SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.06930 · 2023

mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 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; 4 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/2307.06930.json · badge: /api/ran/2307.06930.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
gregor-ge/mblip canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
output_loss Ran
ran; our contract draft was wrong, not the code
gregor-ge/mblip/src/tasks/vllm/evaluation.py
code served (permissive licence) · get_code("41a387217f0d60ae")
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 torch
import torch.nn as nn

torch.manual_seed(0)

# The function takes self, outputs dict, *args, **kwargs
# It accesses outputs['loss'] and calls .detach() and .unsqueeze(0) on it
# So outputs['loss'] must be a tensor

# Create a small loss tensor (scalar or 1D)
loss_tensor = torch.tensor([0.5], dtype=torch.float32)

# Create the outputs dict with the required 'loss' key
outputs = {
    'loss': loss_tensor,
    'other_key': torch.randn(2, 4)  # additional data that might be in outputs
}

# Create a dummy self (a simple class instance)
class DummyModel:
    def output_loss(self, outputs, *args, **kwargs):
        outputs['loss'] = outputs['loss'].detach().unsqueeze(0)
        return outputs

self_instance = DummyModel()

# Define ARGS mapping all parameters
ARGS = {
    'self': self_instance,
    'outputs': outputs,
    'args': (),
    'kwargs': {}
}
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 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
processing_hooks Ran
ran; our contract draft was wrong, not the code
gregor-ge/mblip/src/utils/hooks.py
code served (permissive licence) · get_code("257b04b56ec523a2")
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 function. 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
from functools import wraps

torch.manual_seed(0)

def processing_hooks(func):
    """
    Hooks for preprocessing that are pre- and appended to any built-in preprocessing function.

    Args:
        on_before_processing (:obj:`Callable`):
            Hook executed before the function call with the following function signature.

            .. code-block:: python

                def on_before_processing(inputs: dict, *args, **kwargs) -> dict:
                    ...
                    return inputs

        on_after_processing (:obj:`Callable`):
            Hook executed after the function call with the following function signature.

            .. code-block:: python

                def on_after_processing(inputs: dict, *args, **kwargs) -> BatchEncoding:
                    ...
                    return inputs
    """

    @wraps(func)
    def hooks(inputs: dict, *args, **kwargs):
        on_before_processing = kwargs.pop('on_before_processing', None)
        on_after_processing = kwargs.pop('on_after_processing', None)
        if on_before_processing is not None:
            inputs = on_before_processing(inputs, *args, **kwargs)
        inputs = func(inputs, *args, **kwargs)
        if on_after_processing is not None:
            inputs = on_after_processing(inputs, *args, **kwargs)
        return inputs
    return hooks


# Define a simple preprocessing function to be wrapped
def simple_preprocess(inputs: dict, *args, **kwargs) -> dict:
    """Simple preprocessing function that returns the inputs unchanged."""
    return inputs


# Define hook functions
def before_hook(inputs: dict, *args, **kwargs) -> dict:
    """Hook executed before processing."""
    inputs['before_called'] = True
    return inputs


def after_hook(inputs: dict, *args, **kwargs) -> dict:
    """Hook executed after processing."""
    inputs['after_called'] = True
    return inputs


# Create the wrapped function
wrapped_func = processing_hooks(simple_preprocess)

# Define arguments for the function
ARGS = {
    'func': simple_preprocess
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
set_generation_mode Ran
ran; our contract draft was wrong, not the code
gregor-ge/mblip/src/tasks/vllm/evaluation.py
code served (permissive licence) · get_code("1113dcafe2eb33cb")
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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal batch dictionary
batch = {
    'input_ids': torch.randint(0, 100, (2, 4)),
    'attention_mask': torch.ones(2, 4),
}

# Define ARGS mapping all parameters
ARGS = {
    'self': None,  # Will be bound to an instance
    'batch': batch,
    'mode': 'generate',
    'args': (),
    'kwargs': {'max_length': 50, 'top_p': 0.9}
}
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 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
validation_loss Ran
ran; our contract draft was wrong, not the code
gregor-ge/mblip/src/tasks/vllm/evaluation.py
code served (permissive licence) · get_code("c215d2db990935a2")
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 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.

import torch

torch.manual_seed(0)

# The function takes a 'loss' parameter and computes torch.mean on it
# So we need to pass a tensor (typically loss values)
loss_tensor = torch.randn(2, 4)

ARGS = {
    "loss": loss_tensor
}
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 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 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Modular vision-language models (Vision-LLMs) align pretrained image encoders with (frozen) large language models (LLMs) and post-hoc condition LLMs to `understand' the image input. With the abundance of readily available high-quality English image-text data as well as strong monolingual English LLMs, the research focus has been on English-only Vision-LLMs. Multilingual vision-language models are still predominantly obtained via expensive end-to-end pretraining, resulting in comparatively smaller models, trained on limited multilingual image data supplemented with text-only multilingual corpora. We present mBLIP, the first Vision-LLM leveraging multilingual LLMs, which we obtain in a computationally efficient manner on consumer-level hardware. To this end, we \textit{re-align} an image encoder previously tuned to an English LLM to a new, multilingual LLM using only a few million multilingual training examples derived from a mix of vision-and-language tasks, which we obtain by machine-translating high-quality English data to 95 languages. On the IGLUE benchmark and XM3600, mBLIP yields results competitive with state-of-the-art models and it greatly outperforms strong English-only Vision-LLMs like Llava 1.5. We release our model, code, and train data at \url{https://github.com/gregor-ge/mBLIP}.

For agents

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

get_harvested_code_for_paper("2307.06930")
get_code_for_paper("2307.06930")
have("2307.06930")

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

curl https://syntology.ai/api/ran/2307.06930.json

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

[![Syntology run record](https://syntology.ai/api/ran/2307.06930.svg)](https://syntology.ai/paper/2307.06930)

Connect an agent — have() is free.