Update app.py
Browse files
app.py
CHANGED
|
@@ -1,37 +1,45 @@
|
|
| 1 |
-
""" Basic Agent Evaluation Runner"""
|
| 2 |
import os
|
| 3 |
-
import inspect
|
| 4 |
import gradio as gr
|
| 5 |
import requests
|
|
|
|
| 6 |
import pandas as pd
|
| 7 |
-
from langchain_core.messages import HumanMessage
|
| 8 |
from agent import build_graph
|
| 9 |
-
|
| 10 |
-
|
|
|
|
| 11 |
|
| 12 |
# (Keep Constants as is)
|
| 13 |
# --- Constants ---
|
| 14 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
| 15 |
|
| 16 |
# --- Basic Agent Definition ---
|
| 17 |
-
# ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
|
| 18 |
-
|
| 19 |
-
|
| 20 |
class BasicAgent:
|
| 21 |
-
"""A langgraph agent."""
|
| 22 |
def __init__(self):
|
| 23 |
print("BasicAgent initialized.")
|
| 24 |
-
self.
|
| 25 |
-
|
| 26 |
def __call__(self, question: str) -> str:
|
| 27 |
print(f"Agent received question (first 50 chars): {question[:50]}...")
|
| 28 |
-
# Wrap the question in a HumanMessage from langchain_core
|
| 29 |
messages = [HumanMessage(content=question)]
|
| 30 |
-
messages = self.
|
| 31 |
answer = messages['messages'][-1].content
|
| 32 |
-
return answer[14:]
|
| 33 |
-
|
| 34 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
def run_and_submit_all( profile: gr.OAuthProfile | None):
|
| 36 |
"""
|
| 37 |
Fetches all questions, runs the BasicAgent on them, submits all answers,
|
|
@@ -45,23 +53,24 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 45 |
print(f"User logged in: {username}")
|
| 46 |
else:
|
| 47 |
print("User not logged in.")
|
| 48 |
-
return "Please
|
| 49 |
|
| 50 |
api_url = DEFAULT_API_URL
|
| 51 |
questions_url = f"{api_url}/questions"
|
| 52 |
submit_url = f"{api_url}/submit"
|
| 53 |
|
| 54 |
-
# 1. Instantiate Agent (
|
| 55 |
try:
|
| 56 |
agent = BasicAgent()
|
| 57 |
except Exception as e:
|
| 58 |
print(f"Error instantiating agent: {e}")
|
| 59 |
return f"Error initializing agent: {e}", None
|
| 60 |
-
|
|
|
|
| 61 |
agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
|
| 62 |
print(agent_code)
|
| 63 |
|
| 64 |
-
# 2. Fetch
|
| 65 |
print(f"Fetching questions from: {questions_url}")
|
| 66 |
try:
|
| 67 |
response = requests.get(questions_url, timeout=15)
|
|
@@ -82,9 +91,12 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 82 |
print(f"An unexpected error occurred fetching questions: {e}")
|
| 83 |
return f"An unexpected error occurred fetching questions: {e}", None
|
| 84 |
|
| 85 |
-
# 3. Run your
|
| 86 |
results_log = []
|
| 87 |
answers_payload = []
|
|
|
|
|
|
|
|
|
|
| 88 |
print(f"Running agent on {len(questions_data)} questions...")
|
| 89 |
for item in questions_data:
|
| 90 |
task_id = item.get("task_id")
|
|
@@ -93,23 +105,27 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 93 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 94 |
continue
|
| 95 |
try:
|
| 96 |
-
submitted_answer = agent(question_text)
|
|
|
|
|
|
|
| 97 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
| 98 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
|
|
|
|
| 99 |
except Exception as e:
|
| 100 |
-
|
| 101 |
-
|
|
|
|
| 102 |
|
| 103 |
if not answers_payload:
|
| 104 |
print("Agent did not produce any answers to submit.")
|
| 105 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
| 106 |
|
| 107 |
-
# 4. Prepare
|
| 108 |
submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
|
| 109 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 110 |
print(status_update)
|
| 111 |
|
| 112 |
-
# 5. Submit
|
| 113 |
print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
|
| 114 |
try:
|
| 115 |
response = requests.post(submit_url, json=submission_data, timeout=60)
|
|
@@ -147,13 +163,13 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 147 |
results_df = pd.DataFrame(results_log)
|
| 148 |
return status_message, results_df
|
| 149 |
except Exception as e:
|
| 150 |
-
status_message = f"
|
| 151 |
print(status_message)
|
| 152 |
results_df = pd.DataFrame(results_log)
|
| 153 |
return status_message, results_df
|
| 154 |
|
| 155 |
|
| 156 |
-
# --- Build Gradio
|
| 157 |
with gr.Blocks() as demo:
|
| 158 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 159 |
gr.Markdown(
|
|
@@ -203,5 +219,5 @@ if __name__ == "__main__":
|
|
| 203 |
|
| 204 |
print("-"*(60 + len(" App Starting ")) + "\n")
|
| 205 |
|
| 206 |
-
print("Launching Gradio
|
| 207 |
demo.launch(debug=True, share=False)
|
|
|
|
|
|
|
| 1 |
import os
|
|
|
|
| 2 |
import gradio as gr
|
| 3 |
import requests
|
| 4 |
+
import inspect
|
| 5 |
import pandas as pd
|
|
|
|
| 6 |
from agent import build_graph
|
| 7 |
+
from langchain_core.messages import HumanMessage
|
| 8 |
+
import time
|
| 9 |
+
import csv
|
| 10 |
|
| 11 |
# (Keep Constants as is)
|
| 12 |
# --- Constants ---
|
| 13 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
| 14 |
|
| 15 |
# --- Basic Agent Definition ---
|
|
|
|
|
|
|
|
|
|
| 16 |
class BasicAgent:
|
|
|
|
| 17 |
def __init__(self):
|
| 18 |
print("BasicAgent initialized.")
|
| 19 |
+
self.agent = build_graph()
|
| 20 |
+
|
| 21 |
def __call__(self, question: str) -> str:
|
| 22 |
print(f"Agent received question (first 50 chars): {question[:50]}...")
|
|
|
|
| 23 |
messages = [HumanMessage(content=question)]
|
| 24 |
+
messages = self.agent.invoke({"messages": messages})
|
| 25 |
answer = messages['messages'][-1].content
|
|
|
|
|
|
|
| 26 |
|
| 27 |
+
index = answer.find("FINAL ANSWER: ")
|
| 28 |
+
if index == -1:
|
| 29 |
+
return answer
|
| 30 |
+
return answer[index+14:]
|
| 31 |
+
|
| 32 |
+
# --- Upload answers solved locally ---
|
| 33 |
+
def csv_to_dict(file_path):
|
| 34 |
+
result = {}
|
| 35 |
+
with open(file_path, 'r') as file:
|
| 36 |
+
csv_reader = csv.reader(file)
|
| 37 |
+
header = next(csv_reader) # Skip header row
|
| 38 |
+
for row in csv_reader:
|
| 39 |
+
result[row[0]] = row[1]
|
| 40 |
+
return result
|
| 41 |
+
|
| 42 |
+
|
| 43 |
def run_and_submit_all( profile: gr.OAuthProfile | None):
|
| 44 |
"""
|
| 45 |
Fetches all questions, runs the BasicAgent on them, submits all answers,
|
|
|
|
| 53 |
print(f"User logged in: {username}")
|
| 54 |
else:
|
| 55 |
print("User not logged in.")
|
| 56 |
+
return "Please log in to Hugging Face with the button.", None
|
| 57 |
|
| 58 |
api_url = DEFAULT_API_URL
|
| 59 |
questions_url = f"{api_url}/questions"
|
| 60 |
submit_url = f"{api_url}/submit"
|
| 61 |
|
| 62 |
+
# 1. Instantiate Agent (modify this part to create your agent)
|
| 63 |
try:
|
| 64 |
agent = BasicAgent()
|
| 65 |
except Exception as e:
|
| 66 |
print(f"Error instantiating agent: {e}")
|
| 67 |
return f"Error initializing agent: {e}", None
|
| 68 |
+
|
| 69 |
+
# 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)
|
| 70 |
agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
|
| 71 |
print(agent_code)
|
| 72 |
|
| 73 |
+
# 2. Fetch questions
|
| 74 |
print(f"Fetching questions from: {questions_url}")
|
| 75 |
try:
|
| 76 |
response = requests.get(questions_url, timeout=15)
|
|
|
|
| 91 |
print(f"An unexpected error occurred fetching questions: {e}")
|
| 92 |
return f"An unexpected error occurred fetching questions: {e}", None
|
| 93 |
|
| 94 |
+
# 3. Run your agent
|
| 95 |
results_log = []
|
| 96 |
answers_payload = []
|
| 97 |
+
|
| 98 |
+
answers = csv_to_dict("answers.csv")
|
| 99 |
+
|
| 100 |
print(f"Running agent on {len(questions_data)} questions...")
|
| 101 |
for item in questions_data:
|
| 102 |
task_id = item.get("task_id")
|
|
|
|
| 105 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 106 |
continue
|
| 107 |
try:
|
| 108 |
+
#submitted_answer = agent(question_text)
|
| 109 |
+
submitted_answer = answers[task_id]
|
| 110 |
+
|
| 111 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
| 112 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
|
| 113 |
+
time.sleep(10)
|
| 114 |
except Exception as e:
|
| 115 |
+
print(f"Error running agent on task {task_id}: {e}")
|
| 116 |
+
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
|
| 117 |
+
time.sleep(10)
|
| 118 |
|
| 119 |
if not answers_payload:
|
| 120 |
print("Agent did not produce any answers to submit.")
|
| 121 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
| 122 |
|
| 123 |
+
# 4. Prepare submission
|
| 124 |
submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
|
| 125 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 126 |
print(status_update)
|
| 127 |
|
| 128 |
+
# 5. Submit answers
|
| 129 |
print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
|
| 130 |
try:
|
| 131 |
response = requests.post(submit_url, json=submission_data, timeout=60)
|
|
|
|
| 163 |
results_df = pd.DataFrame(results_log)
|
| 164 |
return status_message, results_df
|
| 165 |
except Exception as e:
|
| 166 |
+
status_message = f"Unexpected error occurred during submission: {e}"
|
| 167 |
print(status_message)
|
| 168 |
results_df = pd.DataFrame(results_log)
|
| 169 |
return status_message, results_df
|
| 170 |
|
| 171 |
|
| 172 |
+
# --- Build Gradio interface using Blocks ---
|
| 173 |
with gr.Blocks() as demo:
|
| 174 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 175 |
gr.Markdown(
|
|
|
|
| 219 |
|
| 220 |
print("-"*(60 + len(" App Starting ")) + "\n")
|
| 221 |
|
| 222 |
+
print("Launching Gradio interface for Basic Agent evaluation...")
|
| 223 |
demo.launch(debug=True, share=False)
|