SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2412.00876 · ICLR · 2025

Dynamic-LLaVA: Efficient Multimodal Large Language Models via Dynamic Vision-language Context Sparsification

Yao Hu, Yunhang Shen, Wenxuan Huang, Zijie Zhai, Shaosheng Cao, Fei Zhao, Xiangfeng Xu, Zheyu Ye, Shaohui Lin

arXiv · PDF · Open in the Atlas

Code that ran

13 of 18 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 4 have no contract verdict; 8 failures of our instrument; 5 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/2412.00876.json · badge: /api/ran/2412.00876.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
Osilly/dynamic_llava canonical (classifier label)
classifier basis: census
7 ran · 0 no recorded run
osilly/dynamic_llava community
Papers with Code archive
6 ran · 5 no recorded run
FunctionStatusWhere it lives
LlamaDynamicNTKScalingRotaryEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("68bd42b9c09a9dd1")
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, LlamaDynamicNTKScalingRotaryEmbedding: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    'dim': 4,
    'max_position_embeddings': 2048,
    'base': 10000,
    'device': None,
    'scaling_factor': 1.0
}
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 LlamaDynamicNTKScalingRotaryEmbedding — 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 LlamaDynamicNTKScalingRotaryEmbedding 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
LlamaLinearScalingRotaryEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("1287c98591685f54")
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, LlamaLinearScalingRotaryEmbedding: 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
from torch import nn

torch.manual_seed(0)

# LlamaLinearScalingRotaryEmbedding requires:
# - dim: int (embedding dimension, must be even)
# - max_position_embeddings: int (default 2048)
# - base: int (default 10000)
# - device: optional device
# - scaling_factor: float (default 1.0)

ARGS = {
    "dim": 4,
    "max_position_embeddings": 128,
    "base": 10000,
    "device": None,
    "scaling_factor": 2.0,
}
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 LlamaLinearScalingRotaryEmbedding — 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 LlamaLinearScalingRotaryEmbedding 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
LlamaMLP Ran
ran on a synthesized input · constructed an object; no result was computed
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("c5d348f0b0059642")
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, LlamaMLP: 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.activations import ACT2FN

torch.manual_seed(0)

class Config:
    def __init__(self, hidden_size=4, intermediate_size=8, hidden_act="silu", pretraining_tp=1):
        self.hidden_size = hidden_size
        self.intermediate_size = intermediate_size
        self.hidden_act = hidden_act
        self.pretraining_tp = pretraining_tp

config = Config(hidden_size=4, intermediate_size=8, hidden_act="silu", pretraining_tp=1)

x = torch.randn(2, 3, 4)

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 LlamaMLP — 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 LlamaMLP 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
LlamaRotaryEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("7541f97d86797bda")
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, LlamaRotaryEmbedding: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    'dim': 4,
    'max_position_embeddings': 2048,
    'base': 10000,
    'device': 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 LlamaRotaryEmbedding — 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 LlamaRotaryEmbedding 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
collate_fn Ran
ran; our contract draft was wrong, not the code
Osilly/dynamic_llava/llava/dynamic_eval/model_vqa_loader.py
code served (permissive licence) · get_code("20e4f665698a3d18")
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 collate_fn expects a batch which is a list/iterable of tuples
# Each tuple contains (input_ids, image_tensor, image_size)
# The function unpacks these, stacks input_ids and image_tensors, and returns image_sizes as-is

# Create sample data items
item1 = (
    torch.randint(0, 100, (4,)),  # input_ids: small tensor
    torch.randn(3, 8, 8),          # image_tensor: 3-channel image, 8x8
    (8, 8)                         # image_size: tuple
)

item2 = (
    torch.randint(0, 100, (4,)),  # input_ids: same shape
    torch.randn(3, 8, 8),          # image_tensor: same shape
    (8, 8)                         # image_size: tuple
)

# Create batch as a list of tuples
batch = [item1, item2]

ARGS = {
    "batch": batch
}
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
convert_caps Ran
ran; our contract draft was wrong, not the code
Osilly/dynamic_llava/llava/dynamic_eval/eval_science_qa.py
code served (permissive licence) · get_code("94fda0761ce0d5b9")
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 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

torch.manual_seed(0)

# Construct valid arguments for convert_caps function
results = [
    {
        'question_id': 42,
        'text': 'A dog running in the park'
    },
    {
        'question_id': 123,
        'text': 'A cat sitting on a mat'
    }
]

ARGS = {
    'results': results
}
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 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
get_chunk Ran
ran; our contract draft was wrong, not the code
Osilly/dynamic_llava/llava/dynamic_eval/model_lvis_for_meteor.py
code served (permissive licence) · get_code("42a46570620cd9fa")
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 list, ndim 1, dtype family int; 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 math

def split_list(lst, n):
    """Split a list into n (roughly) equal-sized chunks"""
    chunk_size = math.ceil(len(lst) / n)
    return [lst[i:i + chunk_size] for i in range(0, len(lst), chunk_size)]

def get_chunk(lst, n, k):
    chunks = split_list(lst, n)
    return chunks[k]

# Setup code
lst = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
n = 3
k = 1

ARGS = {
    "lst": lst,
    "n": n,
    "k": k
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 int — 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
get_pred_idx Ran
ran, violated the contract we drafted
Osilly/dynamic_llava/llava/dynamic_eval/eval_science_qa.py
code served (permissive licence) · get_code("36e875f85baa4565")
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 int, ndim 0, dtype family int, 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 random

def get_pred_idx(prediction, choices, options):
    """
    Get the index (e.g. 2) from the prediction (e.g. 'C')
    """
    if prediction in options[:len(choices)]:
        return options.index(prediction)
    else:
        return -1
        return random.choice(range(len(choices)))

# Setup code
prediction = 'B'
choices = ['option1', 'option2', 'option3']
options = ['A', 'B', 'C', 'D']

ARGS = {
    'prediction': prediction,
    'choices': choices,
    'options': options
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 int, ndim 0, dtype family int, 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
prompt_processor Ran
ran; our contract draft was wrong, not the code
Osilly/dynamic_llava/llava/dynamic_eval/eval_textvqa.py
code served (permissive licence) · get_code("1e68840d6bbdface")
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 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 re

def prompt_processor(prompt):
    if prompt.startswith('OCR tokens: '):
        pattern = 'Question: (.*?) Short answer:'
        match = re.search(pattern, prompt, re.DOTALL)
        question = match.group(1)
    elif 'Reference OCR token: ' in prompt and len(prompt.split('\n')) == 3:
        if prompt.startswith('Reference OCR token:'):
            question = prompt.split('\n')[1]
        else:
            question = prompt.split('\n')[0]
    elif len(prompt.split('\n')) == 2:
        question = prompt.split('\n')[0]
    else:
        assert False
    return question.lower()


# Test case 1: prompt starting with 'OCR tokens: '
prompt_1 = 'OCR tokens: Question: What is this? Short answer: test'

# Test case 2: prompt with 'Reference OCR token: ' and 3 lines, starting with it
prompt_2 = 'Reference OCR token: \nWhat is the answer?\nMore info'

# Test case 3: prompt with exactly 2 lines
prompt_3 = 'What is the question?\nSome context'

ARGS = {
    'prompt': prompt_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
scaled_dot_product_attention_with_policy Ran
ran; our contract draft was wrong, not the code
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("bc218bfbe73e286b")
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 Tensor, ndim 4, dtype family float, shape (2, 2, 3, 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 math

torch.manual_seed(0)

# Tiny dimensions: batch=2, seq_len=3, key_len=3, head_dim=4, num_heads=2
B = 2
L = 3  # query sequence length
S = 3  # key/value sequence length
H = 2  # number of heads
D = 4  # head dimension

# Create query, key, value tensors
# Shape: (B, H, L, D) for query and (B, H, S, D) for key/value
query = torch.randn(B, H, L, D)
key = torch.randn(B, H, S, D)
value = torch.randn(B, H, S, D)

# Optional attention mask (can be None or a tensor)
attn_mask = None

# Dropout probability
dropout_p = 0.1

# Causal masking flag
is_causal = False

# Scale factor (None means it will be computed as 1/sqrt(D))
scale = None

# Policy tensor: shape (B, N, 1) where N is sequence length
# This is used in softmax_with_policy to reshape to (B, 1, 1, N)
policy = torch.ones(B, S, 1)

ARGS = {
    "query": query,
    "key": key,
    "value": value,
    "attn_mask": attn_mask,
    "dropout_p": dropout_p,
    "is_causal": is_causal,
    "scale": scale,
    "policy": policy,
}
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 Tensor, ndim 4, dtype family float, shape (2, 2, 3, 4) — shape (2, 2, 3, 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
softmax_with_policy Ran
ran; our contract draft was wrong, not the code
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("aa7e8923d37f93a8")
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 4, dtype family float, shape (2, 3, 4, 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

torch.manual_seed(0)

# Analyze function signature and body:
# - attn: 4D tensor, used with .size() -> (B, H, N, N)
# - policy: 3D tensor, used with .size() -> (B, N, _), .reshape(), .dtype, .device
# - eps: float, default 1e-06

# Small dimensions: B=2, H=3, N=4
B, H, N = 2, 3, 4

# attn shape: (B, H, N, N) = (2, 3, 4, 4)
attn = torch.randn(B, H, N, N, dtype=torch.float32)

# policy shape: (B, N, 1) - the third dim is "_" but needs to exist
policy = torch.randn(B, N, 1, dtype=torch.float32)

# eps parameter
eps = 1e-06

ARGS = {
    "attn": attn,
    "policy": policy,
    "eps": eps,
}
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 4, dtype family float, shape (2, 3, 4, 4) — shape (2, 3, 4, 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
split_list Ran
ran; our contract draft was wrong, not the code
Osilly/dynamic_llava/llava/dynamic_eval/model_lvis_for_meteor.py
code served (permissive licence) · get_code("076c252c52cbb161")
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 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 math

def split_list(lst, n):
    """Split a list into n (roughly) equal-sized chunks"""
    chunk_size = math.ceil(len(lst) / n)
    return [lst[i:i + chunk_size] for i in range(0, len(lst), chunk_size)]

# Setup code
lst = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
n = 3

ARGS = {
    "lst": lst,
    "n": n
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
unpad_image Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Osilly/dynamic_llava/llava/model/dynamic_llava_arch.py
code served (permissive licence) · get_code("55c32993da87759b")
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 Tensor, ndim 3, dtype family float, shape (3, 10, 12) 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)

# Create a padded image tensor in CxHxW format
# Let's say original image was 8x6 (width x height)
# After padding and resizing, it became 12x12 (square)
# C=3 (RGB), H=12, W=12
tensor = torch.randn(3, 12, 12)

# Original size as (width, height) tuple
original_size = (8, 6)

ARGS = {
    "tensor": tensor,
    "original_size": original_size,
}
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 Tensor, ndim 3, dtype family float, shape (3, 10, 12) — shape (3, 10, 12) 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
Cache Not yet run
unverified (harvested, no recorded run)
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("d866787c2702613f")
How it ran

no invocation recorded.

DynamicLlamaDecoderLayer Not yet run
unverified (harvested, no recorded run)
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("2ee5fbf088ba5b7e")
How it ran

no invocation recorded.

DynamicLlamaSdpaAttention Not yet run
unverified (harvested, no recorded run)
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("c0464d1e0119ec7f")
How it ran

no invocation recorded.

LlamaAttention Not yet run
unverified (harvested, no recorded run)
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("64c5b45fc73cecfd")
How it ran

no invocation recorded.

LlamaFlashAttention2 Not yet run
unverified (harvested, no recorded run)
osilly/dynamic_llava/llava/model/language_model/dynamic_modeling_llama.py
code served (permissive licence) · get_code("58c176deebe204d2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 9 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 4 the same month, 6 later) and 9 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
split_list 197 other papers
133 dated earlier, 11 the same month, 46 later, 6 same year, order not determinable, 1 undated
2112.05682 (2021-12)
Self-attention Does Not Need $O(n^2)$ Memory
2306.17107 (2023-06)
LLaVAR: Enhanced Visual Instruction Tuning for Text-Rich Image Understanding
2307.10490 (2023-07)
Abusing Images and Sounds for Indirect Instruction Injection in Multi-Modal LLMs
and 194 more not named here
get_chunk 189 other papers
128 dated earlier, 11 the same month, 44 later, 6 same year, order not determinable
2112.05682 (2021-12)
Self-attention Does Not Need $O(n^2)$ Memory
2306.17107 (2023-06)
LLaVAR: Enhanced Visual Instruction Tuning for Text-Rich Image Understanding
2307.10490 (2023-07)
Abusing Images and Sounds for Indirect Instruction Injection in Multi-Modal LLMs
and 186 more not named here
unpad_image 69 other papers
34 dated earlier, 7 the same month, 28 later · 5 only as an assembly brick
2304.08485 (2023-04)
Visual Instruction Tuning
2309.09958 (2023-09)
An Empirical Study of Scaling Instruct-Tuned Large Multimodal Models
2310.03744 (2023-10)
Improved Baselines with Visual Instruction Tuning
and 66 more not named here
collate_fn 54 other papers
34 dated earlier, 3 the same month, 17 later
2308.13566 (2023-08)
MLLM-DataEngine: An Iterative Refinement Approach for MLLM
2310.08588 (2023-10)
Octopus: Embodied Vision-Language Programmer from Environmental Feedback
2311.07362 (2023-11)
Volcano: Mitigating Multimodal Hallucination through Self-Feedback Guided Revision
and 51 more not named here
convert_caps 3 other papers
2 dated earlier, 1 later
2304.08485 (2023-04)
Visual Instruction Tuning
2403.09792 (2024-03)
Images are Achilles' Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large Language Models
2503.15621 (2025-03)
LLaVA-MORE: A Comparative Study of LLMs and Visual Backbones for Enhanced Visual Instruction Tuning
get_pred_idx 3 other papers
2 dated earlier, 1 later
2304.08485 (2023-04)
Visual Instruction Tuning
2403.09792 (2024-03)
Images are Achilles' Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large Language Models
2503.15621 (2025-03)
LLaVA-MORE: A Comparative Study of LLMs and Visual Backbones for Enhanced Visual Instruction Tuning
LlamaMLP
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
2203.08913 (2022-03)
Memorizing Transformers
2310.03668 (2023-10)
GoLLIE: Annotation Guidelines improve Zero-Shot Information-Extraction
Cache
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2309.17453 (2023-09)
Efficient Streaming Language Models with Attention Sinks
prompt_processor 1 other paper
1 dated earlier
2403.09792 (2024-03)
Images are Achilles' Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large 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

Multimodal Large Language Models (MLLMs) have achieved remarkable success in vision understanding, reasoning, and interaction. However, the inference computation and memory increase progressively with the generation of output tokens during decoding, directly affecting the efficacy of MLLMs. Existing methods attempt to reduce the vision context redundancy to achieve efficient MLLMs. Unfortunately, the efficiency benefits of the vision context reduction in the prefill stage gradually diminish during the decoding stage. To address this problem, we proposed a dynamic vision-language context sparsification framework Dynamic-LLaVA, which dynamically reduces the redundancy of vision context in the prefill stage and decreases the memory and computation overhead of the generated language context during decoding. Dynamic-LLaVA designs a tailored sparsification inference scheme for different inference modes, i.e., prefill, decoding with and without KV cache, to achieve efficient inference of MLLMs. In practice, Dynamic-LLaVA can reduce computation consumption by ∼75% in the prefill stage. Meanwhile, throughout the entire generation process of MLLMs, Dynamic-LLaVA reduces the ∼50% computation consumption under decoding without KV cache, while saving ∼50% GPU memory overhead when decoding with KV cache, due to the vision-language context sparsification. Extensive experiments also demonstrate that Dynamic-LLaVA achieves efficient inference for MLLMs with negligible understanding and generation ability degradation or even performance gains compared to the full-context inference baselines. Code is available at https://github.com/Osilly/dynamic_llava .

For agents

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

get_harvested_code_for_paper("2412.00876")
get_code_for_paper("2412.00876")
have("2412.00876")

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

curl https://syntology.ai/api/ran/2412.00876.json

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

[![Syntology run record](https://syntology.ai/api/ran/2412.00876.svg)](https://syntology.ai/paper/2412.00876)

Connect an agent — have() is free.