SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2411.16375 · ICML · 2025

Ca2-VDM: Efficient Autoregressive Video Diffusion Model with Causal Generation and Cache Sharing

Jiaxin Shi, Hanwang Zhang, Chunping Wang, Long Chen, Kaifeng Gao, Jun Xiao

arXiv · PDF · Open in the Atlas

Code that ran

15 of 21 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (5 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 8 failures of our instrument; 6 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/2411.16375.json · badge: /api/ran/2411.16375.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
universome/stylegan-v canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
dawn-lx/causal-videogen canonical (classifier label)
classifier basis: census
2 ran · 4 no recorded run
songweige/TATS community
Papers with Code archive
9 ran · 1 no recorded run
FunctionStatusWhere it lives
Block Ran
ran on a synthesized input
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("c36e0664ff375f5b")
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 3, 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
import torch.nn as nn
import math

class Config:
    def __init__(self):
        self.n_embd = 4
        self.n_head = 2
        self.block_size = 8
        self.attn_pdrop = 0.1
        self.resid_pdrop = 0.1
        self.n_unmasked = 0

INIT_ARGS = {
    'config': Config()
}

ARGS = {
    'x': torch.randn(2, 6, 4),
    'layer_past': None,
    'return_present': False
}
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 3, 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)
CausalSelfAttention Ran
ran on a synthesized input
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("0dcc3d2b5989bb48")
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 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
import torch.nn as nn
import math

class Config:
    def __init__(self):
        self.n_embd = 4
        self.n_head = 2
        self.block_size = 3
        self.attn_pdrop = 0.1
        self.resid_pdrop = 0.1
        self.n_unmasked = 0

INIT_ARGS = {
    'config': Config()
}

ARGS = {
    'x': torch.randn(2, 3, 4),
    'layer_past': None
}
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 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
GPT Ran
ran on a synthesized input · constructed an object; no result was computed
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("20e50a0db3aadd39")
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, GPT: 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)

class Args:
    """Stand-in for args object"""
    def __init__(self):
        self.sequence_length = 8
        self.resolution = 2

ARGS = {
    'args': Args(),
    'vocab_size': 50,
    'block_size': 16,
    'n_layer': 2,
    'n_head': 4,
    'n_embd': 64,
    'embd_pdrop': 0.1,
    'resid_pdrop': 0.1,
    'attn_pdrop': 0.1,
    'n_unmasked': 0,
    'vtokens_pos': False,
}
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 GPT — 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 GPT 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
GPTConfig Ran
ran on a synthesized input · constructed an object; no result was computed
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("cd407b6b077bc83e")
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, GPTConfig: 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": 256,
    "block_size": 128,
    "n_layer": 12,
    "n_head": 8,
    "n_embd": 64,
}
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 GPTConfig — 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 GPTConfig 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
Identity Ran
ran on a synthesized input · constructed an object; no result was computed
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("f020bb1b02990123")
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, Identity: 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 = {
    "quantize_interface": True
}
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 Identity — 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 Identity 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
Labelator Ran
ran on a synthesized input · constructed an object; no result was computed
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("b9f40f41b93875c6")
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, Labelator: 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)

# Labelator uses:
# - n_classes: an integer (used as parameter)
# - quantize_interface: a boolean (used in conditional)
# - encode method: takes c (a tensor) and returns either (c, c.long()) or c
#   The method does c[:, None] so c should be 2D, and indexing suggests batch dimension

ARGS = {
    "n_classes": 5,
    "quantize_interface": True,
}
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 Labelator — 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 Labelator 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
SOSProvider Ran
ran on a synthesized input · constructed an object; no result was computed
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("6e1c0c9a43690fdb")
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, SOSProvider: 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)

class SOSProvider(nn.Module):
    def __init__(self, sos_token, quantize_interface=True):
        super().__init__()
        self.sos_token = sos_token
        self.quantize_interface = quantize_interface

    def encode(self, x, **kwargs):
        c = torch.ones(x.shape[0], 1) * self.sos_token
        c = c.long().to(x.device)
        if self.quantize_interface:
            return (c, c)
        return c

# Create a small test tensor with batch size 2
x = torch.randn(2, 4)

ARGS = {
    'sos_token': 1,
    'quantize_interface': True
}
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 SOSProvider — 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 SOSProvider 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
accuracy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("085a09050f4f41dc")
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 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

torch.manual_seed(0)

# Create small tensors for testing
batch_size = 2
num_classes = 4

# output: logits tensor of shape (batch_size, num_classes)
# The function calls output.topk(), so it needs to be a 2D tensor
output = torch.randn(batch_size, num_classes)

# target: class indices of shape (batch_size,)
# The function calls target.size(0) and target.reshape(1, -1)
target = torch.tensor([0, 2])

# topk: tuple of k values to compute accuracy for
topk = (1, 2)

ARGS = {
    "output": output,
    "target": target,
    "topk": topk,
}
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 2, 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
build_module Ran
ran; our contract draft was wrong, not the code
dawn-lx/causal-videogen/opensora/registry.py
code served (permissive licence) · get_code("496f3cd2af0df2e1")
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 Linear. 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 copy import deepcopy


# Create a simple Registry class that can build modules
class SimpleRegistry:
    def build(self, cfg):
        """Build a module from config dict."""
        cfg_copy = deepcopy(cfg)
        module_type = cfg_copy.pop('type', None)
        
        if module_type == 'linear':
            return nn.Linear(cfg_copy.get('in_features', 4), cfg_copy.get('out_features', 3))
        elif module_type == 'relu':
            return nn.ReLU()
        else:
            # Default: return a simple Linear module
            return nn.Linear(4, 3)


# Create instances for the function arguments
torch.manual_seed(0)

# A simple module config dict
module_dict = {
    'type': 'linear',
    'in_features': 4,
    'out_features': 3
}

# A simple nn.Module instance
simple_module = nn.Linear(4, 3)

# A registry instance
builder = SimpleRegistry()

# Define ARGS with all parameters
ARGS = {
    'module': module_dict,
    'builder': builder,
}
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 Linear — a Linear output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Linear 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
cfg_to_args_str Ran
ran; our contract draft was wrong, not the code
universome/stylegan-v/src/infra/utils.py
pointer only (licence: NONE) · get_code("2dc4f95b756cb2b3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, universome/stylegan-v when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

construct_experiments_args Ran
ran; our contract draft was wrong, not the code
universome/stylegan-v/src/infra/slurm_batch_launch.py
pointer only (licence: NONE) · get_code("2c35749b6eaa5bfe")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, universome/stylegan-v when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

gather_forward_split_backward Ran
ran; our contract draft was wrong, not the code
dawn-lx/causal-videogen/opensora/acceleration/communications.py
code served (permissive licence) · get_code("97c082db7efd0240")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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.distributed as dist

torch.manual_seed(0)

# Mock ProcessGroup for CPU-only testing
class MockProcessGroup:
    pass

# Initialize a minimal distributed setup (CPU-only)
# We'll create a mock process group that works without actual distributed setup
mock_pg = MockProcessGroup()

# Monkey-patch dist functions to work without actual distributed initialization
original_get_world_size = dist.get_world_size
original_get_rank = dist.get_rank
original_all_gather = dist.all_gather

def mock_get_world_size(pg=None):
    return 1

def mock_get_rank(pg=None):
    return 0

def mock_all_gather(tensor_list, input_, group=None):
    # For world_size=1, just copy the input
    if len(tensor_list) > 0:
        tensor_list[0].copy_(input_)

dist.get_world_size = mock_get_world_size
dist.get_rank = mock_get_rank
dist.all_gather = mock_all_gather

# Create input tensor - small, CPU-only
input_tensor = torch.randn(2, 4, requires_grad=True)

# Define arguments
ARGS = {
    'input_': input_tensor,
    'process_group': mock_pg,
    'dim': -1,
    'grad_scale': 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 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
num_gpus_to_mem Ran
ran; our contract draft was wrong, not the code
universome/stylegan-v/src/infra/utils.py
pointer only (licence: NONE) · get_code("f29fb27caa548814")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, universome/stylegan-v when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

shift_dim Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("791774067a83bf99")
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, 5) 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 small tensor with distinct dimensions for probing
# Using shape (2, 3, 4, 5) to have distinct sizes for each dimension
x = torch.randn(2, 3, 4, 5)

ARGS = {
    'x': x,
    'src_dim': -1,
    'dest_dim': -1,
    'make_contiguous': True
}
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, 5) — shape (2, 3, 4, 5) 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
to_dict Ran
ran; our contract draft was wrong, not the code
universome/stylegan-v/src/infra/slurm_batch_launch.py
pointer only (licence: NONE) · get_code("90b1bec775ebe652")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, universome/stylegan-v when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Net2NetTransformer Not yet run
unverified (harvested, no recorded run)
songweige/TATS/tats/tats_transformer.py
code served (permissive licence) · get_code("af840b748b65873d")
How it ran

no invocation recorded.

all_to_all Not yet run
unverified (harvested, no recorded run)
dawn-lx/causal-videogen/opensora/acceleration/communications.py
code served (permissive licence) · get_code("b51f59051d16f7df")
How it ran

no invocation recorded.

auto_grad_checkpoint Not yet run
unverified (harvested, no recorded run)
dawn-lx/causal-videogen/opensora/acceleration/checkpoint.py
code served (permissive licence) · get_code("20336f4981288332")
How it ran

no invocation recorded.

compute_fvd Not yet run
unverified (harvested, no recorded run)
universome/stylegan-v/src/metrics/frechet_video_distance.py
pointer only (licence: NONE) · get_code("0bf504441d3b1c0e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, universome/stylegan-v when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

partial_causal_flash_attn Not yet run
unverified (harvested, no recorded run)
dawn-lx/causal-videogen/opensora/models/causal_stdit2/attention.py
code served (permissive licence) · get_code("5cbd445824659be5")
How it ran

no invocation recorded.

split_forward_gather_backward Not yet run
unverified (harvested, no recorded run)
dawn-lx/causal-videogen/opensora/acceleration/communications.py
code served (permissive licence) · get_code("a771d32bab25e50b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 21 samples lifted from this paper's linked repositories, 21 carry a function identity and 0 do not. Of the 21 distinct functions with an identity, 6 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 later) and 15 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
accuracy
brick of this paper's assembly: helper
13 other papers
13 dated earlier
1706.02677 (2017-06)
Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
1911.00068 (2019-11)
Confident Learning: Estimating Uncertainty in Dataset Labels
2011.09157 (2020-11)
Dense Contrastive Learning for Self-Supervised Visual Pre-Training
and 10 more not named here
shift_dim
brick of this paper's assembly: helper
2 other papers
2 dated earlier
2402.19470 (2024-02)
Towards Generalizable Tumor Synthesis
2404.12391 (2024-04)
On the Content Bias in Fréchet Video Distance
all_to_all 1 other paper
1 later
2503.09642 (2025-03)
Open-Sora 2.0: Training a Commercial-Level Video Generation Model in $200k
auto_grad_checkpoint 1 other paper
1 dated earlier
2405.14979 (2024-05)
CraftsMan3D: High-fidelity Mesh Generation with 3D Native Generation and Interactive Geometry Refiner
gather_forward_split_backward 1 other paper
1 later
2503.09642 (2025-03)
Open-Sora 2.0: Training a Commercial-Level Video Generation Model in $200k
split_forward_gather_backward 1 other paper
1 later
2503.09642 (2025-03)
Open-Sora 2.0: Training a Commercial-Level Video Generation Model in $200k

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

With the advance of diffusion models, today's video generation has achieved impressive quality. To extend the generation length and facilitate real-world applications, a majority of video diffusion models (VDMs) generate videos in an autoregressive manner, i.e., generating subsequent clips conditioned on the last frame(s) of the previous clip. However, existing autoregressive VDMs are highly inefficient and redundant: The model must re-compute all the conditional frames that are overlapped between adjacent clips. This issue is exacerbated when the conditional frames are extended autoregressively to provide the model with long-term context. In such cases, the computational demands increase significantly (i.e., with a quadratic complexity w.r.t. the autoregression step). In this paper, we propose Ca2-VDM, an efficient autoregressive VDM with Causal generation and Cache sharing. For causal generation, it introduces unidirectional feature computation, which ensures that the cache of conditional frames can be precomputed in previous autoregression steps and reused in every subsequent step, eliminating redundant computations. For cache sharing, it shares the cache across all denoising steps to avoid the huge cache storage cost. Extensive experiments demonstrated that our Ca2-VDM achieves state-of-the-art quantitative and qualitative video generation results and significantly improves the generation speed. Code is available: https://github.com/ Dawn-LX/CausalCache-VDM

For agents

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

get_harvested_code_for_paper("2411.16375")
get_code_for_paper("2411.16375")
have("2411.16375")

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

curl https://syntology.ai/api/ran/2411.16375.json

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

[![Syntology run record](https://syntology.ai/api/ran/2411.16375.svg)](https://syntology.ai/paper/2411.16375)

Connect an agent — have() is free.