Update tooling.py

tooling.py

@@ -1,92 +1,92 @@
-from smolagents import DuckDuckGoSearchTool, HfApiModel, load_tool, CodeAgent, PythonInterpreterTool, VisitWebpageTool, \
-    Tool
-import hashlib
-import json
-from transformers import AutoTokenizer, AutoModelForCausalLM
-import os
-
-
-class ModelMathTool(Tool):
-    name = "math_model"
-    description = "Answers advanced math questions using a pretrained math model."
-
-    inputs = {
-        "problem": {
-            "type": "string",
-            "description": "Math problem to solve.",
-        }
-    }
-
-    output_type = "string"
-
-    def __init__(self, model_id="Qwen/Qwen2.5-Math-7B"):
-        print(f"Loading math model: {model_id}")
-        self.tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
-        self.model = HfApiModel(model_id=model_id, max_tokens=512)
-
-    def forward(self, problem: str) -> str:
-        print(f"[MathModelTool] Question: {problem}")
-        response = self.model.__call__(problem)
-        return response
-
-
-# (Keep Constants as is)
-# --- Constants ---
-DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
-
-web_search = DuckDuckGoSearchTool()
-python_interpreter = PythonInterpreterTool()
-visit_webpage_tool = VisitWebpageTool()
-model_math_tool = ModelMathTool()
-
-# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
-# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud'
-
-model = HfApiModel(model_id="HuggingFaceH4/zephyr-7b-beta", max_tokens=512, token=tok)
-
-
-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)
-
-
-class BasicAgent:
-    def __init__(self):
-        print("BasicAgent initialized.")
-        self.agent = CodeAgent(
-            model=model,
-            tools=[model_math_tool],
-            max_steps=1,
-            verbosity_level=0,
-            grammar=None,
-            planning_interval=3,
-
-        )
-
-    def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
-        answer = self.agent.run(question)
-        return answer
-
-
-
-agent = BasicAgent()
-
-response = agent.__call__(question="How much is 2*2?")
+from smolagents import DuckDuckGoSearchTool, HfApiModel, load_tool, CodeAgent, PythonInterpreterTool, VisitWebpageTool, \
+    Tool
+import hashlib
+import json
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import os
+
+
+class ModelMathTool(Tool):
+    name = "math_model"
+    description = "Answers advanced math questions using a pretrained math model."
+
+    inputs = {
+        "problem": {
+            "type": "string",
+            "description": "Math problem to solve.",
+        }
+    }
+
+    output_type = "string"
+
+    def __init__(self, model_id="Qwen/Qwen2.5-Math-7B"):
+        print(f"Loading math model: {model_id}")
+        self.tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+        self.model = HfApiModel(model_id=model_id, max_tokens=512)
+
+    def forward(self, problem: str) -> str:
+        print(f"[MathModelTool] Question: {problem}")
+        response = self.model.__call__(problem)
+        return response
+
+
+# (Keep Constants as is)
+# --- Constants ---
+DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+
+web_search = DuckDuckGoSearchTool()
+python_interpreter = PythonInterpreterTool()
+visit_webpage_tool = VisitWebpageTool()
+model_math_tool = ModelMathTool()
+
+# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
+# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud'
+
+model = HfApiModel(model_id="HuggingFaceH4/zephyr-7b-beta", max_tokens=512)
+
+
+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)
+
+
+class BasicAgent:
+    def __init__(self):
+        print("BasicAgent initialized.")
+        self.agent = CodeAgent(
+            model=model,
+            tools=[model_math_tool],
+            max_steps=1,
+            verbosity_level=0,
+            grammar=None,
+            planning_interval=3,
+
+        )
+
+    def __call__(self, question: str) -> str:
+        print(f"Agent received question (first 50 chars): {question[:50]}...")
+        answer = self.agent.run(question)
+        return answer
+
+
+
+agent = BasicAgent()
+
+response = agent.__call__(question="How much is 2*2?")
 print(response)