SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2004.05160 · EMNLP Findings · 2020

On the Language Neutrality of Pre-trained Multilingual Representations

Alexander Fraser, Jindřich Libovický, Rudolf Rosa

arXiv · PDF · Open in the Atlas

Code that ran

4 of 6 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; 3 failures of our instrument; 2 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/2004.05160.json · badge: /api/ran/2004.05160.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
jlibovicky/assess-multilingual-bert marked official in Papers with Code
Papers with Code archive
4 ran · 2 no recorded run
FunctionStatusWhere it lives
batch_generator Ran
ran; our contract draft was wrong, not the code
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("49169613387052ef")
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 generator. The image it ran in was not recorded. If your copy returns the same type 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

torch.manual_seed(0)

class SimpleTokenizer:
    def __init__(self, pad_token_id=0):
        self.pad_token_id = pad_token_id

def simple_generator():
    """Generator that yields token sequences of varying lengths"""
    yield torch.tensor([1, 2, 3], dtype=torch.int64)
    yield torch.tensor([4, 5], dtype=torch.int64)
    yield torch.tensor([6, 7, 8, 9], dtype=torch.int64)

tokenizer = SimpleTokenizer(pad_token_id=0)

ARGS = {
    "generator": simple_generator(),
    "size": 2,
    "tokenizer": tokenizer,
    "padding": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type generator — a generator output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a generator output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
load_bert Ran
ran; our contract draft was wrong, not the code
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("e4b951d62725e6fc")
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. The image it ran in was not recorded. If your copy returns the same type 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 os
from unittest.mock import MagicMock

torch.manual_seed(0)

# Create a mock tokenizer
mock_tokenizer = MagicMock()
mock_tokenizer.from_pretrained = MagicMock(return_value=MagicMock())

# Create a mock model with the necessary attributes
mock_model = MagicMock()
mock_model.config = MagicMock()
mock_model.config.hidden_size = 768
mock_model.embeddings = MagicMock()
mock_model.embeddings.word_embeddings = MagicMock()
mock_model.embeddings.word_embeddings.weight = torch.randn(30522, 768)
mock_model.eval = MagicMock(return_value=None)
mock_model.to = MagicMock(return_value=mock_model)

# Mock AutoModel and AutoTokenizer
import transformers
original_auto_model = transformers.AutoModel
original_auto_tokenizer = transformers.AutoTokenizer

transformers.AutoModel.from_pretrained = MagicMock(return_value=mock_model)
transformers.AutoTokenizer.from_pretrained = MagicMock(return_value=mock_tokenizer)

ARGS = {
    'bert_spec': 'bert-base-uncased',
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
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
pad_sentences Ran
ran, honoured the contract we drafted
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("58b8bef26d1b590e")
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 Tensor, ndim 2, dtype family int, shape (2, 3) 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

torch.manual_seed(0)

class SimpleTokenizer:
    def __init__(self, pad_token_id=0):
        self.pad_token_id = pad_token_id

sentences = [
    torch.tensor([1, 2, 3], dtype=torch.int64),
    torch.tensor([4, 5], dtype=torch.int64),
]

tokenizer = SimpleTokenizer(pad_token_id=0)

ARGS = {
    "sentences": sentences,
    "tokenizer": tokenizer,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family int, shape (2, 3) — shape (2, 3) 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
text_data_generator Ran
ran; our contract draft was wrong, not the code
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("2974bc2df4e4d420")
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 generator. The image it ran in was not recorded. If your copy returns the same type 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 tempfile
import os

class MockTokenizer:
    def encode(self, text, max_length=510):
        # Simple mock: return a list of token ids
        tokens = [ord(c) % 100 for c in text[:max_length]]
        return tokens[:max_length]

# Create a temporary file with sample text
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, encoding='utf-8', suffix='.txt')
temp_file.write("This is the first sentence.\n")
temp_file.write("This is the second sentence.\n")
temp_file.write("This is the third sentence.\n")
temp_file.close()

ARGS = {
    'path': temp_file.name,
    'tokenizer': MockTokenizer(),
    'epochs': 2,
    'max_len': 510
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type generator — a generator output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a generator output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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_repr_from_layer Not yet run
unverified (harvested, no recorded run)
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("6516b2ef42579edb")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
jlibovicky/assess-multilingual-bert/save_centroids.py
code served (permissive licence) · get_code("f141e5aaf7d9dd5b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 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 6 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

Multilingual contextual embeddings, such as multilingual BERT and XLM-RoBERTa, have proved useful for many multi-lingual tasks. Previous work probed the cross-linguality of the representations indirectly using zero-shot transfer learning on morphological and syntactic tasks. We instead investigate the languageneutrality of multilingual contextual embeddings directly and with respect to lexical semantics. Our results show that contextual embeddings are more language-neutral and, in general, more informative than aligned static word-type embeddings, which are explicitly trained for language neutrality. Contextual embeddings are still only moderately languageneutral by default, so we propose two simple methods for achieving stronger language neutrality: first, by unsupervised centering of the representation for each language and second, by fitting an explicit projection on small parallel data. Besides, we show how to reach stateof-the-art accuracy on language identification and match the performance of statistical methods for word alignment of parallel sentences without using parallel data.

For agents

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

get_harvested_code_for_paper("2004.05160")
get_code_for_paper("2004.05160")
have("2004.05160")

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

curl https://syntology.ai/api/ran/2004.05160.json

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

[![Syntology run record](https://syntology.ai/api/ran/2004.05160.svg)](https://syntology.ai/paper/2004.05160)

Connect an agent — have() is free.