reasoning-gym/reasoning_gym/algebra/simple_equations.py
Adefioye 5bfe730c3b
Simple equations curriculum (#340)
* Add curriculum to simple_equations and unit tests
* Add difficulty to metadata
2025-03-13 21:17:14 +01:00

184 lines
7 KiB
Python

import random
import string
from dataclasses import dataclass, field
from typing import Optional
from sympy import Symbol
from ..coaching import AttributeType, BaseCurriculum, ScalarAttributeDefinition
from ..factory import ProceduralDataset, register_dataset
@dataclass
class SimpleEquationsConfig:
"""Configuration for simple equation task generation"""
min_terms: int = 2 # Minimum number of terms in expression
max_terms: int = 4 # Maximum number of terms
min_value: int = 1 # Minimum value for constants
max_value: int = 100 # Maximum value for constants
operators: tuple = ("+", "-", "*") # Allowed operators
operators_weights: list[float] = field(default_factory=lambda: [0.4, 0.4, 0.2]) # Weights for each operator
seed: Optional[int] = None
size: int = 500
def validate(self) -> None:
"""Validate configuration parameters"""
assert self.min_terms > 0, "min_terms must be positive"
assert self.max_terms >= self.min_terms, "max_terms must be >= min_terms"
assert self.min_value > 0, "min_value must be positive"
assert self.max_value >= self.min_value, "max_value must be >= min_value"
assert len(self.operators) > 0, "must specify at least one operator"
assert all(op in ("+", "-", "*") for op in self.operators), "invalid operator specified"
assert round(sum(self.operators_weights), 1) == 1.0, "operators_weights must sum to 1.0"
class SimpleEquationsDataset(ProceduralDataset):
"""Generates simple equations with one variable to solve"""
def __init__(self, config: SimpleEquationsConfig):
self._prompt_templates = [
"Find the value of {variable} in the equation: {equation}",
"Solve for {variable}: {equation}",
"Determine the value of {variable} that satisfies: {equation}",
]
super().__init__(config=config, seed=config.seed, size=config.size)
def __getitem__(self, idx: int) -> dict:
"""Generate a single equation task
Returns:
dict with keys:
- question: str, the equation to solve (e.g. "3 * x = 12")
- answer: str, the solution value (e.g. "4")
- metadata: dict with generation parameters
"""
rng = random.Random(self.seed + idx)
# Get variable and generate equation
variable = self._get_variable(rng)
equation, solution = self._generate_equation(rng, variable)
return {
"question": rng.choice(self._prompt_templates).format(variable=variable, equation=equation),
"answer": str(solution),
"metadata": {
"equation": equation,
"variable": variable,
"difficulty": {
"min_terms": self.config.min_terms,
"max_terms": self.config.max_terms,
"min_value": self.config.min_value,
"max_value": self.config.max_value,
"operators_weights": self.config.operators_weights,
},
},
}
def _get_variable(self, rng: random.Random) -> str:
"""Get a random lowercase variable name"""
return rng.choice(string.ascii_lowercase)
def _generate_equation(self, rng: random.Random, variable: str) -> tuple[str, int]:
"""Generate an equation and its solution
Args:
rng: Random number generator
variable: Variable symbol to use in equation
Returns:
Tuple of (equation string, solution integer)
"""
x = Symbol(variable)
# Generate terms for left side
num_terms = rng.randint(self.config.min_terms, self.config.max_terms)
terms = []
# Generate all constant terms first
for _ in range(num_terms):
value = rng.randint(self.config.min_value, self.config.max_value)
terms.append(value)
# Replace one random term with the variable term
var_pos = rng.randint(0, num_terms - 1)
coef = rng.randint(self.config.min_value, self.config.max_value)
if "*" in self.config.operators:
terms[var_pos] = coef * x
else:
terms[var_pos] = x
# Apply operators between terms
expr = terms[0]
for i in range(1, num_terms):
op = rng.choices(self.config.operators, weights=self.config.operators_weights, k=1)[0]
if op == "+":
expr = expr + terms[i]
elif op == "-":
expr = expr - terms[i]
else: # '*'
expr = expr * terms[i]
left_side = expr
solution_value = rng.randint(self.config.min_value, self.config.max_value)
right_side = left_side.subs(x, solution_value)
return f"{left_side} = {right_side}", solution_value
class SimpleEquationsCurriculum(BaseCurriculum):
"""Curriculum for simple equations task"""
def __init__(self):
super().__init__(SimpleEquationsCurriculum.__name__, SimpleEquationsConfig)
# Define attributes
self._define_attributes(
ScalarAttributeDefinition(
name="min_terms",
field_name="min_terms",
levels=[2, 3, 4, 5],
default_level=0,
description="Minimum number of terms in simple equations",
attr_type=AttributeType.STATIC,
min_value=-2,
),
ScalarAttributeDefinition(
name="max_terms",
field_name="max_terms",
levels=[5, 10, 15, 20],
default_level=0,
description="Maximum number of terms in simple equations",
attr_type=AttributeType.STATIC,
min_value=5,
),
ScalarAttributeDefinition(
name="min_value",
field_name="min_value",
levels=[1, 10, 100, 1000],
default_level=0,
description="Minimum value for constants in simple equations",
attr_type=AttributeType.STATIC,
min_value=1,
),
ScalarAttributeDefinition(
name="max_value",
field_name="max_value",
levels=[100, 10000, 1000000, 100000000],
default_level=0,
description="Maximum value for constants in simple equations",
attr_type=AttributeType.STATIC,
min_value=100,
),
ScalarAttributeDefinition(
name="operators_weights",
field_name="operators_weights",
levels=[[0.4, 0.4, 0.2], [0.35, 0.35, 0.3], [0.3, 0.3, 0.4], [0.2, 0.2, 0.6]],
default_level=0,
description="Weights for each operator in simple equations",
attr_type=AttributeType.STATIC,
min_value=[0.4, 0.4, 0.2],
),
)
register_dataset("simple_equations", SimpleEquationsDataset, SimpleEquationsConfig, SimpleEquationsCurriculum)