File size: 3,032 Bytes
fc9f76d |
1 2 3 4 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 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 |
#!/usr/bin/env python3
"""
Script to generate __cached_results.json.gz for the MTEB leaderboard.
This pre-generates the cached results file that the leaderboard uses,
which can save 100+ seconds on fresh leaderboard builds.
Usage:
python generate_cached_results.py
Output:
Creates __cached_results.json.gz in the remote repo root directory
"""
import gzip
import json
import logging
import sys
import time
from pathlib import Path
import mteb
from mteb.cache import ResultCache
logging.basicConfig(
level=logging.INFO,
format='[%(asctime)s] %(levelname)s - %(message)s',
datefmt='%H:%M:%S'
)
logger = logging.getLogger(__name__)
def generate_cached_results():
"""Generate the cached results JSON file."""
start_time = time.time()
logger.info("Initializing ResultCache...")
cache = ResultCache(Path(__file__).parent.parent)
# The remote repo should already be cloned from previous runs
logger.info("Using existing remote results repository...")
# Load all model names
logger.info("Getting all model names...")
models_start = time.time()
all_model_names = [model_meta.name for model_meta in mteb.get_model_metas()]
models_time = time.time() - models_start
logger.info(f"Found {len(all_model_names)} models in {models_time:.2f}s")
# Load results for all models
logger.info("Loading results from cache...")
load_start = time.time()
all_results = cache.load_results(
models=all_model_names,
only_main_score=True,
require_model_meta=False,
include_remote=True,
)
load_time = time.time() - load_start
logger.info(f"Loaded results in {load_time:.2f}s")
# Serialize to JSON and write to gzip file
repo_root = Path(__file__).parent.parent
output_path = repo_root / "__cached_results.json.gz"
logger.info(f"Serializing to JSON and writing to {output_path}...")
write_start = time.time()
json_str = all_results.model_dump_json()
with gzip.open(output_path, 'wt', encoding='utf-8') as f:
f.write(json_str)
write_time = time.time() - write_start
logger.info(f"Serialized and written in {write_time:.2f}s")
# Report file size
file_size_mb = output_path.stat().st_size / (1024 * 1024)
uncompressed_size_mb = len(json_str) / (1024 * 1024)
compression_ratio = (1 - file_size_mb / uncompressed_size_mb) * 100
logger.info(f"Generated {output_path} ({file_size_mb:.1f} MB)")
logger.info(f"Uncompressed size: {uncompressed_size_mb:.1f} MB")
logger.info(f"Compression ratio: {compression_ratio:.1f}%")
total_time = time.time() - start_time
logger.info(f"Total time: {total_time:.2f}s")
return output_path
if __name__ == "__main__":
try:
output_file = generate_cached_results()
logger.info(f"✅ Success! Generated {output_file}")
except Exception as e:
logger.error(f"❌ Failed: {e}")
import traceback
traceback.print_exc()
sys.exit(1)
|