kpalatel commited on
Commit
34e66dc
·
verified ·
1 Parent(s): 22221ae

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +5 -197
app.py CHANGED
@@ -1,202 +1,10 @@
1
- import os
2
- import gradio as gr
3
- import requests
4
- import inspect
5
- import pandas as pd
6
- from smolagents import CodeAgent, HfApiEngine, DuckDuckGoSearchTool
7
 
8
- # --- Constants ---
9
- DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
10
-
11
- # --- Smart Agent Definition ---
12
  class BasicAgent:
13
  def __init__(self):
14
- print("Initializing powerful CodeAgent for evaluation...")
15
- # Using a highly capable reasoning model via serverless API
16
- # It automatically looks for the HF_TOKEN secret in your environment variables
17
- self.engine = HfApiEngine(model_id="meta-llama/Llama-3.3-70B-Instruct")
18
-
19
- # Initializing the CodeAgent with web search capabilities
20
- # A CodeAgent can write and execute Python code blocks to solve complex reasoning/math
21
- self.agent = CodeAgent(
22
- tools=[DuckDuckGoSearchTool()],
23
- engine=self.engine,
24
- max_steps=5 # Cap steps to prevent timeouts and infinite loops
25
- )
26
 
27
  def __call__(self, question: str) -> str:
28
- print(f"\n--- New Evaluation Task ---")
29
- print(f"Agent received question (first 50 chars): {question[:50]}...")
30
- try:
31
- # Run the agent through its thinking steps to find the correct answer
32
- result = self.agent.run(question)
33
- print(f"Agent successfully resolved answer: {str(result)[:100]}...")
34
- return str(result)
35
- except Exception as e:
36
- print(f"Error during agent execution: {e}")
37
- return "Error calculating answer."
38
-
39
- def run_and_submit_all(profile: gr.OAuthProfile | None):
40
- """
41
- Fetches all questions, runs the BasicAgent on them, submits all answers,
42
- and displays the results.
43
- """
44
- # --- Determine HF Space Runtime URL and Repo URL ---
45
- space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
46
- if profile:
47
- username = f"{profile.username}"
48
- print(f"User logged in: {username}")
49
- else:
50
- print("User not logged in.")
51
- return "Please Login to Hugging Face with the button.", None
52
-
53
- api_url = DEFAULT_API_URL
54
- questions_url = f"{api_url}/questions"
55
- submit_url = f"{api_url}/submit"
56
-
57
- # 1. Instantiate Agent
58
- try:
59
- agent = BasicAgent()
60
- except Exception as e:
61
- print(f"Error instantiating agent: {e}")
62
- return f"Error initializing agent: {e}", None
63
-
64
- # Link pointing toward your public codebase
65
- agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
66
- print(agent_code)
67
-
68
- # 2. Fetch Questions
69
- print(f"Fetching questions from: {questions_url}")
70
- try:
71
- response = requests.get(questions_url, timeout=15)
72
- response.raise_for_status()
73
- questions_data = response.json()
74
- if not questions_data:
75
- print("Fetched questions list is empty.")
76
- return "Fetched questions list is empty or invalid format.", None
77
- print(f"Fetched {len(questions_data)} questions.")
78
- except requests.exceptions.RequestException as e:
79
- print(f"Error fetching questions: {e}")
80
- return f"Error fetching questions: {e}", None
81
- except requests.exceptions.JSONDecodeError as e:
82
- print(f"Error decoding JSON response from questions endpoint: {e}")
83
- print(f"Response text: {response.text[:500]}")
84
- return f"Error decoding server response for questions: {e}", None
85
- except Exception as e:
86
- print(f"An unexpected error occurred fetching questions: {e}")
87
- return f"An unexpected error occurred fetching questions: {e}", None
88
-
89
- # 3. Run your Agent
90
- results_log = []
91
- answers_payload = []
92
- print(f"Running agent on {len(questions_data)} questions...")
93
- for item in questions_data:
94
- task_id = item.get("task_id")
95
- question_text = item.get("question")
96
- if not task_id or question_text is None:
97
- print(f"Skipping item with missing task_id or question: {item}")
98
- continue
99
- try:
100
- submitted_answer = agent(question_text)
101
- answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
102
- results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
103
- except Exception as e:
104
- print(f"Error running agent on task {task_id}: {e}")
105
- results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
106
-
107
- if not answers_payload:
108
- print("Agent did not produce any answers to submit.")
109
- return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
110
-
111
- # 4. Prepare Submission
112
- submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
113
- status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
114
- print(status_update)
115
-
116
- # 5. Submit
117
- print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
118
- try:
119
- response = requests.post(submit_url, json=submission_data, timeout=60)
120
- response.raise_for_status()
121
- result_data = response.json()
122
- final_status = (
123
- f"Submission Successful!\n"
124
- f"User: {result_data.get('username')}\n"
125
- f"Overall Score: {result_data.get('score', 'N/A')}% "
126
- f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
127
- f"Message: {result_data.get('message', 'No message received.')}"
128
- )
129
- print("Submission successful.")
130
- results_df = pd.DataFrame(results_log)
131
- return final_status, results_df
132
- except requests.exceptions.HTTPError as e:
133
- error_detail = f"Server responded with status {e.response.status_code}."
134
- try:
135
- error_json = e.response.json()
136
- error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
137
- except requests.exceptions.JSONDecodeError:
138
- error_detail += f" Response: {e.response.text[:500]}"
139
- status_message = f"Submission Failed: {error_detail}"
140
- print(status_message)
141
- results_df = pd.DataFrame(results_log)
142
- return status_message, results_df
143
- except requests.exceptions.Timeout:
144
- status_message = "Submission Failed: The request timed out."
145
- print(status_message)
146
- results_df = pd.DataFrame(results_log)
147
- return status_message, results_df
148
- except requests.exceptions.RequestException as e:
149
- status_message = f"Submission Failed: Network error - {e}"
150
- print(status_message)
151
- results_df = pd.DataFrame(results_log)
152
- return status_message, results_df
153
- except Exception as e:
154
- status_message = f"An unexpected error occurred during submission: {e}"
155
- print(status_message)
156
- results_df = pd.DataFrame(results_log)
157
- return status_message, results_df
158
-
159
- # --- Build Gradio Interface using Blocks ---
160
- with gr.Blocks() as demo:
161
- gr.Markdown("# Smart Agent Evaluation Runner")
162
- gr.Markdown(
163
- """
164
- **Instructions:**
165
- 1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
166
- 2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
167
- 3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
168
- ---
169
- **Disclaimers:**
170
- Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
171
- This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
172
- """
173
- )
174
- gr.LoginButton()
175
- run_button = gr.Button("Run Evaluation & Submit All Answers")
176
- status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
177
- results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
178
-
179
- run_button.click(
180
- fn=run_and_submit_all,
181
- outputs=[status_output, results_table]
182
- )
183
-
184
- if __name__ == "__main__":
185
- print("\n" + "-"*30 + " App Starting " + "-"*30)
186
- space_host_startup = os.getenv("SPACE_HOST")
187
- space_id_startup = os.getenv("SPACE_ID")
188
- if space_host_startup:
189
- print(f"✅ SPACE_HOST found: {space_host_startup}")
190
- print(f" Runtime URL should be: https://{space_host_startup}.hf.space")
191
- else:
192
- print("ℹ️ SPACE_HOST environment variable not found (running locally?).")
193
- if space_id_startup:
194
- print(f"✅ SPACE_ID found: {space_id_startup}")
195
- print(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}")
196
- print(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
197
- else:
198
- print("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")
199
- print("-"*(60 + len(" App Starting ")) + "\n")
200
-
201
- print("Launching Gradio Interface for Smart Agent Evaluation...")
202
- demo.launch(debug=True, share=False)
 
1
+ # Replace your previous BasicAgent definition inside app.py with this import strategy
2
+ from agent import CourseEvaluationAgent
 
 
 
 
3
 
 
 
 
 
4
  class BasicAgent:
5
  def __init__(self):
6
+ # Initializes our new externalized architecture agent
7
+ self.eval_agent = CourseEvaluationAgent()
 
 
 
 
 
 
 
 
 
 
8
 
9
  def __call__(self, question: str) -> str:
10
+ return self.eval_agent.run_task(question)