Create app.py

app.py

@@ -0,0 +1,32 @@
+from fastapi import FastAPI
+from pydantic import BaseModel
+from llama_cpp import Llama
+
+app = FastAPI()
+
+# Download and initialize the model when the server starts
+llm = Llama.from_pretrained(
+    repo_id="Qwen/Qwen2.5-Coder-1.5B-Instruct-GGUF",
+    filename="*q4_k_m.gguf", # 4-bit quantization for speed and low memory
+    n_ctx=2048 # Context window size
+)
+
+class EvalRequest(BaseModel):
+    task_description: str
+    python_code: str
+
+@app.post("/evaluate")
+async def evaluate_code(request: EvalRequest):
+    prompt = f"Task Description:\n{request.task_description}\n\nSubmitted Code:\n{request.python_code}\n\nEvaluate the code against the task. Assign a final score out of 10. Keep your feedback concise and helpful."
+    
+    response = llm.create_chat_completion(
+        messages=[
+            # Framing the model specifically for grading student submissions
+            {"role": "system", "content": "You are an expert Python instructor. You evaluate student code submissions accurately, checking for logical correctness and task completion."},
+            {"role": "user", "content": prompt}
+        ],
+        max_tokens=250, # Limit response length to keep API fast
+        temperature=0.2 # Low temperature for consistent scoring
+    )
+    
+    return {"evaluation": response['choices'][0]['message']['content']}