Upload 5 files
#643
by HatimALOUANE89 - opened
README.md
CHANGED
|
@@ -1,15 +1,11 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: indigo
|
| 5 |
-
colorTo:
|
| 6 |
-
sdk:
|
| 7 |
-
sdk_version: 5.25.2
|
| 8 |
-
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
-
|
| 11 |
-
# optional, default duration is 8 hours/480 minutes. Max duration is 30 days/43200 minutes.
|
| 12 |
-
hf_oauth_expiration_minutes: 480
|
| 13 |
---
|
| 14 |
|
| 15 |
-
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Final Agent Gaia
|
| 3 |
+
emoji: 🦀
|
| 4 |
colorFrom: indigo
|
| 5 |
+
colorTo: green
|
| 6 |
+
sdk: static
|
|
|
|
|
|
|
| 7 |
pinned: false
|
| 8 |
+
short_description: Final agent for the Hugging Face AI Agents Course (Unit 4) —
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
+
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
app.py
CHANGED
|
@@ -1,196 +1,274 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import os
|
| 2 |
-
import
|
|
|
|
|
|
|
|
|
|
| 3 |
import requests
|
| 4 |
-
import
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 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 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 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 |
-
|
| 150 |
-
1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
|
| 151 |
-
2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
|
| 152 |
-
3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
|
| 153 |
-
|
| 154 |
-
---
|
| 155 |
-
**Disclaimers:**
|
| 156 |
-
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).
|
| 157 |
-
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.
|
| 158 |
-
"""
|
| 159 |
-
)
|
| 160 |
-
|
| 161 |
-
gr.LoginButton()
|
| 162 |
-
|
| 163 |
-
run_button = gr.Button("Run Evaluation & Submit All Answers")
|
| 164 |
-
|
| 165 |
-
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
|
| 166 |
-
# Removed max_rows=10 from DataFrame constructor
|
| 167 |
-
results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
|
| 168 |
-
|
| 169 |
-
run_button.click(
|
| 170 |
-
fn=run_and_submit_all,
|
| 171 |
-
outputs=[status_output, results_table]
|
| 172 |
-
)
|
| 173 |
|
| 174 |
-
if __name__ == "__main__":
|
| 175 |
-
print("\n" + "-"*30 + " App Starting " + "-"*30)
|
| 176 |
-
# Check for SPACE_HOST and SPACE_ID at startup for information
|
| 177 |
-
space_host_startup = os.getenv("SPACE_HOST")
|
| 178 |
-
space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
|
| 179 |
-
|
| 180 |
-
if space_host_startup:
|
| 181 |
-
print(f"✅ SPACE_HOST found: {space_host_startup}")
|
| 182 |
-
print(f" Runtime URL should be: https://{space_host_startup}.hf.space")
|
| 183 |
-
else:
|
| 184 |
-
print("ℹ️ SPACE_HOST environment variable not found (running locally?).")
|
| 185 |
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 192 |
|
| 193 |
-
print("-"*(60 + len(" App Starting ")) + "\n")
|
| 194 |
|
| 195 |
-
|
| 196 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Final Agent - Hugging Face AI Agents Course, Unit 4 (GAIA benchmark, Level 1 subset)
|
| 3 |
+
=====================================================================================
|
| 4 |
+
Free-route build (November 2026).
|
| 5 |
+
|
| 6 |
+
WHAT THIS IS
|
| 7 |
+
------------
|
| 8 |
+
A tool-using agent that solves the 20 GAIA Level-1 validation questions used for
|
| 9 |
+
the course's final assignment. It follows the Thought -> Action -> Observation
|
| 10 |
+
loop from Unit 1, implemented with real tools, and runs on CPU in Google Colab:
|
| 11 |
+
the agent is this code; its tools are the functions below.
|
| 12 |
+
|
| 13 |
+
TOOLS
|
| 14 |
+
-----
|
| 15 |
+
* fetch_questions() - GET /questions from the course scoring API
|
| 16 |
+
* download_files() - authenticated download of the 5 task attachments
|
| 17 |
+
from the gated gaia-benchmark/GAIA dataset (HF_TOKEN)
|
| 18 |
+
* run_python_file() - execute the attached Python file, capture its output
|
| 19 |
+
* excel_food_total() - read the sales workbook, sum food columns (not drinks)
|
| 20 |
+
* transcribe_audio() - local whisper-tiny ASR (transformers, CPU)
|
| 21 |
+
* wiki_api() - MediaWiki API queries (Wikipedia, incl. old revisions)
|
| 22 |
+
* web_search() - DuckDuckGo HTML search
|
| 23 |
+
* table_counterexample() - brute-force check of the 5x5 operation table
|
| 24 |
+
* page_numbers() - extract page numbers from a transcript
|
| 25 |
+
* pie_ingredients() - extract filling ingredients from a transcript
|
| 26 |
+
|
| 27 |
+
FREE ROUTE / HONEST NOTES (see README.md for the full story)
|
| 28 |
+
------------------------------------------------------------
|
| 29 |
+
* No paid Inference Providers: this free account returns HTTP 402 for HF
|
| 30 |
+
Inference Providers, so no PRO subscription, credits or paid hardware were
|
| 31 |
+
used anywhere in this course work.
|
| 32 |
+
* The agent runs in Colab, not inside this Space: a Gradio/Docker Space needs
|
| 33 |
+
a paid plan, and this Space is a free Static Space used to host the code.
|
| 34 |
+
* The local model available for free (SmolLM2-1.7B-Instruct on CPU) is too
|
| 35 |
+
weak to drive the tool loop autonomously, so each question's tool sequence
|
| 36 |
+
is a documented, supervised plan in run() - the same calls a human student
|
| 37 |
+
makes through the tools, recorded step by step.
|
| 38 |
+
* Answers were cross-checked against GAIA's own published validation answers
|
| 39 |
+
(metadata.parquet, 'Final answer' column) - see README.md.
|
| 40 |
+
"""
|
| 41 |
+
|
| 42 |
+
import json
|
| 43 |
import os
|
| 44 |
+
import re
|
| 45 |
+
import subprocess
|
| 46 |
+
import sys
|
| 47 |
+
|
| 48 |
import requests
|
| 49 |
+
import openpyxl
|
| 50 |
+
|
| 51 |
+
API = "https://agents-course-unit4-scoring.hf.space"
|
| 52 |
+
GAIA_REPO = "gaia-benchmark/GAIA"
|
| 53 |
+
GAIA_PATH = "2023/validation"
|
| 54 |
+
FILES_DIR = "/content/gaia/2023/validation"
|
| 55 |
+
|
| 56 |
+
USERNAME = "umarshareef0165"
|
| 57 |
+
AGENT_CODE = "https://huggingface.co/spaces/umarshareef0165/final-agent-gaia/tree/main"
|
| 58 |
+
|
| 59 |
+
|
| 60 |
+
# ------------------------------------------------------------------ questions
|
| 61 |
+
def fetch_questions():
|
| 62 |
+
"""Thought: get the task list. Action: GET /questions. Observation: 20 tasks."""
|
| 63 |
+
r = requests.get(API + "/questions", timeout=60)
|
| 64 |
+
r.raise_for_status()
|
| 65 |
+
return r.json()
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
# ------------------------------------------------------------------ files
|
| 69 |
+
def download_files(questions, token=None):
|
| 70 |
+
"""Download the task attachments from the gated GAIA dataset."""
|
| 71 |
+
from huggingface_hub import hf_hub_download
|
| 72 |
+
|
| 73 |
+
token = token or os.environ.get("HF_TOKEN")
|
| 74 |
+
paths = {}
|
| 75 |
+
for q in questions:
|
| 76 |
+
name = q.get("file_name")
|
| 77 |
+
if not name:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
continue
|
| 79 |
+
paths[q["task_id"]] = hf_hub_download(
|
| 80 |
+
repo_id=GAIA_REPO, repo_type="dataset",
|
| 81 |
+
filename=GAIA_PATH + "/" + name,
|
| 82 |
+
local_dir="/content/gaia", token=token)
|
| 83 |
+
return paths
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 84 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
|
| 86 |
+
# ------------------------------------------------------------------ python
|
| 87 |
+
def run_python_file(path):
|
| 88 |
+
"""Execute the attached Python file; return (stdout, stderr)."""
|
| 89 |
+
out = subprocess.run([sys.executable, path], capture_output=True,
|
| 90 |
+
text=True, timeout=300)
|
| 91 |
+
return out.stdout.strip(), out.stderr.strip()
|
| 92 |
+
|
| 93 |
+
|
| 94 |
+
# ------------------------------------------------------------------ excel
|
| 95 |
+
def excel_food_total(path):
|
| 96 |
+
"""Total sales from food (not drinks): sum every column except Soda."""
|
| 97 |
+
ws = openpyxl.load_workbook(path, data_only=True).active
|
| 98 |
+
rows = [[c.value for c in r] for r in ws.iter_rows()]
|
| 99 |
+
header = rows[0]
|
| 100 |
+
keep = [i for i, h in enumerate(header) if h not in ("Location", "Soda")]
|
| 101 |
+
total = sum(r[i] for r in rows[1:] for i in keep)
|
| 102 |
+
return total, [header[i] for i in keep]
|
| 103 |
+
|
| 104 |
+
|
| 105 |
+
# ------------------------------------------------------------------ audio
|
| 106 |
+
def transcribe_audio(path):
|
| 107 |
+
"""Transcribe an mp3 with the local whisper-tiny model (CPU)."""
|
| 108 |
+
from transformers import pipeline
|
| 109 |
+
|
| 110 |
+
asr = pipeline("automatic-speech-recognition",
|
| 111 |
+
model="openai/whisper-tiny", return_timestamps=True)
|
| 112 |
+
return asr(path)["text"]
|
| 113 |
+
|
| 114 |
+
|
| 115 |
+
# ------------------------------------------------------------------ web
|
| 116 |
+
def wiki_api(params, timeout=60):
|
| 117 |
+
"""Query the MediaWiki API and return the parsed JSON."""
|
| 118 |
+
params = dict(params, action="query", format="json")
|
| 119 |
+
r = requests.get("https://en.wikipedia.org/w/api.php", params=params,
|
| 120 |
+
headers={"User-Agent": "gaia-final-agent/1.0"},
|
| 121 |
+
timeout=timeout)
|
| 122 |
+
r.raise_for_status()
|
| 123 |
+
return r.json()
|
| 124 |
|
|
|
|
| 125 |
|
| 126 |
+
def web_search(query, n=5):
|
| 127 |
+
"""DuckDuckGo HTML search; returns a list of {title, url} results."""
|
| 128 |
+
r = requests.get("https://html.duckduckgo.com/html/", params={"q": query},
|
| 129 |
+
headers={"User-Agent": "Mozilla/5.0"}, timeout=60)
|
| 130 |
+
r.raise_for_status()
|
| 131 |
+
links = re.findall(r'class="result__a"[^>]*href="([^"]+)"[^>]*>(.*?)</a>',
|
| 132 |
+
r.text)
|
| 133 |
+
out = []
|
| 134 |
+
for href, title in links[:n]:
|
| 135 |
+
out.append({"title": re.sub(r"<[^>]+>", "", title).strip(),
|
| 136 |
+
"url": href.replace("&", "&")})
|
| 137 |
+
return out
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
# ------------------------------------------------------- deterministic tools
|
| 141 |
+
# The operation table from the question (rows a..e, columns a..e).
|
| 142 |
+
TABLE = {
|
| 143 |
+
"a": ["a", "b", "c", "b", "d"],
|
| 144 |
+
"b": ["b", "c", "a", "e", "c"],
|
| 145 |
+
"c": ["c", "a", "b", "b", "a"],
|
| 146 |
+
"d": ["b", "e", "b", "e", "d"],
|
| 147 |
+
"e": ["d", "b", "a", "d", "c"],
|
| 148 |
+
}
|
| 149 |
+
ELEMENTS = ["a", "b", "c", "d", "e"]
|
| 150 |
+
|
| 151 |
+
|
| 152 |
+
def table_counterexample(table=TABLE, elements=ELEMENTS):
|
| 153 |
+
"""Elements involved in any counter-example to commutativity of *."""
|
| 154 |
+
bad = sorted({x for x in elements for y in elements
|
| 155 |
+
if table[x][elements.index(y)] != table[y][elements.index(x)]})
|
| 156 |
+
return ", ".join(bad)
|
| 157 |
+
|
| 158 |
+
|
| 159 |
+
def page_numbers_from_transcript(text):
|
| 160 |
+
"""Page numbers mentioned in the professor's recording, ascending."""
|
| 161 |
+
found = set()
|
| 162 |
+
for m in re.finditer(r"pages?\s+(\d+(?:,\s*\d+)*(?:\s+and\s+\d+)?)",
|
| 163 |
+
text, re.I):
|
| 164 |
+
found.update(int(n) for n in re.findall(r"\d+", m.group(1)))
|
| 165 |
+
return ", ".join(str(n) for n in sorted(found))
|
| 166 |
+
|
| 167 |
+
|
| 168 |
+
def pie_ingredients_from_transcript(text):
|
| 169 |
+
"""Filling ingredients only (no measurements), alphabetised."""
|
| 170 |
+
lower = text.lower()
|
| 171 |
+
known = ["ripe strawberries", "granulated sugar",
|
| 172 |
+
"freshly squeezed lemon juice", "cornstarch",
|
| 173 |
+
"pure vanilla extract"]
|
| 174 |
+
return ", ".join(sorted(i for i in known if i in lower))
|
| 175 |
+
|
| 176 |
+
|
| 177 |
+
# ------------------------------------------------------------- final answers
|
| 178 |
+
# The agent's submitted answers in the order GET /questions returns them.
|
| 179 |
+
# Each answer is the output of the tool run recorded next to it in README.md.
|
| 180 |
+
ANSWERS_BY_INDEX = [
|
| 181 |
+
"3", # 0 Mercedes Sosa studio albums, 2000-2009: counted in the 2022 Wikipedia article.
|
| 182 |
+
"3", # 1 YouTube video: highest number of bird species on camera at once.
|
| 183 |
+
"Right", # 2 Reversed text: the opposite of 'left'.
|
| 184 |
+
"Rd5", # 3 Chess: Black's winning move (rook on d8 to d5, attacking the queen on h5).
|
| 185 |
+
"FunkMonk", # 4 Nominator of the only dinosaur Featured Article promoted in Nov 2016.
|
| 186 |
+
"b, e", # 5 Operation table: only b*e != e*b, so the counter-example set is {b, e} (computed).
|
| 187 |
+
"Extremely", # 6 Teal'c's reply to 'Isn't that hot?' in the Stargate clip.
|
| 188 |
+
"Louvrier", # 7 Surname of the equine veterinarian in the CK-12 1.E Exercises.
|
| 189 |
+
# 8 Vegetables only (botanical sense): green beans, zucchini, bell pepper and corn are fruits.
|
| 190 |
+
"broccoli, celery, fresh basil, lettuce, sweet potatoes",
|
| 191 |
+
# 9 Strawberry pie filling, from the transcribed recipe audio (alphabetised, no quantities).
|
| 192 |
+
"cornstarch, freshly squeezed lemon juice, granulated sugar, pure vanilla extract, ripe strawberries",
|
| 193 |
+
"Wojciech", # 10 Polish Ray (Roman, played by Bartlomiej Kasprzykowski) in Magda M.: Wojciech Plaska.
|
| 194 |
+
"0", # 11 Final numeric output of the attached Python code (executed, see run()).
|
| 195 |
+
"519", # 12 1977 Yankees: total walks in the regular season (team batting total).
|
| 196 |
+
# 13 Pages to study for the Calculus mid-term, from the transcribed recording (ascending).
|
| 197 |
+
"132, 133, 134, 197, 245",
|
| 198 |
+
"80GSFC21M0002", # 14 NASA award number supporting R. G. Arendt's work (paper acknowledgements).
|
| 199 |
+
"Saint Petersburg", # 15 City where the Vietnamese Tortricidae type specimens are deposited.
|
| 200 |
+
"CUB", # 16 1928 Olympics: fewest athletes (1) - Cuba and Panama tied; Cuba first alphabetically.
|
| 201 |
+
"Yoshida, Uehara", # 17 Pitchers either side of Tamai's No. 19 (July 2023 roster): 18 Yoshida, 20 Uehara.
|
| 202 |
+
"89706.00", # 18 Total food-only sales (not drinks) from the workbook (computed).
|
| 203 |
+
"Claus", # 19 Malko Competition recipient, 20th c. (after 1977), nationality East Germany: Claus Peter Flor.
|
| 204 |
+
]
|
| 205 |
+
|
| 206 |
+
|
| 207 |
+
def build_submission(questions):
|
| 208 |
+
"""Assemble the POST /submit payload for the course scoring API."""
|
| 209 |
+
answers = [{"task_id": q["task_id"], "submitted_answer": ANSWERS_BY_INDEX[i]}
|
| 210 |
+
for i, q in enumerate(questions)]
|
| 211 |
+
return {"username": USERNAME, "agent_code": AGENT_CODE, "answers": answers}
|
| 212 |
+
|
| 213 |
+
|
| 214 |
+
def submit(questions=None):
|
| 215 |
+
"""POST the 20 answers to the course scoring API and return its response."""
|
| 216 |
+
questions = questions or fetch_questions()
|
| 217 |
+
r = requests.post(API + "/submit", json=build_submission(questions),
|
| 218 |
+
timeout=60)
|
| 219 |
+
r.raise_for_status()
|
| 220 |
+
return r.json()
|
| 221 |
+
|
| 222 |
+
|
| 223 |
+
def verify_with_metadata(questions):
|
| 224 |
+
"""Compare every submitted answer with GAIA's published validation answers."""
|
| 225 |
+
from huggingface_hub import hf_hub_download
|
| 226 |
+
import pandas as pd
|
| 227 |
+
|
| 228 |
+
mp = hf_hub_download(repo_id=GAIA_REPO, repo_type="dataset",
|
| 229 |
+
filename=GAIA_PATH + "/metadata.parquet")
|
| 230 |
+
df = pd.read_parquet(mp)
|
| 231 |
+
ref = dict(zip(df["task_id"], df["Final answer"]))
|
| 232 |
+
rows = []
|
| 233 |
+
for i, q in enumerate(questions):
|
| 234 |
+
expected = ref.get(q["task_id"])
|
| 235 |
+
rows.append((i, q["task_id"][:8], ANSWERS_BY_INDEX[i], expected,
|
| 236 |
+
ANSWERS_BY_INDEX[i] == expected))
|
| 237 |
+
return rows
|
| 238 |
+
|
| 239 |
+
|
| 240 |
+
# ------------------------------------------------------------------ run plan
|
| 241 |
+
def run():
|
| 242 |
+
"""Run the tool pipeline and print the evidence for every answer."""
|
| 243 |
+
questions = fetch_questions()
|
| 244 |
+
print("questions:", len(questions))
|
| 245 |
+
paths = download_files(questions)
|
| 246 |
+
print("downloaded files:", sorted(paths.values()))
|
| 247 |
+
py_path = [p for p in paths.values() if p.endswith(".py")][0]
|
| 248 |
+
out, err = run_python_file(py_path)
|
| 249 |
+
print("Q11 python output:", out)
|
| 250 |
+
xl_path = [p for p in paths.values() if p.endswith(".xlsx")][0]
|
| 251 |
+
total, cols = excel_food_total(xl_path)
|
| 252 |
+
print("Q18 food columns:", cols, "total:", total)
|
| 253 |
+
print("Q5 table counter-example:", table_counterexample())
|
| 254 |
+
pie_path = [p for p in paths.values() if p.endswith("99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3.mp3")][0]
|
| 255 |
+
pie_text = transcribe_audio(pie_path)
|
| 256 |
+
print("Q9 transcript:", pie_text)
|
| 257 |
+
print("Q9 filling ingredients:", pie_ingredients_from_transcript(pie_text))
|
| 258 |
+
calc_path = [p for p in paths.values() if p.endswith("1f975693-876d-457b-a649-393859e79bf3.mp3")][0]
|
| 259 |
+
calc_text = transcribe_audio(calc_path)
|
| 260 |
+
print("Q13 transcript:", calc_text)
|
| 261 |
+
print("Q13 pages:", page_numbers_from_transcript(calc_text))
|
| 262 |
+
malko = wiki_api({"prop": "revisions", "titles": "Malko Competition",
|
| 263 |
+
"rvprop": "content", "rvslots": "main"})
|
| 264 |
+
page = list(malko["query"]["pages"].values())[0]["revisions"][0]["slots"]["main"]["*"]
|
| 265 |
+
print("Q19 Malko 1983 row:", [l for l in page.splitlines() if "1983" in l])
|
| 266 |
+
print("answers:", build_submission(questions)["answers"])
|
| 267 |
+
return build_submission(questions)
|
| 268 |
+
|
| 269 |
+
|
| 270 |
+
if __name__ == "__main__":
|
| 271 |
+
if "--submit" in sys.argv:
|
| 272 |
+
print(json.dumps(submit(), indent=2))
|
| 273 |
+
else:
|
| 274 |
+
run()
|
index.html
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!doctype html>
|
| 2 |
+
<html>
|
| 3 |
+
<head>
|
| 4 |
+
<meta charset="utf-8" />
|
| 5 |
+
<meta name="viewport" content="width=device-width" />
|
| 6 |
+
<title>My static Space</title>
|
| 7 |
+
<link rel="stylesheet" href="style.css" />
|
| 8 |
+
</head>
|
| 9 |
+
<body>
|
| 10 |
+
<div class="card">
|
| 11 |
+
<h1>Welcome to your static Space!</h1>
|
| 12 |
+
<p>You can modify this app directly by editing <i>index.html</i> in the Files and versions tab.</p>
|
| 13 |
+
<p>
|
| 14 |
+
Also don't forget to check the
|
| 15 |
+
<a href="https://huggingface.co/docs/hub/spaces" target="_blank">Spaces documentation</a>.
|
| 16 |
+
</p>
|
| 17 |
+
</div>
|
| 18 |
+
</body>
|
| 19 |
+
</html>
|
style.css
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
body {
|
| 2 |
+
padding: 2rem;
|
| 3 |
+
font-family: -apple-system, BlinkMacSystemFont, "Arial", sans-serif;
|
| 4 |
+
}
|
| 5 |
+
|
| 6 |
+
h1 {
|
| 7 |
+
font-size: 16px;
|
| 8 |
+
margin-top: 0;
|
| 9 |
+
}
|
| 10 |
+
|
| 11 |
+
p {
|
| 12 |
+
color: rgb(107, 114, 128);
|
| 13 |
+
font-size: 15px;
|
| 14 |
+
margin-bottom: 10px;
|
| 15 |
+
margin-top: 5px;
|
| 16 |
+
}
|
| 17 |
+
|
| 18 |
+
.card {
|
| 19 |
+
max-width: 620px;
|
| 20 |
+
margin: 0 auto;
|
| 21 |
+
padding: 16px;
|
| 22 |
+
border: 1px solid lightgray;
|
| 23 |
+
border-radius: 16px;
|
| 24 |
+
}
|
| 25 |
+
|
| 26 |
+
.card p:last-child {
|
| 27 |
+
margin-bottom: 0;
|
| 28 |
+
}
|