SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.03511 · ACL · 2024

MoExtend: Tuning New Experts for Modality and Task Extension

Zhongzhan Huang, Pan Zhou, Marinka Zitnik, Shanghua Gao, Shanshan Zhong, Wushao Wen

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 1 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/2408.03511.json · badge: /api/ran/2408.03511.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
zhongshsh/MoExtend community
Papers with Code archive
3 ran · 1 no recorded run
FunctionStatusWhere it lives
MixtralConfig Ran
ran on a synthesized input · constructed an object; no result was computed
zhongshsh/MoExtend/src/model_moextend.py
code served (permissive licence) · get_code("f251388ba91a580d")
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 an instance of its own class, MixtralConfig: the run constructed the object and did not call it, so no result was computed. 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)

ARGS = {
    "vocab_size": 32000,
    "hidden_size": 4096,
    "intermediate_size": 14336,
    "num_hidden_layers": 32,
    "num_attention_heads": 32,
    "num_key_value_heads": 8,
    "hidden_act": "silu",
    "max_position_embeddings": 131072,
    "initializer_range": 0.02,
    "rms_norm_eps": 1e-05,
    "use_cache": True,
    "pad_token_id": 0,
    "bos_token_id": 1,
    "eos_token_id": 2,
    "tie_word_embeddings": False,
    "rope_theta": 1000000.0,
    "sliding_window": 4096,
    "attention_dropout": 0.0,
    "num_experts_per_tok": 2,
    "num_local_experts": 8,
    "output_router_logits": False,
    "router_aux_loss_coef": 0.001,
}
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 MixtralConfig — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MixtralConfig 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
MixtralSparseMoeBlock Ran
ran on a synthesized input · constructed an object; no result was computed
zhongshsh/MoExtend/src/model_moextend.py
code served (permissive licence) · get_code("0983b4b1438b8ac2")
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 an instance of its own class, MixtralSparseMoeBlock: the run constructed the object and did not call it, so no result was computed. 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 transformers.configuration_utils import PretrainedConfig

torch.manual_seed(0)

class MixtralConfig(PretrainedConfig):
    def __init__(
        self,
        vocab_size=32000,
        hidden_size=4096,
        intermediate_size=14336,
        num_hidden_layers=32,
        num_attention_heads=32,
        num_key_value_heads=8,
        hidden_act="silu",
        max_position_embeddings=131072,
        initializer_range=0.02,
        rms_norm_eps=1e-5,
        use_cache=True,
        pad_token_id=None,
        bos_token_id=1,
        eos_token_id=2,
        tie_word_embeddings=False,
        rope_theta=1000000.0,
        sliding_window=4096,
        attention_dropout=0.0,
        num_experts_per_tok=2,
        num_local_experts=8,
        output_router_logits=False,
        router_aux_loss_coef=0.001,
        **kwargs
    ):
        self.vocab_size = vocab_size
        self.hidden_size = hidden_size
        self.intermediate_size = intermediate_size
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.num_key_value_heads = num_key_value_heads
        self.hidden_act = hidden_act
        self.max_position_embeddings = max_position_embeddings
        self.initializer_range = initializer_range
        self.rms_norm_eps = rms_norm_eps
        self.use_cache = use_cache
        self.pad_token_id = pad_token_id
        self.bos_token_id = bos_token_id
        self.eos_token_id = eos_token_id
        self.tie_word_embeddings = tie_word_embeddings
        self.rope_theta = rope_theta
        self.sliding_window = sliding_window
        self.attention_dropout = attention_dropout
        self.num_experts_per_tok = num_experts_per_tok
        self.num_local_experts = num_local_experts
        self.output_router_logits = output_router_logits
        self.router_aux_loss_coef = router_aux_loss_coef
        super().__init__(**kwargs)

config = MixtralConfig(
    vocab_size=256,
    hidden_size=64,
    intermediate_size=128,
    num_hidden_layers=2,
    num_attention_heads=4,
    num_key_value_heads=2,
    hidden_act="silu",
    max_position_embeddings=512,
    initializer_range=0.02,
    rms_norm_eps=1e-5,
    use_cache=True,
    pad_token_id=0,
    bos_token_id=1,
    eos_token_id=2,
    tie_word_embeddings=False,
    rope_theta=1000000.0,
    sliding_window=256,
    attention_dropout=0.1,
    num_experts_per_tok=2,
    num_local_experts=4,
    output_router_logits=False,
    router_aux_loss_coef=0.001,
)

ARGS = {
    "config": config,
}
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 MixtralSparseMoeBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MixtralSparseMoeBlock 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
WeightScale Ran
ran on a synthesized input
zhongshsh/MoExtend/src/model_moextend.py
code served (permissive licence) · get_code("60cbc5cb83d64366")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, 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
from torch import nn

class WeightScale(nn.Module):

    def __init__(self):
        super().__init__()
        self.weight_scale = nn.Sequential(nn.Linear(9, 16, bias=False), nn.GELU(), nn.Linear(16, 9, bias=False))

    def forward(self, hidden_states):
        return self.weight_scale(hidden_states)


INIT_ARGS = {}

ARGS = {
    'hidden_states': torch.randn(2, 9)
}

if __name__ == '__main__':
    model = WeightScale(**INIT_ARGS)
    output = model(**ARGS)
    print(output.shape)
    print(output)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MixtralBLockSparseTop2MLP Not yet run
unverified (harvested, no recorded run)
zhongshsh/MoExtend/src/model_moextend.py
code served (permissive licence) · get_code("2f745d254ea9ba37")
How it ran

no invocation recorded.

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

Large language models (LLMs) excel in various tasks but are primarily trained on text data, limiting their application scope. Expanding LLM capabilities to include visionlanguage understanding is vital, yet training them on multimodal data from scratch is challenging and costly. Existing instruction tuning methods, e.g., LLAVA, often connects a pretrained CLIP vision encoder and LLMs via fully fine-tuning LLMs to bridge the modality gap. However, full fine-tuning is plagued by catastrophic forgetting, i.e., forgetting previous knowledge, and high training costs particularly in the era of increasing tasks and modalities. To solve this issue, we introduce MoExtend, an effective framework designed to streamline the modality adaptation and extension of Mixtureof-Experts (MoE) models. MoExtend seamlessly integrates new experts into pre-trained MoE models, endowing them with novel knowledge without the need to tune pretrained models such as MoE and vision encoders. This approach enables rapid adaptation and extension to new modal data or tasks, effectively addressing the challenge of accommodating new modalities within LLMs. Furthermore, MoExtend avoids tuning pretrained models, thus mitigating the risk of catastrophic forgetting. Experimental results demonstrate the efficacy and efficiency of MoExtend in enhancing the multimodal capabilities of LLMs, contributing to advancements in multimodal AI research. https://github.com/zhongshsh/MoExtend.

For agents

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

get_harvested_code_for_paper("2408.03511")
get_code_for_paper("2408.03511")
have("2408.03511")

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

curl https://syntology.ai/api/ran/2408.03511.json

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

[![Syntology run record](https://syntology.ai/api/ran/2408.03511.svg)](https://syntology.ai/paper/2408.03511)

Connect an agent — have() is free.