SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2501.19252 · 2025

Inference-Time Text-to-Video Alignment with Diffusion Latent Beam Search

arXiv · PDF · Open in the Atlas

Code that ran

9 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; 3 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 5 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/2501.19252.json · badge: /api/ran/2501.19252.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
shim0114/T2V-Diffusion-Search community (classifier label)
classifier basis not recorded on the link
9 ran · 1 no recorded run
FunctionStatusWhere it lives
approx_standard_normal_cdf Ran
ran; our contract draft was wrong, not the code
shim0114/T2V-Diffusion-Search/diffusion/diffusion_utils.py
code served (permissive licence) · get_code("d6a68e210556f857")
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 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch as th
import numpy as np

th.manual_seed(0)

# The function takes a single parameter 'x' which should be a torch tensor
# It performs mathematical operations on it: tanh, pow, addition, multiplication
# So x should be a tensor of any shape

x = th.randn(2, 4)

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
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 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
continuous_gaussian_log_likelihood Ran
ran; our contract draft was wrong, not the code
shim0114/T2V-Diffusion-Search/diffusion/diffusion_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from rerv/vdt, is pointer only (licence: NOASSERTION) · get_code("ab1c9568b4e13899")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from rerv/vdt, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

create_named_schedule_sampler Ran
ran; our contract draft was wrong, not the code
shim0114/T2V-Diffusion-Search/diffusion/timestep_sampler.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from vainf/remix-dit, is pointer only (licence: NOASSERTION) · get_code("e48218d7d73db0b3")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from vainf/remix-dit, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

dd_mapping_func Ran
ran on a synthesized input
shim0114/T2V-Diffusion-Search/CogVideoX/my_reward_model.py
code served (permissive licence) · get_code("e6adc74db7935f1a")
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 ndarray, ndim 1, dtype family float, shape (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 numpy as np

def dd_mapping_func(a):
    return np.log(a) / 16.0

# Create a small numpy array as input
a = np.array([1.0, 2.0, 3.0, 4.0, 5.0])

ARGS = {
    "a": a
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (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
get_beta_schedule Ran
ran, honoured the contract we drafted
shim0114/T2V-Diffusion-Search/diffusion/gaussian_diffusion.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from xuyanwu/SIDDMs-UFOGen, is pointer only (licence: NONE) · get_code("3e0fa4efc22272d4")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from xuyanwu/SIDDMs-UFOGen, and GitHub detected no licence in that repository 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.

get_named_beta_schedule Ran
ran, honoured the contract we drafted
shim0114/T2V-Diffusion-Search/diffusion/gaussian_diffusion.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from xuyanwu/SIDDMs-UFOGen, is pointer only (licence: NONE) · get_code("36e30c7fb679ec78")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from xuyanwu/SIDDMs-UFOGen, and GitHub detected no licence in that repository 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.

mean_flat Ran
ran; our contract draft was wrong, not the code
shim0114/T2V-Diffusion-Search/diffusion/gaussian_diffusion.py
code served (permissive licence) · get_code("f6d7c009a8efb8b7")
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 1, dtype family float, shape (2,) 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)

# The function takes a tensor and computes mean over all non-batch dimensions
# It expects a tensor with shape (batch_size, *other_dims)
# and returns a tensor of shape (batch_size,)

# Create a small tensor with batch dimension and other dimensions
# Using distinct sizes: batch=2, dim1=4, dim2=3
tensor = torch.randn(2, 4, 3)

ARGS = {
    "tensor": tensor
}
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 1, dtype family float, shape (2,) — shape (2,) 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
normal_kl Ran
ran, honoured the contract we drafted
shim0114/T2V-Diffusion-Search/diffusion/diffusion_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from rerv/vdt, is pointer only (licence: NOASSERTION) · get_code("8afbfc42c6ea0448")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from rerv/vdt, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

space_timesteps Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
shim0114/T2V-Diffusion-Search/diffusion/respace.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from ianyeung/mgld-vsr, is pointer only (licence: NOASSERTION) · get_code("ea9dbc131adf582e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from ianyeung/mgld-vsr, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

prompt_enc Not yet run
unverified (harvested, no recorded run)
shim0114/T2V-Diffusion-Search/Wan2.1/gradio/t2v_1.3B_singleGPU.py
code served (permissive licence) · get_code("9ab1110bfcad5f91")
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, 8 are also present in other papers whose code we hold (8 with at least one paper dated earlier, 0 the same month, 8 later) and 2 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
normal_kl 160 other papers
139 dated earlier, 19 later, 2 same year, order not determinable · 19 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 157 more not named here
approx_standard_normal_cdf 154 other papers
133 dated earlier, 19 later, 2 same year, order not determinable · 17 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 151 more not named here
mean_flat 60 other papers
49 dated earlier, 11 later · 25 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2201.06503 (2022-01)
Analytic-DPM: an Analytic Estimate of the Optimal Reverse Variance in Diffusion Probabilistic Models
and 57 more not named here
space_timesteps 57 other papers
51 dated earlier, 6 later · 2 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
and 54 more not named here
create_named_schedule_sampler 45 other papers
40 dated earlier, 5 later
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
and 42 more not named here
get_beta_schedule 31 other papers
25 dated earlier, 6 later
2202.09671 (2022-02)
TRUNCATED DIFFUSION PROBABILISTIC MODELS AND DIFFUSION-BASED ADVERSARIAL AUTO-ENCODERS
2210.07574 (2022-10)
Is synthetic data from generative models ready for image recognition?
2306.02562 (2023-06)
Video Diffusion Models with Local-Global Context Guidance
and 28 more not named here
continuous_gaussian_log_likelihood 26 other papers
23 dated earlier, 3 later
2305.13311 (2023-05)
VDT: General-purpose Video Diffusion Transformers via Mask Modeling
2306.14770 (2023-06)
ProtoDiff: Learning to Learn Prototypical Networks by Task-Guided Diffusion
2312.02139 (2023-12)
DiffiT: Diffusion Vision Transformers for Image Generation
and 23 more not named here
get_named_beta_schedule 25 other papers
21 dated earlier, 4 later
2210.07574 (2022-10)
Is synthetic data from generative models ready for image recognition?
2306.02562 (2023-06)
Video Diffusion Models with Local-Global Context Guidance
2306.12511 (2023-06)
Semi-Implicit Denoising Diffusion Models (SIDDMs)
and 22 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The remarkable progress in text-to-video diffusion models enables the generation of photorealistic videos, although the content of these generated videos often includes unnatural movement or deformation, reverse playback, and motionless scenes. Recently, an alignment problem has attracted huge attention, where we steer the output of diffusion models based on some measure of the content's goodness. Because there is a large room for improvement of perceptual quality along the frame direction, we should address which metrics we should optimize and how we can optimize them in the video generation. In this paper, we propose diffusion latent beam search with lookahead estimator, which can select a better diffusion latent to maximize a given alignment reward at inference time. We then point out that improving perceptual video quality with respect to alignment to prompts requires reward calibration by weighting existing metrics. This is because when humans or vision language models evaluate outputs, many previous metrics to quantify the naturalness of video do not always correlate with the evaluation. We demonstrate that our method improves the perceptual quality evaluated on the calibrated reward, VLMs, and human assessment, without model parameter update, and outputs the best generation compared to greedy search and best-of-N sampling under much more efficient computational cost. The experiments highlight that our method is beneficial to many capable generative models, and provide a practical guideline: we should prioritize the inference-time compute allocation into enabling the lookahead estimator and increasing the search budget, rather than expanding the denoising steps.

For agents

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

get_harvested_code_for_paper("2501.19252")
get_code_for_paper("2501.19252")
have("2501.19252")

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

curl https://syntology.ai/api/ran/2501.19252.json

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

[![Syntology run record](https://syntology.ai/api/ran/2501.19252.svg)](https://syntology.ai/paper/2501.19252)

Connect an agent — have() is free.