doc

src/leaderboard/read_evals.py

@@ -14,6 +14,8 @@ from src.submission.check_validity import is_model_on_hub
 
 @dataclass
 class EvalResult:
+    """Represents one full evaluation. Built from a combination of the result and request file for a given run.
+    """
     eval_name: str # org_model_precision (uid)
     full_model: str # org/model (path on hub)
     org: str 

src/populate.py

@@ -9,6 +9,7 @@ from src.leaderboard.read_evals import get_raw_eval_results
 
 
 def get_leaderboard_df(results_path: str, requests_path: str, cols: list, benchmark_cols: list) -> pd.DataFrame:
+    """Creates a dataframe from all the individual experiment results"""
     raw_data = get_raw_eval_results(results_path, requests_path)
     all_data_json = [v.to_dict() for v in raw_data]
 
@@ -22,6 +23,7 @@ def get_leaderboard_df(results_path: str, requests_path: str, cols: list, benchm
 
 
 def get_evaluation_queue_df(save_path: str, cols: list) -> list[pd.DataFrame]:
+    """Creates the different dataframes for the evaluation queues requestes"""
     entries = [entry for entry in os.listdir(save_path) if not entry.startswith(".")]
     all_evals = []
 

src/submission/check_validity.py

@@ -32,6 +32,7 @@ def check_model_card(repo_id: str) -> tuple[bool, str]:
     return True, ""
 
 def is_model_on_hub(model_name: str, revision: str, token: str = None, trust_remote_code=False, test_tokenizer=False) -> tuple[bool, str]:
+    """Checks if the model model_name is on the hub, and whether it (and its tokenizer) can be loaded with AutoClasses."""
     try:
         config = AutoConfig.from_pretrained(model_name, revision=revision, trust_remote_code=trust_remote_code, token=token)
         if test_tokenizer:
@@ -74,6 +75,7 @@ def get_model_arch(model_info: ModelInfo):
     return model_info.config.get("architectures", "Unknown")
 
 def already_submitted_models(requested_models_dir: str) -> set[str]:
+    """Gather a list of already submitted models to avoid duplicates"""
     depth = 1
     file_names = []
     users_to_submission_dates = defaultdict(list)