SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.13476 · EMNLP · 2024

LLMs Are Zero-Shot Context-Aware Simultaneous Translators

Satoshi Nakamura, Roman Koshkin, Katsuhito Sudoh

arXiv · PDF · Open in the Atlas

Code that ran

9 of 9 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 0 have no contract verdict; 9 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/2406.13476.json · badge: /api/ran/2406.13476.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
RomanKoshkin/toLLMatch canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
FunctionStatusWhere it lives
count_matches Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/utils/utils.py
code served (permissive licence) · get_code("e004d713b242a34e")
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

torch.manual_seed(0)

# The function count_matches takes two strings and compares them character by character,
# counting matches and building a string of matched characters.
# It skips comparisons when either string has '_' at that position.

i = "hello_world"
j = "hello_there"

ARGS = {
    "i": i,
    "j": j,
}
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 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
dot_product_attention Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
RomanKoshkin/toLLMatch/whisper-jax/whisper_jax/layers.py
code served (permissive licence) · get_code("2000f34f95c683d6")
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 ArrayImpl, ndim 4, dtype family float, shape (2, 3, 2, 6) 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 jax
import jax.numpy as jnp
from jax import random

# Set seed for reproducibility
key = random.PRNGKey(0)

# Define small tensor dimensions
batch_size = 2
q_length = 3
kv_length = 4
num_heads = 2
qk_depth_per_head = 5
v_depth_per_head = 6

# Create query, key, value tensors with distinct shapes
query = jnp.ones((batch_size, q_length, num_heads, qk_depth_per_head), dtype=jnp.float32)
key = jnp.ones((batch_size, kv_length, num_heads, qk_depth_per_head), dtype=jnp.float32)
value = jnp.ones((batch_size, kv_length, num_heads, v_depth_per_head), dtype=jnp.float32)

# Create optional bias with shape broadcastable to [batch, num_heads, q_length, kv_length]
bias = jnp.zeros((batch_size, num_heads, q_length, kv_length), dtype=jnp.float32)

# Create dropout RNG key
dropout_rng = random.PRNGKey(1)

# Define ARGS dictionary with all parameters
ARGS = {
    'query': query,
    'key': key,
    'value': value,
    'bias': bias,
    'dropout_rng': dropout_rng,
    'dropout_rate': 0.1,
    'deterministic': False,
    'dtype': jnp.float32,
    'float32_logits': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 4, dtype family float, shape (2, 3, 2, 6) — shape (2, 3, 2, 6) 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
general_parser Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/SimulEval/simuleval/options.py
code served (permissive licence) · get_code("d56dce8864438c34")
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 ArgumentParser. 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 argparse
import sys
from typing import Optional

torch_imported = False
try:
    import torch
    torch_imported = True
    torch.manual_seed(0)
except ImportError:
    pass

# Minimal setup - the function signature has only two parameters
ARGS = {
    "config_dict": None,
    "parser": None,
}
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 ArgumentParser — a ArgumentParser output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ArgumentParser 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_last_word Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/evaluation/s2tt_agent.py
code served (permissive licence) · get_code("24bbf49accb51e6b")
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 list, 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

torch.manual_seed(0)

# Analyzing the function:
# - outputs: list of dicts with 'generated_text' key (string)
# - prompt: string to be used for slicing
# The function extracts text after the prompt and removes the last word

# Construct minimal valid arguments
prompt = "The quick brown"

outputs = [
    {
        'generated_text': "The quick brown fox jumps over"
    },
    {
        'generated_text': "The quick brown dog runs fast"
    }
]

ARGS = {
    'outputs': outputs,
    'prompt': prompt
}
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 list, ndim 1, 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
pad Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/utils/utils.py
code served (permissive licence) · get_code("9b6358ba7555d251")
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 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.

import torch

torch.manual_seed(0)

# Analyzing the function:
# - _a: used with len() and list concatenation with ['_'] * (...), so it's a list
# - _b: used with len() and list concatenation, so it's a list
# - overlap: used in arithmetic (len(_b) - overlap, len(_a) - overlap), so it's an int

_a = ['x', 'y', 'z']
_b = ['p', 'q']
overlap = 1

ARGS = {
    '_a': _a,
    '_b': _b,
    'overlap': overlap
}
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 tuple, ndim 2, 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
recov Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/utils/utils.py
code served (permissive licence) · get_code("9571938451acf1c3")
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 str. 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)

# The function recov takes a single parameter 'it' and checks if its first element
# (it[0]) equals the string '_'. If so, it returns it[1], otherwise it returns it[0].
# So 'it' needs to be indexable with at least 2 elements.

# Create a simple list that satisfies the function's requirements
it = ['_', 'value_at_index_1']

ARGS = {
    'it': it
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
template_0 Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/chat_templates/templates.py
code served (permissive licence) · get_code("ddad9955649d7c05")
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 str. 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 numpy as np

torch.manual_seed(0)

# Define arguments for template_0
SRC_LANG = "English"
TARG_LANG = "French"
BGD_INFO = "The conference is about climate policy. Key terms: carbon tax, emissions trading, renewable energy."

ARGS = {
    "SRC_LANG": SRC_LANG,
    "TARG_LANG": TARG_LANG,
    "BGD_INFO": BGD_INFO
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
template_1 Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/chat_templates/templates.py
code served (permissive licence) · get_code("962f03f540593974")
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 str. 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 numpy as np

# Examine the function signature and body
# template_1 takes three parameters: SRC_LANG, TARG_LANG, BGD_INFO
# SRC_LANG: string for source language
# TARG_LANG: string for target language
# BGD_INFO: optional string for background information (defaults to None)

# The function uses these parameters as:
# - SRC_LANG: formatted into f-string
# - TARG_LANG: formatted into f-string
# - BGD_INFO: checked for None, then formatted into f-string if not None

# Construct minimal valid arguments
SRC_LANG = "English"
TARG_LANG = "French"
BGD_INFO = "This is a conference about linguistics and translation technologies."

ARGS = {
    "SRC_LANG": SRC_LANG,
    "TARG_LANG": TARG_LANG,
    "BGD_INFO": BGD_INFO
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
template_2 Ran
ran; our contract draft was wrong, not the code
RomanKoshkin/toLLMatch/chat_templates/templates.py
code served (permissive licence) · get_code("984587db1419180f")
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 str. 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.

# Setup code for template_2

ARGS = {
    'SRC_LANG': 'English',
    'TARG_LANG': 'Russian',
    'BGD_INFO': 'Climate crisis refers to rapid global warming caused by greenhouse gases. The atmosphere contains oxygen and other gases that refract light.'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 8 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
dot_product_attention 5 other papers
5 dated earlier
2106.03598 (2021-06)
SciFive: a text-to-text transformer model for biomedical literature
2112.00791 (2021-12)
Controlling Conditional Language Models without Catastrophic Forgetting
2206.05408 (2022-06)
Multi-instrument Music Synthesis with Spectrogram Diffusion
and 2 more not named here

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

The advent of transformers has fueled progress in machine translation. More recently large language models (LLMs) have come to the spotlight thanks to their generality and strong performance in a wide range of language tasks, including translation. Here we show that open-source LLMs perform on par with or better than some state-of-the-art baselines in simultaneous machine translation (SiMT) tasks, zero-shot. We also demonstrate that injection of minimal background information, which is easy with an LLM, brings further performance gains, especially on challenging technical subject-matter. This highlights LLMs' potential for building next generation of massively multilingual, context-aware and terminologically accurate SiMT systems that require no resource-intensive training or fine-tuning. The code is available at https://github.com/RomanKoshkin/toLLMatch.

For agents

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

get_harvested_code_for_paper("2406.13476")
get_code_for_paper("2406.13476")
have("2406.13476")

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

curl https://syntology.ai/api/ran/2406.13476.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.13476.svg)](https://syntology.ai/paper/2406.13476)

Connect an agent — have() is free.