File size: 5,236 Bytes
5755408 | 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 | """Generate ToolFault-Bench.
python scripts/generate.py --out data --seed 20260923
Writes data/test-00000-of-00001.parquet and data/validation-00000-of-00001.parquet.
Deterministic for a given seed.
"""
from __future__ import annotations
import argparse
import json
import os
import random
import sys
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import pandas as pd # noqa: E402
from common import Ctx, dumps, iso, person, sample_now # noqa: E402
from domains import DOMAINS, DOMAIN_BY_KEY # noqa: E402
from templates import PLAN # noqa: E402
VAL_FRACTION = 0.15
def assignments(subtype: str, n: int, rng: random.Random):
"""Yield (domain, extra) pairs for each item of a subtype."""
if subtype == "invalid_args":
for rep in range(n // (len(DOMAINS) * 3)):
for d in DOMAINS:
for k in range(3):
yield d, {"k": k}
return
if subtype == "none_old_timestamp_ok":
for _ in range(4):
yield DOMAIN_BY_KEY["brokerage"], {"weekend": True}
for i in range(n - 4):
yield DOMAINS[i % len(DOMAINS)], {}
return
off = rng.randrange(len(DOMAINS))
for i in range(n):
yield DOMAINS[(i + off) % len(DOMAINS)], {}
def build_messages(d, c, it):
msgs = [{"role": "system", "content": d.system(c)}, {"role": "user", "content": it["request"]}]
k = 0
def add(call, result, note=None):
nonlocal k
k += 1
cid = f"call_{k}"
msgs.append({"role": "assistant", "content": note or "",
"tool_calls": [{"id": cid, "type": "function",
"function": {"name": call.tool, "arguments": json.dumps(call.args, ensure_ascii=False)}}]})
msgs.append({"role": "tool", "tool_call_id": cid, "name": call.tool, "content": dumps(result)})
for p in it["prefix"]:
add(p, p.response)
for call, err, note in it["history"]:
add(call, err, note)
add(it["call"], it["result"], it.get("prior_note"))
return msgs
def generate(seed: int):
rows = []
for subtype, (builder, n) in PLAN.items():
srng = random.Random(f"{seed}:{subtype}")
count = 0
for i, (d, extra) in enumerate(assignments(subtype, n, srng)):
rng = random.Random(f"{seed}:{subtype}:{i}")
c = Ctx(rng=rng, now=sample_now(rng, weekend=extra.get("weekend", False)), org=rng.choice(d.orgs),
user=person(rng), extra=extra)
it = builder(d, c)
tools = d.tools(it["overrides"])
msgs = build_messages(d, c, it)
rows.append({
"domain": d.key,
"subtype": it["subtype"],
"fault_type": it["fault_type"],
"next_action": it["next_action"],
"acceptable_next_actions": list(it["acceptable"]),
"critical_next_actions": list(it["critical"]),
"difficulty": it["difficulty"],
"is_hard_negative": bool(it["is_hard_negative"]),
"current_time": iso(c.now),
"user_request": it["request"],
"tools": json.dumps(tools, ensure_ascii=False),
"messages": json.dumps(msgs, ensure_ascii=False),
"num_tool_calls": sum(1 for m in msgs if m["role"] == "tool"),
"last_tool_name": it["call"].tool,
"last_tool_result": msgs[-1]["content"],
"gold_recovery": it["gold_recovery"],
"rationale": it["rationale"],
"meta": json.dumps(it["meta"], ensure_ascii=False),
})
count += 1
assert count == n, (subtype, count, n)
return rows
def split_and_id(rows, seed: int):
rng = random.Random(f"{seed}:split")
by_sub = {}
for r in rows:
by_sub.setdefault(r["subtype"], []).append(r)
for rs in by_sub.values():
rng.shuffle(rs)
k = max(1, round(len(rs) * VAL_FRACTION))
for j, r in enumerate(rs):
r["split"] = "validation" if j < k else "test"
rows = [r for rs in by_sub.values() for r in rs]
rng.shuffle(rows)
for j, r in enumerate(rows):
r["id"] = f"tfb-{j + 1:04d}"
return rows
COLUMNS = ["id", "domain", "subtype", "fault_type", "next_action", "acceptable_next_actions", "critical_next_actions",
"difficulty", "is_hard_negative", "current_time", "user_request", "tools", "messages", "num_tool_calls",
"last_tool_name", "last_tool_result", "gold_recovery", "rationale", "meta"]
def main():
ap = argparse.ArgumentParser()
ap.add_argument("--out", default="data")
ap.add_argument("--seed", type=int, default=20260923)
a = ap.parse_args()
rows = split_and_id(generate(a.seed), a.seed)
os.makedirs(a.out, exist_ok=True)
df = pd.DataFrame(rows)
for split in ("test", "validation"):
part = df[df.split == split][COLUMNS].sort_values("id").reset_index(drop=True)
path = os.path.join(a.out, f"{split}-00000-of-00001.parquet")
part.to_parquet(path, index=False)
print(f"{split}: {len(part)} rows -> {path}")
if __name__ == "__main__":
main()
|