mirror of
https://github.com/open-thought/reasoning-gym.git
synced 2026-04-25 17:10:51 +00:00
Create python generator files for gsm symbolic templates
This commit is contained in:
parent
ff13dc6825
commit
1e0dbc9875
100 changed files with 6919 additions and 0 deletions
53
reasoning_gym/arithmetic/gsm_symbolic/generator_59.py
Normal file
53
reasoning_gym/arithmetic/gsm_symbolic/generator_59.py
Normal file
|
|
@ -0,0 +1,53 @@
|
|||
from random import Random
|
||||
from typing import Dict, Any
|
||||
|
||||
def generate_from_variables(time_per_interval: int, distance_per_interval: int, total_distance: int) -> Dict[str, Any]:
|
||||
intervals = total_distance // distance_per_interval
|
||||
total_time = intervals * time_per_interval
|
||||
|
||||
question = f"A fog bank rolls in from the ocean to cover a city. It takes {time_per_interval} minutes to cover every {distance_per_interval} miles of the city. If the city is {total_distance} miles across from the oceanfront to the opposite inland edge, how many minutes will it take for the fog bank to cover the whole city?"
|
||||
|
||||
answer_cot = f"The city will be covered in {total_distance} / {distance_per_interval} = {intervals} intervals of {time_per_interval} minutes.\nThus, it will take {intervals} * {time_per_interval} = {total_time} minutes for the fog to cover the whole city.\n#### {total_time}"
|
||||
|
||||
return {
|
||||
'question': question,
|
||||
'answer': str(total_time),
|
||||
'answer_cot': answer_cot,
|
||||
'answer_value': total_time,
|
||||
'variables': {
|
||||
'time_per_interval': time_per_interval,
|
||||
'distance_per_interval': distance_per_interval,
|
||||
'total_distance': total_distance,
|
||||
'intervals': intervals
|
||||
}
|
||||
}
|
||||
|
||||
def generate_example(rng: Random, difficulty: float = 1.0) -> Dict[str, Any]:
|
||||
# Generate random values scaled by difficulty
|
||||
distance_per_interval = int(rng.randint(2, int(100 * difficulty)))
|
||||
time_per_interval = int(rng.randint(2, int(500 * difficulty)))
|
||||
|
||||
# Ensure total distance is divisible by distance_per_interval
|
||||
num_intervals = rng.randint(2, int(20 * difficulty))
|
||||
total_distance = distance_per_interval * num_intervals
|
||||
|
||||
# Ensure total_distance is in valid range
|
||||
while total_distance > 100:
|
||||
num_intervals = rng.randint(2, int(20 * difficulty))
|
||||
total_distance = distance_per_interval * num_intervals
|
||||
|
||||
result = generate_from_variables(time_per_interval, distance_per_interval, total_distance)
|
||||
|
||||
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(10, 3, 42)
|
||||
Loading…
Add table
Add a link
Reference in a new issue