wlchee commited on
Commit
7e4e51a
·
verified ·
1 Parent(s): fd7ec87

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +169 -120
app.py CHANGED
@@ -1,145 +1,194 @@
1
  import os
2
  import gradio as gr
3
  import requests
 
4
  import pandas as pd
5
- from datetime import datetime
6
- from smolagents import Tool, ToolCallingAgent
7
- from smolagents.models import InferenceClientModel
8
 
9
- # Constants
 
10
  DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
11
 
12
- # --- Custom Tools ---
13
- class CalculatorTool(Tool):
14
- name = "calculator"
15
- description = "Performs mathematical calculations"
16
- input_schema = {
17
- "expression": {
18
- "type": "string",
19
- "description": "Math expression to evaluate (e.g. '2+2')"
20
- }
21
- }
22
- output_schema = {
23
- "result": {
24
- "type": "string",
25
- "description": "The result of the expression"
26
- }
27
- }
28
-
29
- def use(self, expression: str) -> dict:
30
- try:
31
- return {"result": str(eval(expression))}
32
- except Exception as e:
33
- return {"result": f"Error: {e}"}
34
-
35
-
36
- class TimeTool(Tool):
37
- name = "current_time"
38
- description = "Gets current UTC time"
39
- input_schema = {}
40
- output_schema = {
41
- "time": {
42
- "type": "string",
43
- "description": "Current time in UTC"
44
- }
45
- }
46
-
47
- def use(self) -> dict:
48
- return {"time": datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S UTC")}
49
-
50
- # --- Agent ---
51
- class LocalAgent:
52
  def __init__(self):
53
- self.tools = [CalculatorTool(), TimeTool()]
54
- self.agent = ToolCallingAgent(
55
- tools=self.tools,
56
- model=InferenceClientModel(
57
- model_id="HuggingFaceH4/zephyr-7b-beta",
58
- api_base="https://api-inference.huggingface.co/models"
59
- )
60
- )
61
-
62
  def __call__(self, question: str) -> str:
63
- question_lower = question.lower()
64
- if any(op in question_lower for op in ["calculate", "+", "-", "*", "/", "what is"]):
65
- return CalculatorTool().use(question.replace("?", ""))["result"]
66
- if "time" in question_lower:
67
- return TimeTool().use()["time"]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
68
  try:
69
- return str(self.agent.run(question))
 
 
70
  except Exception as e:
71
- return f"Error: {e}"
 
72
 
73
- # --- Evaluation Function ---
74
- def run_and_submit_all(profile: gr.OAuthProfile | None):
75
- if not profile:
76
- return "Please login first.", None
77
 
78
- space_id = os.getenv("SPACE_ID", "local-test")
79
- api_url = os.getenv("API_URL", DEFAULT_API_URL)
 
 
80
 
 
 
81
  try:
82
- agent = LocalAgent()
83
- questions = requests.get(f"{api_url}/questions", timeout=15).json()
84
-
85
- answers = []
86
- logs = []
87
-
88
- for q in questions:
89
- try:
90
- ans = agent(q["question"])
91
- answers.append({
92
- "task_id": q["task_id"],
93
- "submitted_answer": ans
94
- })
95
- logs.append({
96
- "Task ID": q["task_id"],
97
- "Question": q["question"],
98
- "Answer": ans
99
- })
100
- except Exception as e:
101
- logs.append({
102
- "Task ID": q["task_id"],
103
- "Question": q["question"],
104
- "Answer": f"Error: {e}"
105
- })
106
-
107
- submission = {
108
- "username": profile.username,
109
- "agent_code": f"https://huggingface.co/spaces/{space_id}",
110
- "answers": answers
111
- }
112
-
113
- result = requests.post(f"{api_url}/submit", json=submission, timeout=60).json()
114
-
115
- return (
116
- f"✅ Score: {result.get('score', 'N/A')}%\n"
117
- f"Correct: {result.get('correct_count', '?')}/{result.get('total_attempted', '?')}",
118
- pd.DataFrame(logs)
119
  )
120
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
121
  except Exception as e:
122
- return f"Evaluation failed: {e}", pd.DataFrame([])
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
123
 
124
- # --- Gradio UI ---
125
- with gr.Blocks(title="Agent Evaluation") as app:
126
- gr.Markdown("## 🤖 Agent Evaluation with smolagents")
127
- gr.Markdown("Login, then click 'Run Evaluation' to test your agent.")
128
 
129
- gr.LoginButton() # Login button visible
130
- profile = gr.OAuthProfile() # Profile input (not visible)
131
 
132
- run_btn = gr.Button("🚀 Run Evaluation")
133
- output = gr.Textbox(label="Evaluation Result")
134
- results_table = gr.DataFrame(label="Answer Log")
135
 
136
- run_btn.click(
137
  fn=run_and_submit_all,
138
- inputs=[profile], # ✅ Must include this
139
- outputs=[output, results_table]
140
  )
141
 
142
  if __name__ == "__main__":
143
- app.launch()
144
-
145
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  import os
2
  import gradio as gr
3
  import requests
4
+ import inspect
5
  import pandas as pd
 
 
 
6
 
7
+ # (Keep Constants as is)
8
+ # --- Constants ---
9
  DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
10
 
11
+ # --- Basic Agent Definition ---
12
+ # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
13
+ class BasicAgent:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
  def __init__(self):
15
+ print("BasicAgent initialized.")
 
 
 
 
 
 
 
 
16
  def __call__(self, question: str) -> str:
17
+ print(f"Agent received question (first 50 chars): {question[:50]}...")
18
+ fixed_answer = "This is a default answer."
19
+ print(f"Agent returning fixed answer: {fixed_answer}")
20
+ return fixed_answer
21
+
22
+ def run_and_submit_all( profile: gr.OAuthProfile | None):
23
+ """
24
+ Fetches all questions, runs the BasicAgent on them, submits all answers,
25
+ and displays the results.
26
+ """
27
+ # --- Determine HF Space Runtime URL and Repo URL ---
28
+ space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
29
+
30
+ if profile:
31
+ username= f"{profile.username}"
32
+ print(f"User logged in: {username}")
33
+ else:
34
+ print("User not logged in.")
35
+ return "Please Login to Hugging Face with the button.", None
36
+
37
+ api_url = DEFAULT_API_URL
38
+ questions_url = f"{api_url}/questions"
39
+ submit_url = f"{api_url}/submit"
40
+
41
+ # 1. Instantiate Agent ( modify this part to create your agent)
42
+ try:
43
+ agent = BasicAgent()
44
+ except Exception as e:
45
+ print(f"Error instantiating agent: {e}")
46
+ return f"Error initializing agent: {e}", None
47
+ # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
48
+ agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
49
+ print(agent_code)
50
+
51
+ # 2. Fetch Questions
52
+ print(f"Fetching questions from: {questions_url}")
53
+ try:
54
+ response = requests.get(questions_url, timeout=15)
55
+ response.raise_for_status()
56
+ questions_data = response.json()
57
+ if not questions_data:
58
+ print("Fetched questions list is empty.")
59
+ return "Fetched questions list is empty or invalid format.", None
60
+ print(f"Fetched {len(questions_data)} questions.")
61
+ except requests.exceptions.RequestException as e:
62
+ print(f"Error fetching questions: {e}")
63
+ return f"Error fetching questions: {e}", None
64
+ except requests.exceptions.JSONDecodeError as e:
65
+ print(f"Error decoding JSON response from questions endpoint: {e}")
66
+ print(f"Response text: {response.text[:500]}")
67
+ return f"Error decoding server response for questions: {e}", None
68
+ except Exception as e:
69
+ print(f"An unexpected error occurred fetching questions: {e}")
70
+ return f"An unexpected error occurred fetching questions: {e}", None
71
+
72
+ # 3. Run your Agent
73
+ results_log = []
74
+ answers_payload = []
75
+ print(f"Running agent on {len(questions_data)} questions...")
76
+ for item in questions_data:
77
+ task_id = item.get("task_id")
78
+ question_text = item.get("question")
79
+ if not task_id or question_text is None:
80
+ print(f"Skipping item with missing task_id or question: {item}")
81
+ continue
82
  try:
83
+ submitted_answer = agent(question_text)
84
+ answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
85
+ results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
86
  except Exception as e:
87
+ print(f"Error running agent on task {task_id}: {e}")
88
+ results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
89
 
90
+ if not answers_payload:
91
+ print("Agent did not produce any answers to submit.")
92
+ return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
 
93
 
94
+ # 4. Prepare Submission
95
+ submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
96
+ status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
97
+ print(status_update)
98
 
99
+ # 5. Submit
100
+ print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
101
  try:
102
+ response = requests.post(submit_url, json=submission_data, timeout=60)
103
+ response.raise_for_status()
104
+ result_data = response.json()
105
+ final_status = (
106
+ f"Submission Successful!\n"
107
+ f"User: {result_data.get('username')}\n"
108
+ f"Overall Score: {result_data.get('score', 'N/A')}% "
109
+ f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
110
+ f"Message: {result_data.get('message', 'No message received.')}"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  )
112
+ print("Submission successful.")
113
+ results_df = pd.DataFrame(results_log)
114
+ return final_status, results_df
115
+ except requests.exceptions.HTTPError as e:
116
+ error_detail = f"Server responded with status {e.response.status_code}."
117
+ try:
118
+ error_json = e.response.json()
119
+ error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
120
+ except requests.exceptions.JSONDecodeError:
121
+ error_detail += f" Response: {e.response.text[:500]}"
122
+ status_message = f"Submission Failed: {error_detail}"
123
+ print(status_message)
124
+ results_df = pd.DataFrame(results_log)
125
+ return status_message, results_df
126
+ except requests.exceptions.Timeout:
127
+ status_message = "Submission Failed: The request timed out."
128
+ print(status_message)
129
+ results_df = pd.DataFrame(results_log)
130
+ return status_message, results_df
131
+ except requests.exceptions.RequestException as e:
132
+ status_message = f"Submission Failed: Network error - {e}"
133
+ print(status_message)
134
+ results_df = pd.DataFrame(results_log)
135
+ return status_message, results_df
136
  except Exception as e:
137
+ status_message = f"An unexpected error occurred during submission: {e}"
138
+ print(status_message)
139
+ results_df = pd.DataFrame(results_log)
140
+ return status_message, results_df
141
+
142
+
143
+ # --- Build Gradio Interface using Blocks ---
144
+ with gr.Blocks() as demo:
145
+ gr.Markdown("# Basic Agent Evaluation Runner")
146
+ gr.Markdown(
147
+ """
148
+ **Instructions:**
149
+ 1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
150
+ 2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
151
+ 3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
152
+ ---
153
+ **Disclaimers:**
154
+ 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).
155
+ 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.
156
+ """
157
+ )
158
 
159
+ gr.LoginButton()
 
 
 
160
 
161
+ run_button = gr.Button("Run Evaluation & Submit All Answers")
 
162
 
163
+ status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
164
+ # Removed max_rows=10 from DataFrame constructor
165
+ results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
166
 
167
+ run_button.click(
168
  fn=run_and_submit_all,
169
+ outputs=[status_output, results_table]
 
170
  )
171
 
172
  if __name__ == "__main__":
173
+ print("\n" + "-"*30 + " App Starting " + "-"*30)
174
+ # Check for SPACE_HOST and SPACE_ID at startup for information
175
+ space_host_startup = os.getenv("SPACE_HOST")
176
+ space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
177
+
178
+ if space_host_startup:
179
+ print(f"✅ SPACE_HOST found: {space_host_startup}")
180
+ print(f" Runtime URL should be: https://{space_host_startup}.hf.space")
181
+ else:
182
+ print("ℹ️ SPACE_HOST environment variable not found (running locally?).")
183
+
184
+ if space_id_startup: # Print repo URLs if SPACE_ID is found
185
+ print(f"✅ SPACE_ID found: {space_id_startup}")
186
+ print(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}")
187
+ print(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
188
+ else:
189
+ print("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")
190
+
191
+ print("-"*(60 + len(" App Starting ")) + "\n")
192
+
193
+ print("Launching Gradio Interface for Basic Agent Evaluation...")
194
+ demo.launch(debug=True, share=False)