Spaces:
Sleeping
Sleeping
File size: 1,494 Bytes
37ff7c9 | 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 | from __future__ import annotations
import json
import time
import joblib
import pandas as pd
from credexp.config import settings
from credexp.data.io import processed_dir
def load_data():
holdout_path = processed_dir() / "api_holdout.parquet"
df = pd.read_parquet(holdout_path)
if "TARGET" in df.columns:
df = df.drop(columns=["TARGET"])
return df.head(100).copy()
def main() -> None:
pipe = joblib.load(settings.artifacts_dir / "models" / "pipeline.joblib")
df = load_data()
single = df.head(1).copy()
batch = df.head(50).copy()
# 50 single-row calls
t0 = time.perf_counter()
for _ in range(50):
_ = pipe.predict_proba(single)
single_total_ms = (time.perf_counter() - t0) * 1000
# 1 batch call of 50 rows
t0 = time.perf_counter()
_ = pipe.predict_proba(batch)
batch_total_ms = (time.perf_counter() - t0) * 1000
result = {
"single_total_ms_for_50_calls": single_total_ms,
"single_avg_ms_per_row": single_total_ms / 50,
"batch_total_ms_for_50_rows": batch_total_ms,
"batch_avg_ms_per_row": batch_total_ms / 50,
"speedup_factor_per_row": (single_total_ms / 50) / (batch_total_ms / 50),
}
out_path = "reports/performance/batching_benchmark.json"
with open(out_path, "w", encoding="utf-8") as f:
json.dump(result, f, indent=2)
print(json.dumps(result, indent=2))
print(f"Saved to {out_path}")
if __name__ == "__main__":
main()
|