SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.12709 · 2024

MoME: Mixture of Multimodal Experts for Generalist Multimodal Large Language Models

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 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, 1 violated it and 0 have no contract verdict; 4 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/2407.12709.json · badge: /api/ran/2407.12709.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
jiutian-vl/mome canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
box_extract Ran
ran; our contract draft was wrong, not the code
jiutian-vl/mome/demo_mome.py
code served (permissive licence) · get_code("5653e2bbfb29e58c")
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. 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 re

middle_brackets_pat = re.compile('\\[\\d(?:\\.\\d*)?(?:,\\d(?:\\.\\d*)?){3}(?:;\\d(?:\\.\\d*)?(?:,\\d(?:\\.\\d*)?){3})*\\]')

def box_extract(string: str):
    ret = []
    for bboxes_str in middle_brackets_pat.findall(string):
        bboxes = []
        bbox_strs = bboxes_str.replace('(', '').replace(')', '').replace('[', '').replace(']', '').split(';')
        for bbox_str in bbox_strs:
            bbox = list(map(float, bbox_str.split(',')))
            bboxes.append(bbox)
        ret.append(bboxes)
    return ret

# Test strings with valid bbox patterns matching the regex
# Pattern: [d,d,d,d] or [d,d,d,d;d,d,d,d;...]
# where d is an integer or float like 1, 1.5, etc.

test_string = "Some text with [1,2,3,4] and [5.5,6.7,7.8,8.9;1.1,2.2,3.3,4.4] boxes"

ARGS = {
    "string": test_string
}
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 list — a list 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
disabled_train Ran
ran, violated the contract we drafted
jiutian-vl/mome/models/mome_model.py
code served (permissive licence) · get_code("4cb732f513d69dfd")
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) 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

torch.manual_seed(0)

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

model = SimpleModel()

ARGS = {
    "self": model,
    "mode": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
extract_ans Ran
ran, honoured the contract we drafted
jiutian-vl/mome/demo_mome.py
code served (permissive licence) · get_code("439482b5435bb1d9")
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 re

middle_brackets_pat = re.compile('\\[\\d(?:\\.\\d*)?(?:,\\d(?:\\.\\d*)?){3}(?:;\\d(?:\\.\\d*)?(?:,\\d(?:\\.\\d*)?){3})*\\]')

def box_extract(string: str):
    ret = []
    for bboxes_str in middle_brackets_pat.findall(string):
        bboxes = []
        bbox_strs = bboxes_str.replace('(', '').replace(')', '').replace('[', '').replace(']', '').split(';')
        for bbox_str in bbox_strs:
            bbox = list(map(float, bbox_str.split(',')))
            bboxes.append(bbox)
        ret.append(bboxes)
    return ret

def extract_ans(string: str):
    try:
        list_of_boxes = box_extract(string)
        if len(list_of_boxes) != 1 or len(list_of_boxes[0]) != 1:
            return None
        box = list_of_boxes[0][0]
        if len(box) != 4:
            return None
        return box
    except Exception as e:
        print(f'extract_ans for {string} but get exception: {e}')
        return None

ARGS = {
    "string": "The bounding box is [1.5,2.3,3.7,4.2] in the image."
}
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 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
forward_mlp_llama Ran
ran; our contract draft was wrong, not the code
jiutian-vl/mome/models/moe_adapters/mole_adapter.py
code served (permissive licence) · get_code("433e6abade4a9ee0")
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 float, shape (2, 4) 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
import torch.nn as nn

torch.manual_seed(0)

class Adapter(nn.Module):
    def __init__(self, hidden_size):
        super().__init__()
        self.linear = nn.Linear(hidden_size, hidden_size)
    
    def forward(self, x, add_residual=False):
        return self.linear(x)

class MLPLlama(nn.Module):
    def __init__(self):
        super().__init__()
        hidden_size = 4
        self.adapter = Adapter(hidden_size)
        self.gate_proj = nn.Linear(hidden_size, hidden_size)
        self.up_proj = nn.Linear(hidden_size, hidden_size)
        self.down_proj = nn.Linear(hidden_size, hidden_size)
        self.act_fn = nn.ReLU()
    
    def forward_mlp_llama(self, x):
        adapt_x = self.adapter(x, add_residual=False)
        x = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
        return x + adapt_x

model = MLPLlama()
x = torch.randn(2, 4)

ARGS = {
    "self": model,
    "x": x
}
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 float, shape (2, 4) — shape (2, 4) 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
named_apply Ran
ran; our contract draft was wrong, not the code
jiutian-vl/mome/models/dinov2_models/vision_transformer.py
code served (permissive licence) · get_code("e7fcb6d9ac9deaf4")
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) 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

torch.manual_seed(0)

# Create a simple nested module structure
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.layer1 = nn.Linear(4, 3)
        self.layer2 = nn.Sequential(
            nn.Linear(3, 2),
            nn.ReLU()
        )
    
    def forward(self, x):
        x = self.layer1(x)
        x = self.layer2(x)
        return x

# Create a simple callback function that tracks module names
collected_modules = []

def callback_fn(module, name):
    collected_modules.append((name, type(module).__name__))

# Instantiate the module
module = SimpleModel()

# Define ARGS dictionary with all parameters
ARGS = {
    'fn': callback_fn,
    'module': module,
    'name': '',
    'depth_first': True,
    'include_root': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
top1_gate Ran
ran; our contract draft was wrong, not the code
jiutian-vl/mome/models/moe_adapters/mole_adapter.py
code served (permissive licence) · get_code("eb82d58de85115b1")
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

torch.manual_seed(0)

# The function takes logits as input
# It performs softmax, max, scatter operations on tensors
# It expects logits to be a tensor with shape (batch_size, num_classes)

logits = torch.randn(2, 4)  # batch_size=2, num_classes=4

ARGS = {
    "logits": logits
}
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 — 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
create_eva_vit_g Not yet run
unverified (harvested, no recorded run)
jiutian-vl/mome/models/eva_vit.py
code served (permissive licence) · get_code("6d14779e4c4e5d4f")
How it ran

no invocation recorded.

get_image_from_plt Not yet run
unverified (harvested, no recorded run)
jiutian-vl/mome/demo_mome.py
code served (permissive licence) · get_code("c36ac37873d36891")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 2 the same month, 2 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.

FunctionAlso present inEarliest-datedOthers named
disabled_train 173 other papers
127 dated earlier, 3 the same month, 35 later, 8 same year, order not determinable · 5 only as an assembly brick
2012.09841 (2020-12)
Taming Transformers for High-Resolution Image Synthesis
2102.05379 (2021-02)
Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions
2104.07652 (2021-04)
Geometry-Free View Synthesis: Transformers and no 3D Priors
and 170 more not named here
named_apply 26 other papers
10 dated earlier, 1 the same month, 13 later, 2 undated
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
2204.07683 (2022-04)
Safe Self-Refinement for Transformer-based Domain Adaptation
2304.07193 (2023-04)
DINOv2: Learning Robust Visual Features without Supervision
and 23 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

Multimodal large language models (MLLMs) have demonstrated impressive capabilities across various vision-language tasks. However, a generalist MLLM typically underperforms compared with a specialist MLLM on most VL tasks, which can be attributed to task interference. In this paper, we propose a mixture of multimodal experts (MoME) to mitigate task interference and obtain a generalist MLLM. Our MoME is composed of two key components, a mixture of vision experts (MoVE) and a mixture of language experts (MoLE). MoVE can adaptively modulate the features transformed from various vision encoders, and has a strong compatibility in transformation architecture. MoLE incorporates sparsely gated experts into LLMs to achieve painless improvements with roughly unchanged inference costs. In response to task interference, our MoME specializes in both vision and language modality to adapt to task discrepancies. Extensive experiments show that MoME significantly improves the performance of generalist MLLMs across various VL tasks. The source code is released at https://github.com/JiuTian-VL/MoME

For agents

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

get_harvested_code_for_paper("2407.12709")
get_code_for_paper("2407.12709")
have("2407.12709")

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

curl https://syntology.ai/api/ran/2407.12709.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.12709.svg)](https://syntology.ai/paper/2407.12709)

Connect an agent — have() is free.