Update main.py

main.py

@@ -3,11 +3,9 @@ from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel
 from llama_cpp import Llama
 from huggingface_hub import hf_hub_download
-import os
 
 app = FastAPI()
 
-# Enable CORS for your React/Lovable frontend
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -15,50 +13,36 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# Configuration
-REPO_ID = "devray11/Aevis-Medical-SLM"
-MODEL_FILENAME = "DeepSeek-R1-Distill-Llama-8B.Q4_K_M.gguf"
-
-print(f"🚀 Initializing Aevis Medical SLM...")
-
-# Download model from Hugging Face Hub
+# Download Aevis-Medical-SLM
 try:
-    model_path = hf_hub_download(repo_id=REPO_ID, filename=MODEL_FILENAME)
-    print(f"✅ Model downloaded to: {model_path}")
+    print("📥 Downloading model...")
+    model_path = hf_hub_download(
+        repo_id="devray11/Aevis-Medical-SLM",
+        filename="DeepSeek-R1-Distill-Llama-8B.Q4_K_M.gguf"
+    )
+    # Initialize Model (Optimized for 2-core CPU)
+    llm = Llama(model_path=model_path, n_ctx=1024, n_threads=2)
+    print("✅ Model Loaded Successfully")
 except Exception as e:
-    print(f"❌ Error downloading model: {e}")
-    model_path = None
+    print(f"❌ Error: {e}")
+    llm = None
 
-# Initialize Model with optimized CPU settings
-# n_ctx=1024 (Saves RAM), n_threads=2 (Matches HF Free Tier)
-llm = Llama(
-    model_path=model_path,
-    n_ctx=1024,
-    n_threads=2,
-    n_batch=512,
-    verbose=True
-) if model_path else None
-
-class ChatRequest(BaseModel):
+class Query(BaseModel):
     prompt: str
 
 @app.post("/generate")
-async def generate(request: ChatRequest):
+async def generate(query: Query):
     if not llm:
-        return {"error": "Model not loaded properly"}
-        
-    # Standard instruction format for fine-tuned medical models
-    prompt = f"### Instruction:\n{request.prompt}\n\n### Response:\n"
+        return {"error": "Model not initialized"}
     
-    response = llm(
-        prompt,
-        max_tokens=512,
-        stop=["###", "</s>"],
+    output = llm(
+        f"### Instruction:\n{query.prompt}\n\n### Response:\n",
+        max_tokens=256,
+        stop=["###"],
         echo=False
     )
-    
-    return {"response": response["choices"][0]["text"].strip()}
+    return {"response": output["choices"][0]["text"]}
 
 @app.get("/")
-def home():
-    return {"message": "Aevis Medical SLM API is Online", "model": REPO_ID}
+def health():
+    return {"status": "Aevis API is running"}