SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2401.10440 · EMNLP · 2024

Breaking the Curse of Multilinguality with Cross-lingual Expert Language Models

Luke Zettlemoyer, Noah Smith, Margaret Li, Suchin Gururangan, Tomasz Limisiewicz, Hila Gonen, Terra Blevins

arXiv · PDF · Open in the Atlas

Code that ran

6 of 10 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, 0 violated it and 2 have no contract verdict; 4 failures of our instrument; 4 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/2401.10440.json · badge: /api/ran/2401.10440.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
blvns/x-elm canonical (classifier label)
classifier basis: census
6 ran · 4 no recorded run
FunctionStatusWhere it lives
average Ran
ran; our contract draft was wrong, not the code
blvns/x-elm/metaseq/sequence_scorer_btm.py
code served (permissive licence) · get_code("8060f2ec5c0f9539")
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 SimpleModel. 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 torch.nn as nn
from copy import deepcopy

torch.manual_seed(0)

# Create small test models
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.bias = nn.Parameter(torch.randn(2))
    
    def forward(self, x):
        return self.linear(x)

# Instantiate multiple models
model1 = SimpleModel()
model2 = SimpleModel()
model3 = SimpleModel()

models = [model1, model2, model3]

# Create weights for averaging
weights = [0.5, 0.3, 0.2]

ARGS = {
    "models": models,
    "weights": weights
}
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 SimpleModel — a SimpleModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModel 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
find_offsets Ran
ran on a synthesized input
blvns/x-elm/metaseq/file_chunker_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from leililab/enzygen, is pointer only (licence: NONE) · get_code("840d9070e4fd8ba6")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from leililab/enzygen, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_json Ran
ran; our contract draft was wrong, not the code
blvns/x-elm/metaseq/file_io.py
code served (permissive licence) · get_code("c9fd54c09adfb25d")
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 dict. 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 json
import tempfile
import os

# Create a temporary JSON file with valid JSON content
temp_dir = tempfile.mkdtemp()
temp_json_path = os.path.join(temp_dir, "test.json")

test_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "object"}
}

with open(temp_json_path, 'w') as f:
    json.dump(test_data, f)

ARGS = {
    "p": temp_json_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
recursively_cast_dictconfigs Ran
ran; our contract draft was wrong, not the code
blvns/x-elm/metaseq/file_io.py
code served (permissive licence) · get_code("cec2774d64ea9719")
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 dict. 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.

from omegaconf import DictConfig
import torch

torch.manual_seed(0)

# Create a nested DictConfig structure to test the recursive casting
nested_dict_config = DictConfig({
    "model": {
        "layers": 3,
        "hidden_dim": 64
    },
    "training": {
        "lr": 0.001,
        "epochs": 10
    },
    "data": {
        "batch_size": 32,
        "nested": {
            "value": 42,
            "flag": True
        }
    }
})

ARGS = {
    "cfg": nested_dict_config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
omegaconf, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
url_to_filename Ran
ran; our contract draft was wrong, not the code
blvns/x-elm/metaseq/file_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from agi-content-generation/EKD, is pointer only (licence: NOASSERTION) · get_code("9418b43048e4e4ec")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from agi-content-generation/EKD, and GitHub could not classify the licence of that repository (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

with_incremental_state Ran
ran on a synthesized input
blvns/x-elm/metaseq/incremental_decoding_utils.py
code served (permissive licence) · get_code("1dcaa0202f989ada")
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 type. 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 torch.nn as nn
from typing import Dict, Optional
from torch import Tensor

torch.manual_seed(0)

# The function `with_incremental_state` takes a single parameter: a class `cls`
# We need to create a simple class to pass to it

class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

ARGS = {
    'cls': SimpleModule
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
filename_to_url Not yet run
unverified (harvested, no recorded run)
blvns/x-elm/metaseq/file_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from agi-content-generation/EKD, is pointer only (licence: NOASSERTION) · get_code("aafdc12f01ab388f")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from agi-content-generation/EKD, and GitHub could not classify the licence of that repository (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

load_archive_file Not yet run
unverified (harvested, no recorded run)
blvns/x-elm/metaseq/file_utils.py
code served (permissive licence) · get_code("0d372aa1e290a613")
How it ran

no invocation recorded.

load_model Not yet run
unverified (harvested, no recorded run)
blvns/x-elm/metaseq/sequence_scorer_btm.py
code served (permissive licence) · get_code("e5ec2c0476e20add")
How it ran

no invocation recorded.

torch_load_cpu Not yet run
unverified (harvested, no recorded run)
blvns/x-elm/metaseq/file_io.py
code served (permissive licence) · get_code("b2e331c14695c39b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 5 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 0 the same month, 4 later) and 5 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
url_to_filename 48 other papers
43 dated earlier, 4 later, 1 same year, order not determinable · 8 only as an assembly brick
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
1809.11096 (2018-09)
Large Scale GAN Training for High Fidelity Natural Image Synthesis
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
and 45 more not named here
filename_to_url 12 other papers
9 dated earlier, 3 later
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
2007.07691 (2020-07)
A Multilingual Parallel Corpora Collection Effort for Indian Languages
2021.findings-emnlp.366 (2021 (year only))
Rethinking Zero-shot Neural Machine Translation: From a Perspective of Latent Variables
and 9 more not named here
load_archive_file 12 other papers
9 dated earlier, 3 later
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
2021.findings-emnlp.366 (2021 (year only))
Rethinking Zero-shot Neural Machine Translation: From a Perspective of Latent Variables
2104.08771 (2021-04)
Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translation
and 9 more not named here
find_offsets 5 other papers
4 dated earlier, 1 later
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
2203.10426 (2022-03)
STEMM: Self-learning with Speech-text Manifold Mixup for Speech Translation
2205.01068 (2022-05)
OPT: Open Pre-trained Transformer Language Models
and 2 more not named here
with_incremental_state 1 other paper
1 dated earlier
2205.01068 (2022-05)
OPT: Open Pre-trained Transformer Language Models

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

Despite their popularity in non-English NLP, multilingual language models often underperform monolingual ones due to inter-language competition for model parameters. We propose Cross-lingual Expert Language Models (X-ELM), which mitigate this competition by independently training language models on subsets of the multilingual corpus. This process specializes X-ELMs to different languages while remaining effective as a multilingual ensemble. Our experiments show that when given the same compute budget, X-ELM outperforms jointly trained multilingual models across all 16 considered languages and that these gains transfer to downstream tasks. X-ELM provides additional benefits over performance improvements: new experts can be iteratively added, adapting X-ELM to new languages without catastrophic forgetting. Furthermore, training is asynchronous, reducing the hardware requirements for multilingual training and democratizing multilingual modeling.

For agents

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

get_harvested_code_for_paper("2401.10440")
get_code_for_paper("2401.10440")
have("2401.10440")

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

curl https://syntology.ai/api/ran/2401.10440.json

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

[![Syntology run record](https://syntology.ai/api/ran/2401.10440.svg)](https://syntology.ai/paper/2401.10440)

Connect an agent — have() is free.