Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,49 +1,24 @@
|
|
| 1 |
# 17/08/2025 -AO
|
|
|
|
| 2 |
import os
|
| 3 |
import gradio as gr
|
| 4 |
import requests
|
| 5 |
import inspect
|
| 6 |
import pandas as pd
|
| 7 |
-
import
|
| 8 |
-
|
| 9 |
-
from transformers.agents import HfAgent
|
| 10 |
-
from transformers.tools import DuckDuckGoSearchTool
|
| 11 |
-
from huggingface_hub import InferenceClient
|
| 12 |
|
| 13 |
# --- Constants ---
|
| 14 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
| 15 |
|
| 16 |
-
# --- Advanced Agent Definition ---
|
| 17 |
-
# ----- THIS IS WHERE YOU BUILD YOUR AGENT ------
|
| 18 |
-
class MyAgent:
|
| 19 |
-
def __init__(self):
|
| 20 |
-
# Initialize the model and tokenizer for tool use
|
| 21 |
-
print("Initializing MyAgent with Qwen2.5-72B-Instruct...")
|
| 22 |
-
|
| 23 |
-
# Muista asettaa -> HF_TOKEN secret in your space settings.
|
| 24 |
-
self.agent = HfAgent(
|
| 25 |
-
llm="Qwen/Qwen2.5-72B-Instruct",
|
| 26 |
-
additional_tools=[DuckDuckGoSearchTool()],
|
| 27 |
-
)
|
| 28 |
-
print("Agent initialized with InferenceClient.")
|
| 29 |
-
|
| 30 |
-
def __call__(self, question: str) -> str:
|
| 31 |
-
print(f"Agent received question: {question[:50]}...")
|
| 32 |
-
try:
|
| 33 |
-
result = self.agent.run(question)
|
| 34 |
-
|
| 35 |
-
# The GAIA scoring is an "EXACT MATCH".
|
| 36 |
-
final_answer = str(result)
|
| 37 |
-
print(f"Agent returning answer: {final_answer}")
|
| 38 |
-
return final_answer
|
| 39 |
-
except Exception as e:
|
| 40 |
-
print(f"An error occurred during agent execution: {e}")
|
| 41 |
-
return f"Error: {e}"
|
| 42 |
|
| 43 |
-
|
| 44 |
-
|
|
|
|
|
|
|
|
|
|
| 45 |
# --- Determine HF Space Runtime URL and Repo URL ---
|
| 46 |
-
space_id = os.getenv("SPACE_ID")
|
| 47 |
|
| 48 |
if profile:
|
| 49 |
username= f"{profile.username}"
|
|
@@ -58,7 +33,7 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
|
|
| 58 |
|
| 59 |
# 1. Instantiate Agent ( modify this part to create your agent)
|
| 60 |
try:
|
| 61 |
-
agent =
|
| 62 |
except Exception as e:
|
| 63 |
print(f"Error instantiating agent: {e}")
|
| 64 |
return f"Error initializing agent: {e}", None
|
|
@@ -94,11 +69,12 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
|
|
| 94 |
for item in questions_data:
|
| 95 |
task_id = item.get("task_id")
|
| 96 |
question_text = item.get("question")
|
|
|
|
| 97 |
if not task_id or question_text is None:
|
| 98 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 99 |
continue
|
| 100 |
try:
|
| 101 |
-
submitted_answer = agent(question_text)
|
| 102 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
| 103 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
|
| 104 |
except Exception as e:
|
|
@@ -109,7 +85,7 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
|
|
| 109 |
print("Agent did not produce any answers to submit.")
|
| 110 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
| 111 |
|
| 112 |
-
# 4. Prepare Submission
|
| 113 |
submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
|
| 114 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 115 |
print(status_update)
|
|
@@ -157,16 +133,20 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
|
|
| 157 |
results_df = pd.DataFrame(results_log)
|
| 158 |
return status_message, results_df
|
| 159 |
|
|
|
|
| 160 |
# --- Build Gradio Interface using Blocks ---
|
| 161 |
with gr.Blocks() as demo:
|
| 162 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 163 |
gr.Markdown(
|
| 164 |
"""
|
| 165 |
**Instructions:**
|
| 166 |
-
|
| 167 |
1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
|
| 168 |
2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
|
| 169 |
3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 170 |
"""
|
| 171 |
)
|
| 172 |
|
|
@@ -175,6 +155,7 @@ with gr.Blocks() as demo:
|
|
| 175 |
run_button = gr.Button("Run Evaluation & Submit All Answers")
|
| 176 |
|
| 177 |
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
|
|
|
|
| 178 |
results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
|
| 179 |
|
| 180 |
run_button.click(
|
|
|
|
| 1 |
# 17/08/2025 -AO
|
| 2 |
+
|
| 3 |
import os
|
| 4 |
import gradio as gr
|
| 5 |
import requests
|
| 6 |
import inspect
|
| 7 |
import pandas as pd
|
| 8 |
+
from agent import BasicAgent
|
| 9 |
+
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
# --- Constants ---
|
| 12 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
+
def run_and_submit_all( profile: gr.OAuthProfile | None):
|
| 16 |
+
"""
|
| 17 |
+
Fetches all questions, runs the BasicAgent on them, submits all answers,
|
| 18 |
+
and displays the results.
|
| 19 |
+
"""
|
| 20 |
# --- Determine HF Space Runtime URL and Repo URL ---
|
| 21 |
+
space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
|
| 22 |
|
| 23 |
if profile:
|
| 24 |
username= f"{profile.username}"
|
|
|
|
| 33 |
|
| 34 |
# 1. Instantiate Agent ( modify this part to create your agent)
|
| 35 |
try:
|
| 36 |
+
agent = BasicAgent()
|
| 37 |
except Exception as e:
|
| 38 |
print(f"Error instantiating agent: {e}")
|
| 39 |
return f"Error initializing agent: {e}", None
|
|
|
|
| 69 |
for item in questions_data:
|
| 70 |
task_id = item.get("task_id")
|
| 71 |
question_text = item.get("question")
|
| 72 |
+
file_name_text = item.get("file_name")
|
| 73 |
if not task_id or question_text is None:
|
| 74 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 75 |
continue
|
| 76 |
try:
|
| 77 |
+
submitted_answer = agent(question_text, file_name_text)
|
| 78 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
| 79 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
|
| 80 |
except Exception as e:
|
|
|
|
| 85 |
print("Agent did not produce any answers to submit.")
|
| 86 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
| 87 |
|
| 88 |
+
# 4. Prepare Submission
|
| 89 |
submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
|
| 90 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 91 |
print(status_update)
|
|
|
|
| 133 |
results_df = pd.DataFrame(results_log)
|
| 134 |
return status_message, results_df
|
| 135 |
|
| 136 |
+
|
| 137 |
# --- Build Gradio Interface using Blocks ---
|
| 138 |
with gr.Blocks() as demo:
|
| 139 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 140 |
gr.Markdown(
|
| 141 |
"""
|
| 142 |
**Instructions:**
|
|
|
|
| 143 |
1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
|
| 144 |
2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
|
| 145 |
3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
|
| 146 |
+
---
|
| 147 |
+
**Disclaimers:**
|
| 148 |
+
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).
|
| 149 |
+
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.
|
| 150 |
"""
|
| 151 |
)
|
| 152 |
|
|
|
|
| 155 |
run_button = gr.Button("Run Evaluation & Submit All Answers")
|
| 156 |
|
| 157 |
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
|
| 158 |
+
# Removed max_rows=10 from DataFrame constructor
|
| 159 |
results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
|
| 160 |
|
| 161 |
run_button.click(
|