Implementation:Princeton nlp Tree of thought llm Get Task

Knowledge Sources	tree-of-thought-llm
Domains	Software_Design, NLP
Last Updated	2026-02-14 03:30 GMT

Overview

Concrete tool for instantiating benchmark task objects by name provided by the Tree of Thoughts framework.

Description

The get_task function is a factory that maps a string task name to the corresponding task class and returns a fully initialized instance. It uses lazy imports to avoid loading all task dependencies upfront.

Usage

Import and call this function at the start of any experiment run to convert the user-selected task name from CLI arguments into a usable task object. Called in run.py as the first step of the run() function.

Code Reference

Source Location

• Repository: tree-of-thought-llm
• File: src/tot/tasks/__init__.py
• Lines: 1-12

Signature

def get_task(name):
    """
    Factory function that returns an instantiated Task object.

    Args:
        name (str): Task identifier. One of 'game24', 'text', 'crosswords'.

    Returns:
        Task: Instantiated task object (Game24Task, TextTask, or MiniCrosswordsTask).

    Raises:
        NotImplementedError: If name is not recognized.
    """

Import

from tot.tasks import get_task

I/O Contract

Inputs

Name	Type	Required	Description
name	str	Yes	Task identifier: 'game24', 'text', or 'crosswords'

Outputs

Name	Type	Description
return	Task	Instantiated task with loaded data, self.steps, self.stops, prompt wrap methods, and test_output()

Usage Examples

Instantiating a Task

from tot.tasks import get_task

# Instantiate Game of 24 task
task = get_task('game24')
print(len(task))       # Number of puzzles in dataset
print(task.steps)      # 4 (BFS depth levels)
print(task.stops)      # ['\n', '\n', '\n', '\n']

# Get input for a specific puzzle
x = task.get_input(900)  # e.g., "1 2 3 4"

# Instantiate Creative Writing task
text_task = get_task('text')
print(text_task.steps)  # 2

# Instantiate Crosswords task
xword_task = get_task('crosswords')

Related Pages

Implements Principle

• Principle:Princeton_nlp_Tree_of_thought_llm_Task_Instantiation

Requires Environment

• Environment:Princeton_nlp_Tree_of_thought_llm_Python_OpenAI