File size: 4,189 Bytes
64b0d5e | 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 | from __future__ import annotations
import argparse
import json
import random
from pathlib import Path
SYSTEM_PROMPT = (
"Solve exact-answer reasoning problems. "
"Inside the think block, reason in pseudocode only. "
"Do not use English prose inside the think block. "
"After thinking, end with exactly one line formatted as ANSWER: <answer>."
)
USER_INSTRUCTION = (
"Inside the think block, reason in pseudocode only. "
"Do not use English prose inside the think block. "
"Prefer lines like `x = 51`, `x = x + 34`, `state -> x = 85`."
)
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser()
parser.add_argument("--output-dir", required=True)
parser.add_argument("--train", type=int, default=2000)
parser.add_argument("--valid", type=int, default=200)
parser.add_argument("--test", type=int, default=200)
parser.add_argument("--seed", type=int, default=20260406)
return parser.parse_args()
def make_arithmetic_task(rng: random.Random, task_id: str) -> dict[str, str]:
a = rng.randint(14, 49)
b = rng.randint(5, 17)
c = rng.randint(3, 11)
d = rng.randint(2, 9)
e = rng.randint(6, 15)
start = rng.randint(20, 80)
s1 = start + a
s2 = s1 + b
s3 = s2 * c
sub = d * e
answer = s3 - sub
prompt = (
f"Start with {start}. Add {a}. Add {b}. Multiply the result by {c}. "
f"Subtract {d} times {e}. What number do you get?"
)
think = "\n".join(
[
f"x = {start}",
f"x = x + {a} -> {s1}",
f"x = x + {b} -> {s2}",
f"x = x * {c} -> {s3}",
f"y = {d} * {e} -> {sub}",
f"x = x - y -> {answer}",
]
)
return {"id": task_id, "kind": "arithmetic", "prompt": prompt, "answer": str(answer), "think": think}
def make_state_task(rng: random.Random, task_id: str) -> dict[str, str]:
base_x = rng.randint(2, 8)
base_y = rng.randint(3, 9)
base_z = rng.randint(4, 10)
x = base_x
y = base_y
z = base_z
steps: list[str] = [
f"x = {base_x}",
f"y = {base_y}",
f"z = {base_z}",
]
x = x + y
steps.append(f"x = x + y -> {x}")
y = y * 2
steps.append(f"y = y * 2 -> {y}")
z = z + x - 1
steps.append(f"z = z + x - 1 -> {z}")
x = x * z
steps.append(f"x = x * z -> {x}")
y = y + z
steps.append(f"y = y + z -> {y}")
answer = x - y
steps.append(f"result = x - y -> {answer}")
prompt = (
"Run this exact program and report the final value of x - y.\n\n"
f"x = {base_x}\n"
f"y = {base_y}\n"
f"z = {base_z}\n"
"x = x + y\n"
"y = y * 2\n"
"z = z + x - 1\n"
"x = x * z\n"
"y = y + z"
)
think = "\n".join(steps)
return {"id": task_id, "kind": "state", "prompt": prompt, "answer": str(answer), "think": think}
def make_task(rng: random.Random, task_id: str) -> dict[str, str]:
if rng.random() < 0.5:
return make_arithmetic_task(rng, task_id)
return make_state_task(rng, task_id)
def make_record(task: dict[str, str]) -> dict:
assistant = f"<think>\n{task['think']}\n</think>\n\nANSWER: {task['answer']}"
user = f"{USER_INSTRUCTION}\n\nProblem:\n{task['prompt']}"
return {
"messages": [
{"role": "system", "content": SYSTEM_PROMPT},
{"role": "user", "content": user},
{"role": "assistant", "content": assistant},
]
}
def write_split(path: Path, count: int, seed: int) -> None:
rng = random.Random(seed)
with path.open("w", encoding="utf-8") as handle:
for index in range(count):
task = make_task(rng, f"{path.stem}-{index:05d}")
handle.write(json.dumps(make_record(task)) + "\n")
def main() -> None:
args = parse_args()
output_dir = Path(args.output_dir)
output_dir.mkdir(parents=True, exist_ok=False)
write_split(output_dir / "train.jsonl", args.train, args.seed)
write_split(output_dir / "valid.jsonl", args.valid, args.seed + 1)
write_split(output_dir / "test.jsonl", args.test, args.seed + 2)
if __name__ == "__main__":
main()
|