ab_smoke_bucketed.py
6.7 KB
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
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
#!/usr/bin/env python3
from __future__ import annotations
import argparse
import json
import subprocess
from pathlib import Path
from statistics import mean
PYTHON = "/usr/local/miniconda3/bin/python"
def run(cmd: list[str], cwd: Path) -> str:
return subprocess.check_output(cmd, cwd=str(cwd), text=True)
def collect_files(input_dir: Path, patterns: list[str], limit: int | None = None) -> list[Path]:
seen: set[Path] = set()
files: list[Path] = []
for pattern in patterns:
for path in sorted(input_dir.glob(pattern)):
if not path.is_file() or path.suffix.lower() != ".mp3":
continue
resolved = path.resolve()
if resolved in seen:
continue
seen.add(resolved)
files.append(resolved)
if limit is not None and len(files) >= limit:
return files
return files
def ensure_bucket_subset(input_dir: Path, bucket_dir: Path, patterns: list[str], limit: int | None) -> dict:
bucket_dir.mkdir(parents=True, exist_ok=True)
files = collect_files(input_dir, patterns, limit=limit)
copied: list[str] = []
for src in files:
rel = src.relative_to(input_dir)
dst = bucket_dir / rel
dst.parent.mkdir(parents=True, exist_ok=True)
if not dst.exists():
dst.write_bytes(src.read_bytes())
copied.append(str(dst))
return {
"num_files": len(copied),
"sample_files": copied[:5],
}
def main() -> None:
parser = argparse.ArgumentParser(description="Run bucket/style-aware segmented smoke benchmarks")
parser.add_argument("--dataset", default="fma")
parser.add_argument("--input-dir", default="data/raw/fma_small_audio")
parser.add_argument("--bucket-config", required=True, help="JSON file with {buckets:[{name,patterns,subset_size?}]} or {bucket_name:[patterns]}")
parser.add_argument("--work-root", default="/tmp/ab_smoke_bucketed")
parser.add_argument("--query-duration", type=float, default=8.0)
parser.add_argument("--query-stride", type=float, default=None)
parser.add_argument("--train-epochs", type=int, default=1)
parser.add_argument("--batch-size", type=int, default=2)
parser.add_argument("--device", default="cpu")
parser.add_argument("--seed", type=int, default=42)
parser.add_argument("--max-test-queries", type=int, default=None)
parser.add_argument("--default-subset-size", type=int, default=16)
parser.add_argument("--min-files", type=int, default=2)
parser.add_argument("--strategies", nargs="*", default=["high_energy", "hybrid"])
parser.add_argument("--output-json", default=None)
args = parser.parse_args()
repo = Path(__file__).resolve().parents[1]
input_dir = (repo / args.input_dir).resolve()
work_root = Path(args.work_root).resolve()
config = json.loads(Path(args.bucket_config).read_text())
if isinstance(config, dict) and "buckets" in config:
bucket_specs = config["buckets"]
elif isinstance(config, dict):
bucket_specs = [{"name": k, "patterns": v} for k, v in config.items()]
else:
raise ValueError("bucket config must be an object")
bucket_reports = []
for spec in bucket_specs:
name = spec["name"]
patterns = spec["patterns"]
subset_size = spec.get("subset_size", args.default_subset_size)
bucket_root = work_root / name
subset_dir = bucket_root / "bucket_input"
subset_info = ensure_bucket_subset(input_dir, subset_dir, patterns, subset_size)
if subset_info["num_files"] < args.min_files:
bucket_reports.append({
"bucket": name,
"patterns": patterns,
"subset_size": subset_info["num_files"],
"skipped": True,
"reason": f"num_files<{args.min_files}",
"subset": subset_info,
})
continue
cmd = [
PYTHON,
"scripts/ab_smoke_segmentation.py",
"--dataset", args.dataset,
"--input-dir", str(subset_dir),
"--work-root", str(bucket_root / "run"),
"--subset-size", str(subset_info["num_files"]),
"--query-duration", str(args.query_duration),
"--train-epochs", str(args.train_epochs),
"--batch-size", str(args.batch_size),
"--device", args.device,
"--seed", str(args.seed),
"--strategies", *args.strategies,
]
if args.max_test_queries is not None:
cmd += ["--max-test-queries", str(args.max_test_queries)]
if args.query_stride is not None:
cmd += ["--query-stride", str(args.query_stride)]
out_json = bucket_root / "bucket_report.json"
cmd += ["--output-json", str(out_json)]
run(cmd, cwd=repo)
result = json.loads(out_json.read_text())
bucket_reports.append({
"bucket": name,
"patterns": patterns,
"subset_size": subset_info["num_files"],
"skipped": False,
"subset": subset_info,
"winner": result.get("winner"),
"strategies": result.get("strategies", []),
})
strategy_aggregate: dict[str, dict[str, list[float]]] = {}
for bucket in bucket_reports:
if bucket.get("skipped"):
continue
for row in bucket["strategies"]:
agg = strategy_aggregate.setdefault(row["strategy"], {"top1": [], "topk": [], "num_queries": []})
agg["top1"].append(row["top1"])
agg["topk"].append(row["topk"])
agg["num_queries"].append(row["num_queries"])
aggregate = {
strategy: {
"bucket_runs": len(vals["top1"]),
"mean_top1": round(mean(vals["top1"]), 4),
"mean_topk": round(mean(vals["topk"]), 4),
"mean_num_queries": round(mean(vals["num_queries"]), 4),
}
for strategy, vals in strategy_aggregate.items() if vals["top1"]
}
report = {
"dataset": args.dataset,
"input_dir": str(input_dir),
"bucket_config": str(Path(args.bucket_config).resolve()),
"query_duration": args.query_duration,
"query_stride": args.query_stride,
"train_epochs": args.train_epochs,
"batch_size": args.batch_size,
"device": args.device,
"seed": args.seed,
"max_test_queries": args.max_test_queries,
"strategies": args.strategies,
"buckets": bucket_reports,
"aggregate": aggregate,
}
text = json.dumps(report, ensure_ascii=False, indent=2)
if args.output_json:
out = Path(args.output_json)
out.parent.mkdir(parents=True, exist_ok=True)
out.write_text(text)
print(text)
if __name__ == "__main__":
main()