os
-import torch
+
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForCausalLM
-from peft import PeftModel
 
 os.environ["ROCR_VISIBLE_DEVICES"] = "0"
 os.environ["HIP_VISIBLE_DEVICES"] = "0"
@@ -11,22 +9,7 @@ os.environ["HSA_OVERRIDE_GFX_VERSION"] = "9.4.2"
 BASE_MODEL   = "Qwen/Qwen2-1.5B"
 ADAPTER_PATH = "./outputs"
 
-print("Loading tokenizer...")
-tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True)
-tokenizer.pad_token    = tokenizer.eos_token
-tokenizer.padding_side = "left"
-
-print("Loading model...")
-base = AutoModelForCausalLM.from_pretrained(
-    BASE_MODEL,
-    dtype=torch.bfloat16,
-    device_map="auto",
-    trust_remote_code=True,
-)
-model = PeftModel.from_pretrained(base, ADAPTER_PATH)
-model = model.merge_and_unload()
-model.eval()
-print("Ready!")
+
 
 EXAMPLES = [
     ["Which artery is occluded in inferior MI with ST elevation in II, III, aVF?",
@@ -46,29 +29,20 @@ EXAMPLES = [
 def answer(question, opa, opb, opc, opd):
     if not question.strip():
         return "Please enter a question."
-    if not all([opa.strip(), opb.strip(), opc.strip(), opd.strip()]):
-        return "Please fill in all four options."
-    prompt = (
-        f"### Question:\n{question}\n\n"
-        f"### Options:\n"
-        f"A) {opa}\nB) {opb}\nC) {opc}\nD) {opd}\n\n"
-        f"### Answer:\n"
-    )
-    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
-    with torch.no_grad():
-        out = model.generate(
-            **inputs,
-            max_new_tokens=200,
-            do_sample=True,
-            temperature=0.7,
-            top_p=0.9,
-            top_k=50,
-            repetition_penalty=1.3,
-            eos_token_id=tokenizer.eos_token_id,
-            pad_token_id=tokenizer.eos_token_id,
-        )
-    new = out[0][inputs["input_ids"].shape[-1]:]
-    return tokenizer.decode(new, skip_special_tokens=True)
+
+    # simple mock logic (random-ish but believable)
+    import random
+    options = [opa, opb, opc, opd]
+    letters = ["A", "B", "C", "D"]
+
+    idx = random.randint(0, 3)
+
+    return f"""Answer: {letters[idx]}) {options[idx]}
+
+Explanation:
+This is a mock demo running without the full model.
+In the real system, a fine-tuned medical LLM analyzes the clinical context
+and selects the most appropriate answer based on learned patterns."""
 
 CSS = """
 @import url('https://fonts.googleapis.com/css2?family=Syne:wght@400;600;700;800&family=DM+Sans:wght@300;400;500&display=swap');