SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.02330 · 2024

Enhance Reasoning for Large Language Models in the Game Werewolf

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 3 violated it and 0 have no contract verdict; 0 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2402.02330.json · badge: /api/ran/2402.02330.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
boluoweifenda/werewolf canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
check_audio Ran
ran, violated the contract we drafted
boluoweifenda/werewolf/processor/check_data_en.py
code served (permissive licence) · get_code("a7636863acf66050")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

# Construct valid arguments for check_audio function

game_id = "game_001"

data = {
    'audio': {
        'task[1]': {
            'text': 'Hello world',
            'duration': 3.5
        },
        'task[2]': {
            'text': 'Good morning',
            'duration': 4.2
        },
        'task[3]': {
            'text': 'How are you',
            'duration': 2.8
        }
    },
    'game_state': {
        'Automatically Passed for Not Speaking': []
    }
}

ARGS = {
    'game_id': game_id,
    'data': data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
check_game_state Ran
ran, violated the contract we drafted
boluoweifenda/werewolf/processor/check_data_en.py
code served (permissive licence) · get_code("e1fbd7805b53b2d7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

def check_game_state(game_id, data):
    game_state = data['game_state']
    task_times = list(data['task_time'].keys())
    if 'Day 1 Night' not in game_state.keys():
        print('Incomplete replay information: ', game_id)
        return False
    is_complete = False
    for i in range(len(task_times)):
        task_time = task_times[i]
        if 'Day 1 Night' in task_time or 'Day 1 Daytime - Daybreak has come' in task_time:
            is_complete = True
            break
        elif i == 0 and 'Day 1 Daytime' in task_time and ('is preparing to speak' in task_time):
            if 'Voting Pattern (Round 2)' not in game_state['Day 1 Daytime'].keys():
                is_complete = True
            else:
                exile_id = game_state['Day 1 Daytime']['Voting Result']
                if 'Day 1 Daytime-[%d] is speaking(Round 2)' % exile_id in task_time:
                    is_complete = True
            break
    return is_complete


# Construct valid arguments for check_game_state
game_id = "game_001"

data = {
    'game_state': {
        'Day 1 Night': {
            'action': 'night_phase'
        },
        'Day 1 Daytime': {
            'Voting Pattern (Round 2)': {'player_1': 1, 'player_2': 2},
            'Voting Result': 5
        }
    },
    'task_time': {
        'Day 1 Daytime-[5] is speaking(Round 2)': 100,
        'Day 1 Night': 200,
        'Day 1 Daytime - Daybreak has come': 300
    }
}

ARGS = {
    'game_id': game_id,
    'data': data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
check_label Ran
ran, violated the contract we drafted
boluoweifenda/werewolf/processor/check_data_en.py
code served (permissive licence) · get_code("8233dbc8233c28b8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

# Construct valid arguments for check_label function
game_id = "game_001"

data = {
    'audio': {
        'task_1': {
            'text': 'This is some audio text',
            'summary': 'Summary of audio 1',
            'format': 'wav'
        },
        'task_2': {
            'text': 'Another audio text',
            'summary': 'Summary of audio 2',
            'format': 'mp3'
        }
    }
}

ARGS = {
    'game_id': game_id,
    'data': data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

This paper presents an innovative framework that integrates Large Language Models (LLMs) with an external Thinker module to enhance the reasoning capabilities of LLM-based agents. Unlike augmenting LLMs with prompt engineering, Thinker directly harnesses knowledge from databases and employs various optimization techniques. The framework forms a reasoning hierarchy where LLMs handle intuitive System-1 tasks such as natural language processing, while the Thinker focuses on cognitive System-2 tasks that require complex logical analysis and domain-specific knowledge. Our framework is presented using a 9-player Werewolf game that demands dual-system reasoning. We introduce a communication protocol between LLMs and the Thinker, and train the Thinker using data from 18800 human sessions and reinforcement learning. Experiments demonstrate the framework's effectiveness in deductive reasoning, speech generation, and online game evaluation. Additionally, we fine-tune a 6B LLM to surpass GPT4 when integrated with the Thinker. This paper also contributes the largest dataset for social deduction games to date.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2402.02330")
get_code_for_paper("2402.02330")
have("2402.02330")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2402.02330.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2402.02330.svg)](https://syntology.ai/paper/2402.02330)

Connect an agent — have() is free.