mirror of
https://github.com/open-thought/reasoning-gym.git
synced 2026-04-25 17:10:51 +00:00
68 lines
3 KiB
Python
68 lines
3 KiB
Python
from random import Random
|
|
from typing import Dict, Any
|
|
import math
|
|
|
|
def generate_from_variables(name: str, food: str, peel_rate: int, batch_size: int,
|
|
time_per_batch: int, total_amount: int) -> Dict[str, Any]:
|
|
|
|
peel_time = total_amount // peel_rate
|
|
num_batches = total_amount // batch_size
|
|
cook_time = num_batches * time_per_batch
|
|
total_time = peel_time + cook_time
|
|
|
|
question = f"{name} can peel {peel_rate} {food}s a minute and saute {batch_size} {food}s in {time_per_batch} minutes. How long will it take her to peel and saute {total_amount} {food}s?"
|
|
|
|
answer_cot = f"First find how long it takes {name} to peel the {food}: {total_amount} {food} / {peel_rate} {food}/minute = {peel_time} minutes\n" \
|
|
f"Then find how many batches of {food} she needs to cook: {total_amount} {food} / {batch_size} {food}/batch = {num_batches} batches\n" \
|
|
f"Then multiply the number of batches by the time per batch to find the total cook time: {num_batches} batches * {time_per_batch} minutes/batch = {cook_time} minutes\n" \
|
|
f"Then add the peeling time to find the total time {name} spends: {cook_time} minutes + {peel_time} minutes = {total_time} minutes\n" \
|
|
f"#### {total_time}"
|
|
|
|
return {
|
|
'question': question,
|
|
'answer': str(total_time),
|
|
'answer_cot': answer_cot,
|
|
'answer_value': total_time,
|
|
'variables': {
|
|
'name': name,
|
|
'food': food,
|
|
'peel_rate': peel_rate,
|
|
'batch_size': batch_size,
|
|
'time_per_batch': time_per_batch,
|
|
'total_amount': total_amount,
|
|
'peel_time': peel_time,
|
|
'cook_time': cook_time
|
|
}
|
|
}
|
|
|
|
def generate_example(rng: Random, difficulty: float = 1.0) -> Dict[str, Any]:
|
|
names_female = ["Emily", "Sarah", "Emma", "Sophia", "Olivia", "Ava", "Isabella", "Mia"]
|
|
foods = ["shrimp", "onion", "carrot", "mushroom", "clam"]
|
|
|
|
name = rng.choice(names_female)
|
|
food = rng.choice(foods)
|
|
|
|
peel_rate = int(rng.randint(4, int(15 * difficulty)))
|
|
batch_size = int(rng.randrange(20, int(50 * difficulty), 5))
|
|
time_per_batch = int(rng.randint(5, int(20 * difficulty)))
|
|
|
|
# Ensure total is divisible by both peel_rate and batch_size
|
|
lcm = peel_rate * batch_size // math.gcd(peel_rate, batch_size)
|
|
num_lcm = rng.randint(1, int(4 * difficulty))
|
|
total_amount = lcm * num_lcm
|
|
|
|
result = generate_from_variables(name, food, peel_rate, batch_size, time_per_batch, total_amount)
|
|
|
|
return {
|
|
'question': result['question'],
|
|
'answer': result['answer'],
|
|
'metadata': {
|
|
'difficulty': difficulty,
|
|
'answer_value': result['answer_value'],
|
|
'answer_cot': result['answer_cot'],
|
|
'variables': result['variables']
|
|
}
|
|
}
|
|
|
|
def original_example() -> Dict[str, Any]:
|
|
return generate_from_variables("Emily", "shrimp", 6, 30, 10, 90)
|