hi-melnikov commited on
Commit
d3db3e5
1 Parent(s): d0e8be9

remove ruff cache

Browse files
src/gen/data/arena-hard-v0.1/model_answer/external/gigachat_lite.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
src/gen/data/arena-hard-v0.1/model_answer/external/private/var/folders/ws/s9058_gn5cs181gs2_54lcvc0000gn/T/gradio/4a99fae57971a5f7e281df57ab8739fd979a9345/16.o1.csv DELETED
@@ -1,11 +0,0 @@
1
- Col1.Col2.Col3.Col4.Col5.Col6.Col7.Col8.Col9.Col10
2
- 1.2.5.6.2.6.3.7.8.8
3
- 10.10.10.7.8.3.8.9.4.8
4
- 5.9.2.10.7.7.4.9.2.3
5
- 4.8.2.9.8.7.6.6.9.4
6
- 1.8.7.3.1.6.7.7.6.1
7
- 9.9.6.2.1.5.5.2.5.5
8
- 8.2.10.5.10.10.7.6.3.6
9
- 6.1.8.3.3.4.7.7.8.5
10
- 7.1.3.3.2.4.5.9.5.6
11
- 4.1.4.4.6.1.2.6.9.2
 
 
 
 
 
 
 
 
 
 
 
 
src/gen/data/arena-hard-v0.1/model_answer/internal/gpt-3.5-turbo-0125.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
src/gen/data/arena-hard-v0.1/model_judgement/gpt-4-1106-preview/gigachat_lite.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
src/gen/data/arena-hard-v0.1/model_judgement/gpt-4-1106-preview/gigachat_pro.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
src/leaderboard/build_leaderboard.py CHANGED
@@ -19,7 +19,7 @@ def time_diff_wrapper(func):
19
  result = func(*args, **kwargs)
20
  end_time = time.time()
21
  diff = end_time - start_time
22
- logging.info(f"Time taken for {func.__name__}: {diff} seconds")
23
  return result
24
 
25
  return wrapper
@@ -45,33 +45,34 @@ def download_dataset(repo_id, local_dir, repo_type="dataset", max_attempts=3, ba
45
  return
46
  except Exception as e:
47
  wait_time = backoff_factor**attempt
48
- logging.error(f"Error downloading {repo_id}: {e}, retrying in {wait_time}s")
49
  time.sleep(wait_time)
50
  attempt += 1
51
- logging.error(f"Failed to download {repo_id} after {max_attempts} attempts")
52
 
53
 
54
  def build_leadearboard_df():
55
  """Initializes the application space, loading only necessary data."""
56
 
57
- # download_dataset(QUEUE_REPO, EVAL_REQUESTS_PATH)
58
- # download_dataset(DYNAMIC_INFO_REPO, DYNAMIC_INFO_PATH)
59
  download_dataset("Vikhrmodels/openbench-eval", EVAL_RESULTS_PATH)
60
  # print(subprocess.Popen('ls src'))
 
 
61
  subprocess.run(
62
  [
63
  "rsync",
64
- "-avzP",
65
  "--ignore-existing",
66
- f"{EVAL_RESULTS_PATH}/external/*",
67
- "src/gen/data/arena-hard-v0.1/model_answer/",
68
  ],
69
  check=False,
70
  )
71
  subprocess.run(
72
  [
73
  "rsync",
74
- "-avzP",
75
  "--ignore-existing",
76
  f"{EVAL_RESULTS_PATH}/model_judgment/*",
77
  "src/gen/data/arena-hard-v0.1/model_judgement/",
@@ -80,5 +81,6 @@ def build_leadearboard_df():
80
  )
81
 
82
  # Retrieve the leaderboard DataFrame
83
- leaderboard_df = pd.DataFrame.from_records(json.load(open("eval-results/evals/upd.json", "r")))
 
84
  return leaderboard_df.copy()
 
19
  result = func(*args, **kwargs)
20
  end_time = time.time()
21
  diff = end_time - start_time
22
+ logging.info("Time taken for %s: %s seconds", func.__name__, diff)
23
  return result
24
 
25
  return wrapper
 
45
  return
46
  except Exception as e:
47
  wait_time = backoff_factor**attempt
48
+ logging.error("Error downloading %s: %s, retrying in %ss", repo_id, e, wait_time)
49
  time.sleep(wait_time)
50
  attempt += 1
51
+ logging.error("Failed to download %s after %s attempts", repo_id, max_attempts)
52
 
53
 
54
  def build_leadearboard_df():
55
  """Initializes the application space, loading only necessary data."""
56
 
57
+ # download answers of different models that we trust
 
58
  download_dataset("Vikhrmodels/openbench-eval", EVAL_RESULTS_PATH)
59
  # print(subprocess.Popen('ls src'))
60
+
61
+ # copy the grusted to
62
  subprocess.run(
63
  [
64
  "rsync",
65
+ "-azP",
66
  "--ignore-existing",
67
+ f"{EVAL_RESULTS_PATH}/internal/*.jsonl",
68
+ "src/gen/data/arena-hard-v0.1/model_answer/interla/*",
69
  ],
70
  check=False,
71
  )
72
  subprocess.run(
73
  [
74
  "rsync",
75
+ "-azP",
76
  "--ignore-existing",
77
  f"{EVAL_RESULTS_PATH}/model_judgment/*",
78
  "src/gen/data/arena-hard-v0.1/model_judgement/",
 
81
  )
82
 
83
  # Retrieve the leaderboard DataFrame
84
+ with open("eval-results/evals/upd.json", "r", encoding="utf-8") as eval_file:
85
+ leaderboard_df = pd.DataFrame.from_records(json.load(eval_file))
86
  return leaderboard_df.copy()