SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2312.02730 · NeurIPS · 2023

Towards Measuring Representational Similarity of Large Language Models

Max Mehdi, Ben Amor, Michael Granitzer, Florian Lemmerich

arXiv · PDF · Open in the Atlas

Code that ran

2 of 2 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, 1 violated it and 0 have no contract verdict; 1 failure of our instrument; 0 no recorded run), measured 2026-11-22; 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/2312.02730.json · badge: /api/ran/2312.02730.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
mklabunde/llm_repsim canonical (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
filter_combinations Ran
ran; our contract draft was wrong, not the code
mklabunde/llm_repsim/compare_representations.py
code served (permissive licence) · get_code("8f4e943371e38540")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned list, ndim 2, 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.

from pathlib import Path
from typing import List, Tuple

# Create sample Path combinations
combinations = [
    (Path("/data/train/model_v1/checkpoint.pt"), Path("/data/train/model_v1/config.yaml")),
    (Path("/data/test/model_v2/checkpoint.pt"), Path("/data/test/model_v2/config.yaml")),
    (Path("/data/train/model_v2/weights.pt"), Path("/data/train/model_v2/settings.yaml")),
    (Path("/data/validation/model_v1/checkpoint.pt"), Path("/data/validation/model_v1/config.yaml")),
    (Path("/data/train/baseline/checkpoint.pt"), Path("/data/train/baseline/config.yaml")),
]

# Filter criteria
must_contain_all = ["train"]
must_contain_any = ["model_v1", "model_v2"]
must_not_contain = ["baseline"]
one_must_contain = ["checkpoint", "config"]

ARGS = {
    "combinations": combinations,
    "must_contain_all": must_contain_all,
    "must_contain_any": must_contain_any,
    "must_not_contain": must_not_contain,
    "one_must_contain": one_must_contain,
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family float — a list 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
pair_should_be_compared Ran
ran, violated the contract we drafted
mklabunde/llm_repsim/compare_representations.py
code served (permissive licence) · get_code("3e6d2c3a2e2ebf40")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 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.

from pathlib import Path
import tempfile
import os

# Create a temporary directory for testing
temp_dir = tempfile.mkdtemp()

# Create two info dictionaries with same dataset and split, but different models
info1 = {
    'dataset': 'cifar10',
    'split': 'test',
    'model': 'resnet18'
}

info2 = {
    'dataset': 'cifar10',
    'split': 'test',
    'model': 'vgg16'
}

# Create a path that doesn't exist
pair_results_path = Path(temp_dir) / "pair_results.json"

# Set recompute to True
recompute = True

ARGS = {
    'info1': info1,
    'info2': info2,
    'pair_results_path': pair_results_path,
    'recompute': recompute
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
none outside the standard library — 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 2 samples lifted from this paper's linked repositories, 2 carry a function identity and 0 do not. Of the 2 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 2 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

Understanding the similarity of the numerous released large language models (LLMs) has many uses, e.g., simplifying model selection, detecting illegal model reuse, and advancing our understanding of what makes LLMs perform well. In this work, we measure the similarity of representations of a set of LLMs with 7B parameters. Our results suggest that some LLMs are substantially different from others. We identify challenges of using representational similarity measures that suggest the need of careful study of similarity scores to avoid false conclusions.

For agents

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

get_harvested_code_for_paper("2312.02730")
get_code_for_paper("2312.02730")
have("2312.02730")

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

curl https://syntology.ai/api/ran/2312.02730.json

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

[![Syntology run record](https://syntology.ai/api/ran/2312.02730.svg)](https://syntology.ai/paper/2312.02730)

Connect an agent — have() is free.