SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.13347 · CVPR · 2024

vid-TLDR: Training Free Token merging for Light-weight Video Transformer

Hyunwoo Kim, Sanghyeok Lee, Joonmyung Choi, Jaewon Chu, Minhyuk Choi

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 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/2403.13347.json · badge: /api/ran/2403.13347.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
mlvlab/vid-tldr marked official in Papers with Code
Papers with Code archive
3 ran · 1 no recorded run
FunctionStatusWhere it lives
get_objective_score Ran
ran; our contract draft was wrong, not the code
mlvlab/vid-tldr/multi_modality/utils/vid_TLDR.py
code served (permissive licence) · get_code("23e1c4ddde84286f")
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 3, dtype family float, shape (2, 4, 1) 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)

# score_attn is used with:
# - .mean(dim=1)
# - result has .sum(dim=2) and .unsqueeze(-1)
# - result has .shape unpacked to B, T_R, _
# - result has .amin(dim=1, keepdim=True)
# - result has .amax(dim=1, keepdim=True)
# - result has .mean(dim=1, keepdim=True)
# - result supports indexing with boolean mask

# After mean(dim=1), shape should be (B, T_R, feature_dim)
# Before mean(dim=1), shape is (B, num_heads, T_R, feature_dim)
# Let's use: B=2, num_heads=3, T_R=4, feature_dim=5

score_attn = torch.randn(2, 3, 4, 5)

ARGS = {
    "score_attn": score_attn,
}
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 3, dtype family float, shape (2, 4, 1) — shape (2, 4, 1) 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
merging Ran
ran; our contract draft was wrong, not the code
mlvlab/vid-tldr/multi_modality/utils/vid_TLDR.py
code served (permissive licence) · get_code("35ee8782a6d238a8")
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 function. 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)

# metric: torch.Tensor
# The function expects metric to have shape (..., 2*t1, d) where it splits on dim -2
# We need an even number of tokens in dim -2
# Let's use shape (2, 2, 8, 4) - batch=2, seq=2, tokens=8 (even), embedding_dim=4
metric = torch.randn(2, 2, 8, 4)

# r_merge: int
# This controls how many tokens to merge. Must be < t1 (which is 4 in our case)
r_merge = 2

# score_obj: torch.Tensor or None
# Optional tensor with same structure as metric for weighting
# Shape should match metric: (2, 2, 8, 4)
score_obj = torch.randn(2, 2, 8, 4)

ARGS = {
    "metric": metric,
    "r_merge": r_merge,
    "score_obj": score_obj,
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
vidTLDR Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mlvlab/vid-tldr/multi_modality/utils/vid_TLDR.py
code served (permissive licence) · get_code("d9d992feb83108d2")
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 (2, 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)

# Build minimal arguments for vidTLDR function
B = 2  # batch size
T = 8  # sequence length
D = 4  # feature dimension
H = 3  # number of attention heads

# x: input tensor (B, T, D)
x = torch.randn(B, T, D)

# attn: attention tensor used in get_objective_score
# The function does: score_attn.mean(dim=1) then operations on dim=2
# So attn should be (B, H, T, T) to match typical attention shape
attn = torch.randn(B, H, T, T)
# Ensure attention values are positive for log operation
attn = torch.softmax(attn, dim=-1)

# info: dictionary with required keys
info = {
    'use': True,
    'r': {0: 0.5},  # layer 0 has reduction ratio 0.5
    'source_trace': False,
    'source': None,
    'size': torch.ones(B, T, 1),  # size tensor (B, T, 1)
}

# layer: integer index
layer = 0

ARGS = {
    'x': x,
    'attn': attn,
    'info': info,
    'layer': layer,
}
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 (2, 4, 4) — shape (2, 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
merge_wavg Not yet run
unverified (harvested, no recorded run)
mlvlab/vid-tldr/multi_modality/utils/vid_TLDR.py
code served (permissive licence) · get_code("5c2feb7328e8ea26")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

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

Repositories linked to this paper

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

Abstract

Figure 1. Comparison of vid-TLDR (Ours) with UMT [33]. Without any additional training, vid-TLDR obtains comparable or even better performance than the base model UMT (left) while reducing the considerable computational cost (right). UMT-B (87M) is used.

For agents

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

get_harvested_code_for_paper("2403.13347")
get_code_for_paper("2403.13347")
have("2403.13347")

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

curl https://syntology.ai/api/ran/2403.13347.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.13347.svg)](https://syntology.ai/paper/2403.13347)

Connect an agent — have() is free.