SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2308.07201 · 2023

ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate

arXiv · PDF · Open in the Atlas

Code that ran

2 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 1 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/2308.07201.json · badge: /api/ran/2308.07201.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
chanchimin/chateval canonical (classifier label)
classifier basis: census
2 ran · 1 no recorded run
FunctionStatusWhere it lives
get_accuracy_calibration Ran
ran; our contract draft was wrong, not the code
chanchimin/chateval/calc_adversarial_results_calc_score_calibration.py
code served (permissive licence) · get_code("24114c47bda02e3c")
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, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

# Construct valid arguments for get_accuracy_calibration

# Create human_results - list of dicts with 'human_results' key
human_results = [
    {'human_results': 'gpt35'},
    {'human_results': 'vicuna'},
    {'human_results': 'tie'},
    {'human_results': 'gpt35'},
    {'human_results': 'vicuna'},
]

# Create model_results - list of dicts with 'evaluation' key containing list of role dicts
# Each role dict has 'evaluation' key with string containing scores
model_results = [
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 8.5\nThe score of Assistant 2: 7.2'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 6.0\nThe score of Assistant 2: 9.1'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 7.5\nThe score of Assistant 2: 7.5'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 8.0\nThe score of Assistant 2: 6.5'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 5.0\nThe score of Assistant 2: 8.5'
            }
        ]
    },
]

# Create model_results_reverse - same structure as model_results
model_results_reverse = [
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 7.2\nThe score of Assistant 2: 8.5'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 9.1\nThe score of Assistant 2: 6.0'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 7.5\nThe score of Assistant 2: 7.5'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 6.5\nThe score of Assistant 2: 8.0'
            }
        ]
    },
    {
        'evaluation': [
            {
                'evaluation': 'Some evaluation text\nThe score of Assistant 1: 8.5\nThe score of Assistant 2: 5.0'
            }
        ]
    },
]

ARGS = {
    'human_results': human_results,
    'model_results': model_results,
    'model_results_reverse': model_results_reverse,
    'method': 'single',
    'example_nums': 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
get_kappa Ran
ran, honoured the contract we drafted
chanchimin/chateval/calc_adversarial_results_calc_score_calibration.py
code served (permissive licence) · get_code("f22cc7f6755823a6")
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, through the batch API) and returned float64, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

# Construct valid arguments for get_kappa
# The function expects:
# - meta: dict with 'targets' and 'predictions' keys
# - targets: list of strings from {'gpt35', 'vicuna', 'tie'}
# - predictions: list of strings from {'gpt35', 'vicuna', 'tie'}

meta = {
    'targets': ['gpt35', 'vicuna', 'tie', 'gpt35', 'vicuna'],
    'predictions': ['gpt35', 'gpt35', 'tie', 'vicuna', 'vicuna']
}

ARGS = {
    'meta': meta
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
get_conv_template Not yet run
unverified (harvested, no recorded run)
chanchimin/chateval/FastChat/fastchat/conversation.py
code served (permissive licence) · get_code("a619c795784c4043")
How it ran

no invocation recorded.

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, 3 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 3 later) and 0 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.

FunctionAlso present inEarliest-datedOthers named
get_conv_template 69 other papers
4 dated earlier, 1 the same month, 63 later, 1 undated
1911.03705 (2019-11)
CommonGen: A Constrained Text Generation Challenge for Generative Commonsense Reasoning
2112.05682 (2021-12)
Self-attention Does Not Need $O(n^2)$ Memory
2204.07705 (2022-04)
SUPER-NATURALINSTRUCTIONS: Generalization via Declarative Instructions on 1600+ NLP Tasks
and 66 more not named here
get_accuracy_calibration 1 other paper
1 later
2505.20451 (2025-05)
AMULET: Putting Complex Multi-Turn Conversations on the Stand with LLM Juries
get_kappa 1 other paper
1 later
2505.20451 (2025-05)
AMULET: Putting Complex Multi-Turn Conversations on the Stand with LLM Juries

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

Text evaluation has historically posed significant challenges, often demanding substantial labor and time cost. With the emergence of large language models (LLMs), researchers have explored LLMs' potential as alternatives for human evaluation. While these single-agent-based approaches show promise, experimental results suggest that further advancements are needed to bridge the gap between their current effectiveness and human-level evaluation quality. Recognizing that best practices of human evaluation processes often involve multiple human annotators collaborating in the evaluation, we resort to a multi-agent debate framework, moving beyond single-agent prompting strategies. The multi-agent-based approach enables a group of LLMs to synergize with an array of intelligent counterparts, harnessing their distinct capabilities and expertise to enhance efficiency and effectiveness in handling intricate tasks. In this paper, we construct a multi-agent referee team called ChatEval to autonomously discuss and evaluate the quality of generated responses from different models on open-ended questions and traditional natural language generation (NLG) tasks. Our analysis shows that ChatEval transcends mere textual scoring, offering a human-mimicking evaluation process for reliable assessments. Our code is available at https://github.com/chanchimin/ChatEval.

For agents

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

get_harvested_code_for_paper("2308.07201")
get_code_for_paper("2308.07201")
have("2308.07201")

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

curl https://syntology.ai/api/ran/2308.07201.json

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

[![Syntology run record](https://syntology.ai/api/ran/2308.07201.svg)](https://syntology.ai/paper/2308.07201)

Connect an agent — have() is free.