Jimin Huang
commited on
Commit
·
47f193c
1
Parent(s):
25729c8
fix: alter dataset names
Browse files
backend/app/config/hf_config.py
CHANGED
|
@@ -24,11 +24,11 @@ API = HfApi(token=HF_TOKEN)
|
|
| 24 |
|
| 25 |
# Repository configuration
|
| 26 |
QUEUE_REPO = f"{HF_ORGANIZATION}/requests"
|
| 27 |
-
AGGREGATED_REPO = f"{HF_ORGANIZATION}/contents"
|
| 28 |
VOTES_REPO = f"{HF_ORGANIZATION}/votes"
|
| 29 |
MAINTAINERS_HIGHLIGHT_REPO = f"{HF_ORGANIZATION}/maintainers-highlight"
|
| 30 |
|
| 31 |
# File paths from cache config
|
| 32 |
VOTES_PATH = cache_config.votes_file
|
| 33 |
EVAL_REQUESTS_PATH = cache_config.eval_requests_file
|
| 34 |
-
MODEL_CACHE_DIR = cache_config.models_cache
|
|
|
|
| 24 |
|
| 25 |
# Repository configuration
|
| 26 |
QUEUE_REPO = f"{HF_ORGANIZATION}/requests"
|
| 27 |
+
AGGREGATED_REPO = f"{HF_ORGANIZATION}/greek-contents"
|
| 28 |
VOTES_REPO = f"{HF_ORGANIZATION}/votes"
|
| 29 |
MAINTAINERS_HIGHLIGHT_REPO = f"{HF_ORGANIZATION}/maintainers-highlight"
|
| 30 |
|
| 31 |
# File paths from cache config
|
| 32 |
VOTES_PATH = cache_config.votes_file
|
| 33 |
EVAL_REQUESTS_PATH = cache_config.eval_requests_file
|
| 34 |
+
MODEL_CACHE_DIR = cache_config.models_cache
|
backend/app/services/leaderboard.py
CHANGED
|
@@ -17,10 +17,10 @@ class LeaderboardService:
|
|
| 17 |
"""Fetch raw leaderboard data from HuggingFace dataset"""
|
| 18 |
try:
|
| 19 |
logger.info(LogFormatter.section("FETCHING LEADERBOARD DATA"))
|
| 20 |
-
logger.info(LogFormatter.info(f"Loading dataset from {HF_ORGANIZATION}/contents"))
|
| 21 |
|
| 22 |
dataset = datasets.load_dataset(
|
| 23 |
-
f"{HF_ORGANIZATION}/contents",
|
| 24 |
cache_dir=cache_config.get_cache_path("datasets")
|
| 25 |
)["train"]
|
| 26 |
|
|
|
|
| 17 |
"""Fetch raw leaderboard data from HuggingFace dataset"""
|
| 18 |
try:
|
| 19 |
logger.info(LogFormatter.section("FETCHING LEADERBOARD DATA"))
|
| 20 |
+
logger.info(LogFormatter.info(f"Loading dataset from {HF_ORGANIZATION}/greek-contents"))
|
| 21 |
|
| 22 |
dataset = datasets.load_dataset(
|
| 23 |
+
f"{HF_ORGANIZATION}/greek-contents",
|
| 24 |
cache_dir=cache_config.get_cache_path("datasets")
|
| 25 |
)["train"]
|
| 26 |
|
backend/utils/analyze_prod_models.py
CHANGED
|
@@ -31,13 +31,13 @@ def count_evaluated_models():
|
|
| 31 |
"""Count the number of evaluated models"""
|
| 32 |
try:
|
| 33 |
# Get dataset info
|
| 34 |
-
dataset_info = api.dataset_info(repo_id=f"{HF_ORGANIZATION}/contents", repo_type="dataset")
|
| 35 |
|
| 36 |
# Get file list
|
| 37 |
-
files = api.list_repo_files(f"{HF_ORGANIZATION}/contents", repo_type="dataset")
|
| 38 |
|
| 39 |
# Get last commit info
|
| 40 |
-
commits = api.list_repo_commits(f"{HF_ORGANIZATION}/contents", repo_type="dataset")
|
| 41 |
last_commit = next(commits, None)
|
| 42 |
|
| 43 |
# Count lines in jsonl files
|
|
@@ -47,7 +47,7 @@ def count_evaluated_models():
|
|
| 47 |
try:
|
| 48 |
# Download file content
|
| 49 |
content = api.hf_hub_download(
|
| 50 |
-
repo_id=f"{HF_ORGANIZATION}/contents",
|
| 51 |
filename=file,
|
| 52 |
repo_type="dataset"
|
| 53 |
)
|
|
@@ -103,4 +103,4 @@ def main():
|
|
| 103 |
return {"error": str(e)}
|
| 104 |
|
| 105 |
if __name__ == "__main__":
|
| 106 |
-
main()
|
|
|
|
| 31 |
"""Count the number of evaluated models"""
|
| 32 |
try:
|
| 33 |
# Get dataset info
|
| 34 |
+
dataset_info = api.dataset_info(repo_id=f"{HF_ORGANIZATION}/greek-contents", repo_type="dataset")
|
| 35 |
|
| 36 |
# Get file list
|
| 37 |
+
files = api.list_repo_files(f"{HF_ORGANIZATION}/greek-contents", repo_type="dataset")
|
| 38 |
|
| 39 |
# Get last commit info
|
| 40 |
+
commits = api.list_repo_commits(f"{HF_ORGANIZATION}/greek-contents", repo_type="dataset")
|
| 41 |
last_commit = next(commits, None)
|
| 42 |
|
| 43 |
# Count lines in jsonl files
|
|
|
|
| 47 |
try:
|
| 48 |
# Download file content
|
| 49 |
content = api.hf_hub_download(
|
| 50 |
+
repo_id=f"{HF_ORGANIZATION}/greek-contents",
|
| 51 |
filename=file,
|
| 52 |
repo_type="dataset"
|
| 53 |
)
|
|
|
|
| 103 |
return {"error": str(e)}
|
| 104 |
|
| 105 |
if __name__ == "__main__":
|
| 106 |
+
main()
|
backend/utils/sync_datasets_locally.py
CHANGED
|
@@ -29,7 +29,7 @@ DATASET_NAMES = [
|
|
| 29 |
"votes",
|
| 30 |
"results",
|
| 31 |
"requests",
|
| 32 |
-
"contents",
|
| 33 |
"maintainers-highlight",
|
| 34 |
]
|
| 35 |
|
|
@@ -127,4 +127,4 @@ def copy_datasets():
|
|
| 127 |
logger.error(f"❌ Global error: {str(e)}")
|
| 128 |
|
| 129 |
if __name__ == "__main__":
|
| 130 |
-
copy_datasets()
|
|
|
|
| 29 |
"votes",
|
| 30 |
"results",
|
| 31 |
"requests",
|
| 32 |
+
"greek-contents",
|
| 33 |
"maintainers-highlight",
|
| 34 |
]
|
| 35 |
|
|
|
|
| 127 |
logger.error(f"❌ Global error: {str(e)}")
|
| 128 |
|
| 129 |
if __name__ == "__main__":
|
| 130 |
+
copy_datasets()
|