Update app.py

app.py

@@ -19,8 +19,10 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 import os
 
-os.makedirs("cache", exist_ok=True)
-print("Made dir")
+
+cache= {}
+
+
 web_search = DuckDuckGoSearchTool()
 python_interpreter = PythonInterpreterTool()
 visit_webpage_tool = VisitWebpageTool()
@@ -33,24 +35,14 @@ visit_webpage_tool = VisitWebpageTool()
 
 
 
-def get_cache_key(question: str) -> str:
-    return hashlib.sha256(question.encode()).hexdigest()
-
-def load_cached_answer(question: str) -> str | None:
-    key = get_cache_key(question)
-    path = f"cache/{key}.json"
-    if os.path.exists(path):
-        with open(path, "r") as f:
-            data = json.load(f)
-            return data.get("answer")
-    return None
-
-def cache_answer(question: str, answer: str):
-    key = get_cache_key(question)
-    path = f"cache/{key}.json"
-    with open(path, "w") as f:
-        json.dump({"question": question, "answer": answer}, f)
+def load_cached_answer(question_id: str) -> str:
+    if question_id in cache.keys():
+        return cache[question_id]
+    else:
+        return None
 
+def cache_answer(question_id: str, answer: str):
+    cache[question_id] = answer
 
 
 # --- Model Setup ---
@@ -84,16 +76,16 @@ def load_model(model_name):
 
 # Load the model and tokenizer locally
 #model, tokenizer = load_model()
-#model = HfApiModel()#model_id=MODEL_NAME, max_tokens=512)
+model = HfApiModel()#model_id=MODEL_NAME, max_tokens=512)
 
 from smolagents import TransformersModel
 
 
 
 model_id ="meta-llama/Llama-3.2-3B-Instruct"#"microsoft/phi-2"# not working out of the box"google/gemma-2-2b-it" #toobig"Qwen/Qwen1.5-7B-Chat"#working but stupid: "meta-llama/Llama-3.2-3B-Instruct"
-model = TransformersModel(
-    model_id=model_id,
-    max_new_tokens=256)
+#model = TransformersModel(
+#    model_id=model_id,
+#    max_new_tokens=256)
 
 #model = HfApiModel()
 
@@ -183,13 +175,13 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
             print(f"Skipping item with missing task_id or question: {item}")
             continue
         try:
-            cached = load_cached_answer(question_text)
+            cached = load_cached_answer(task_id)
             if cached:
                 submitted_answer = cached
                 print(f"Loaded cached answer for task {task_id}")
             else:
                 submitted_answer = agent(question_text)
-                cache_answer(question_text, submitted_answer)
+                cache_answer(task_id, submitted_answer)
                 print(f"Generated and cached answer for task {task_id}")
     
             answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})