Alessio-Chiovelli commited on
Commit
3dacb75
·
1 Parent(s): a33dabb

added a function to download the element

Browse files
Files changed (3) hide show
  1. Utils/__init__.py +2 -0
  2. Utils/download_GAIA_file.py +18 -0
  3. app.py +2 -2
Utils/__init__.py CHANGED
@@ -12,8 +12,10 @@ from Utils.chains import (
12
  Base64ImageContent, Base64AudioContent, UrlImageContent,
13
  Message, Content, ROLES_PROMPTS, IMAGE_MIME_TYPES
14
  )
 
15
 
16
  __all__ = [
 
17
  'save_results_to_csv',
18
  'bytes_to_base64',
19
  'open_file_bytes',
 
12
  Base64ImageContent, Base64AudioContent, UrlImageContent,
13
  Message, Content, ROLES_PROMPTS, IMAGE_MIME_TYPES
14
  )
15
+ from Utils.download_GAIA_file import get_file_path_by_task_id
16
 
17
  __all__ = [
18
+ 'get_file_path_by_task_id',
19
  'save_results_to_csv',
20
  'bytes_to_base64',
21
  'open_file_bytes',
Utils/download_GAIA_file.py ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from datasets import load_dataset, DownloadConfig
2
+
3
+ # ⚠️ GAIA ha uno script Python custom → trust_remote_code=True
4
+ ds = load_dataset(
5
+ "gaia-benchmark/GAIA",
6
+ name="2023_level1", # oppure 2023_all, 2023_level2, …
7
+ split="validation",
8
+ trust_remote_code=True, # esegue GAIA.py e usa il DownloadManager
9
+ download_config=DownloadConfig(
10
+ resume_download=False # evita path “fantasma”
11
+ ),
12
+ cache_dir="~/.cache/hf_gaia" # dove vuoi tu
13
+ )
14
+ ds_df = ds.to_pandas()
15
+
16
+ def get_file_path_by_task_id(task_id : str) -> str:
17
+ row = ds_df[ds_df['task_id'] == '99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3'].to_dict() or {}
18
+ return row.get("file_path", "")
app.py CHANGED
@@ -6,7 +6,7 @@ import inspect
6
  import pandas as pd
7
 
8
  from agent import AgentGraph as Alfred
9
- # from Utils import save_results_to_csv
10
 
11
  # (Keep Constants as is)
12
  # --- Constants ---
@@ -91,7 +91,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
91
  print(50 * '*')
92
  print(3 * '\n')
93
  if file_name or file_path:
94
- question_text += f" You can access the file {file_name} at path {file_path} to answer the question"
95
  if not task_id or question_text is None:
96
  print(f"Skipping item with missing task_id or question: {item}")
97
  continue
 
6
  import pandas as pd
7
 
8
  from agent import AgentGraph as Alfred
9
+ from Utils import get_file_path_by_task_id
10
 
11
  # (Keep Constants as is)
12
  # --- Constants ---
 
91
  print(50 * '*')
92
  print(3 * '\n')
93
  if file_name or file_path:
94
+ question_text += f"You can access the file {file_name} at path '{get_file_path_by_task_id(task_id)}' to answer the question"
95
  if not task_id or question_text is None:
96
  print(f"Skipping item with missing task_id or question: {item}")
97
  continue