SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2501.03218 · CVPR · 2025

Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reaction

Rui Qian, Xiaoyi Dong, Dahua Lin, Yuhang Cao, Jiaqi Wang, Yuhang Zang, Pan Zhang, Shuangrui Ding

arXiv · PDF · Open in the Atlas

Code that ran

5 of 10 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 5 have no contract verdict; 0 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/2501.03218.json · badge: /api/ran/2501.03218.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
Mark12Ding/Dispider canonical (classifier label)
classifier basis not recorded on the link
5 ran · 5 no recorded run
FunctionStatusWhere it lives
build_mlp_projector Ran
ran on a synthesized input
Mark12Ding/Dispider/dispider/model/projectors.py
code served (permissive licence) · get_code("bd38f2364600345e")
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 Sequential. 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 input_size and hidden_size as parameters
# These are integers that define the dimensions of the neural network layers
# input_size: dimension of input features
# hidden_size: dimension of hidden layer

ARGS = {
    "input_size": 8,
    "hidden_size": 16,
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
build_vision_projector Ran
ran on a synthesized input
Mark12Ding/Dispider/dispider/model/projectors.py
code served (permissive licence) · get_code("58a4a0132d489ed1")
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 PoolProjector. 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
import torch
from torch import nn

torch.manual_seed(0)

class MockConfig:
    def __init__(self):
        self.perception_projector_type = 'pool'
        self.perception_hidden_size = 8
        self.hidden_size = 16
        self.resolution = 16
        self.pool_num = 4

config = MockConfig()

ARGS = {
    'config': config
}
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 PoolProjector — a PoolProjector output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PoolProjector 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_model_name_from_path Ran
ran on a synthesized input
Mark12Ding/Dispider/dispider/mm_utils.py
code served (permissive licence) · get_code("aa373e6520146232")
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 torch.nn as nn

# Analyzing the function:
# - It takes model_path as a string parameter
# - It strips leading/trailing '/', splits by '/'
# - Checks if last part starts with 'checkpoint-'
# - Returns either combined name or just the last part

# Test case 1: path with checkpoint
model_path_with_checkpoint = "models/bert_base/checkpoint-500/"

# Test case 2: path without checkpoint
model_path_without_checkpoint = "models/gpt2"

ARGS = {
    "model_path": "models/bert_base/checkpoint-500/"
}
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
prepare_reaction_config Ran
ran on a synthesized input
Mark12Ding/Dispider/dispider/model/checkpoint.py
code served (permissive licence) · get_code("b3c79e3a6f04b89c")
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 SimpleConfig. 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 os
from dataclasses import dataclass
from pathlib import Path
from typing import Any, Mapping

@dataclass(frozen=True)
class CheckpointLayout:
    root: Path
    perception_decision_dir: Path
    manifest: Mapping[str, Any]

class SimpleConfig:
    def __init__(self):
        self.perception_decision = None
        self.checkpoint_format_version = None
        self._dispider_checkpoint_root = None

config = SimpleConfig()

layout = CheckpointLayout(
    root=Path("/tmp/checkpoint_root"),
    perception_decision_dir=Path("/tmp/perception_decision"),
    manifest={"key": "value"}
)

ARGS = {
    "config": config,
    "layout": layout
}
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 SimpleConfig — a SimpleConfig output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleConfig 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
process_images Ran
ran on a synthesized input
Mark12Ding/Dispider/dispider/mm_utils.py
code served (permissive licence) · get_code("18015a195a7c0e68")
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
from PIL import Image
import numpy as np

torch.manual_seed(0)

# Create small test images
def create_test_image(width, height):
    """Create a simple PIL Image for testing."""
    arr = np.random.randint(0, 256, (height, width, 3), dtype=np.uint8)
    return Image.fromarray(arr, mode='RGB')

# Create image processor mock
class MockImageProcessor:
    def __init__(self):
        self.image_mean = [0.5, 0.5, 0.5]  # Normalized values
    
    def __call__(self, images, return_tensors=None):
        """Process images and return tensors."""
        if not isinstance(images, list):
            images = [images]
        # Return pixel_values with shape (batch, channels, height, width)
        pixel_values = torch.randn(len(images), 3, 4, 4)
        return {'pixel_values': pixel_values}
    
    def preprocess(self, image, return_tensors=None):
        """Preprocess a single image."""
        # Return shape (1, channels, height, width)
        pixel_values = torch.randn(1, 3, 4, 4)
        return {'pixel_values': pixel_values}

# Create model config mock
class MockModelConfig:
    def __init__(self, image_aspect_ratio=None):
        self.image_aspect_ratio = image_aspect_ratio

# Scenario 1: image_aspect_ratio is 'pad' (will use expand_square path)
images = [create_test_image(5, 3), create_test_image(4, 4)]
image_processor = MockImageProcessor()
model_config = MockModelConfig(image_aspect_ratio='pad')

ARGS = {
    'images': images,
    'image_processor': image_processor,
    'model_config': model_config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, 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
build_perception Not yet run
unverified (harvested, no recorded run)
Mark12Ding/Dispider/dispider/model/perception.py
code served (permissive licence) · get_code("6c116aeefac7b3a5")
How it ran

no invocation recorded.

build_perception_decision Not yet run
unverified (harvested, no recorded run)
Mark12Ding/Dispider/dispider/model/perception_decision.py
code served (permissive licence) · get_code("ff51c0620de3bdbd")
How it ran

no invocation recorded.

load_checkpoint_layout Not yet run
unverified (harvested, no recorded run)
Mark12Ding/Dispider/dispider/model/checkpoint.py
code served (permissive licence) · get_code("1e45ed6e9055aed1")
How it ran

no invocation recorded.

load_pretrained_model Not yet run
unverified (harvested, no recorded run)
Mark12Ding/Dispider/dispider/model/builder.py
code served (permissive licence) · get_code("012f01ab58d287a3")
How it ran

no invocation recorded.

resolve_checkpoint_root Not yet run
unverified (harvested, no recorded run)
Mark12Ding/Dispider/dispider/model/checkpoint.py
code served (permissive licence) · get_code("b2fe0210d7b3fa5a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 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 10 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

Active real-time interaction with video LLMs introduces a new paradigm for human-computer interaction, where the model not only understands user intent but also responds while continuously processing streaming video on the fly. Unlike offline video LLMs, which analyze the entire video before answering questions, active real-time interaction requires three capabilities: 1) Perception: real-time video monitoring and interaction capturing. 2) Decision: raising proactive interaction in proper situations, 3) Reaction: continuous interaction with users. However, inherent conflicts exist among the desired capabilities. The Decision and Reaction require a contrary Perception scale and grain, and the autoregressive decoding blocks the real-time Perception and Decision during the Reaction. To unify the conflicted capabilities within a harmonious system, we present Dispider, a solution built on a Disentangled Perception, Decision, and Reaction framework. Dispider features a lightweight Proactive Streaming Video Processing module that tracks the video stream and identifies optimal moments for interaction. Once the interaction is triggered, an asynchronous Precise Interaction module provides detailed responses, while the processing module continues to monitor the video in the meantime. Our disentangled and asynchronous design ensures timely, contextually accurate, and computationally efficient responses, making Dispider ideal for active real-time interaction for long-duration video streams. Experiments prove that Dispider outperforms existing methods not only in its superior understanding of video content in conventional video QA settings, but also in proactive response capability and temporal awareness under the streaming setting. The code and model are released at https://github.com/Mark12Ding/Dispider.

For agents

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

get_harvested_code_for_paper("2501.03218")
get_code_for_paper("2501.03218")
have("2501.03218")

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

curl https://syntology.ai/api/ran/2501.03218.json

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

[![Syntology run record](https://syntology.ai/api/ran/2501.03218.svg)](https://syntology.ai/paper/2501.03218)

Connect an agent — have() is free.