SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2304.06708 · ICCV · 2023

Verbs in Action: Improving verb understanding in video-language models

Andrew Zisserman, Mathilde Caron, Cordelia Schmid, Arsha Nagrani, Liliane Momeni

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure 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/2304.06708.json · badge: /api/ran/2304.06708.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
google-research/scenic marked official in Papers with Code
Papers with Code archive
2 ran · 1 no recorded run
FunctionStatusWhere it lives
get_contrastive_labels Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
google-research/scenic/scenic/projects/verbs_in_action/losses.py
code served (permissive licence) · get_code("c69cabcefea1515a")
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, 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 jax.numpy as jnp
import numpy as np

# Set seed for reproducibility
np.random.seed(0)

# Parameters
batch_size = 2
max_num_captions = 3

# Create inners: shape (batch_size, batch_size * max_num_captions)
# This represents inner products between videos and captions
inners = jnp.array([
    [0.9, 0.1, 0.2, 0.3, 0.4, 0.5],
    [0.1, 0.8, 0.3, 0.2, 0.5, 0.4]
], dtype=jnp.float32)

# Create text_mask: shape (batch_size * max_num_captions, 1)
# 1 indicates padding, 0 indicates valid caption
text_mask = jnp.array([
    [0.0],
    [0.0],
    [1.0],
    [0.0],
    [0.0],
    [1.0]
], dtype=jnp.float32)

# Test with loss_exclude_hn_other_vid=True
loss_exclude_hn_other_vid = True

ARGS = {
    "inners": inners,
    "text_mask": text_mask,
    "loss_exclude_hn_other_vid": loss_exclude_hn_other_vid
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, dtype family float — a tuple 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
hn_nce_loss_with_hardnegs Ran
ran, honoured the contract we drafted
google-research/scenic/scenic/projects/verbs_in_action/losses.py
code served (permissive licence) · get_code("565678aafb6bf2e9")
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 ArrayImpl, 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 jax.numpy as jnp
import numpy as np

# Construct arguments for hn_nce_loss_with_hardnegs
batch_size = 2
num_classes = 4

# logits: shape (batch_size, num_classes)
# These are raw scores before softmax
logits = jnp.array([
    [0.1, 0.5, 0.2, 0.3],
    [0.4, 0.1, 0.6, 0.2]
], dtype=jnp.float32)

# labels: shape (batch_size, num_classes)
# One-hot encoded labels where 1 indicates the positive class
labels = jnp.array([
    [1.0, 0.0, 0.0, 0.0],
    [0.0, 1.0, 0.0, 0.0]
], dtype=jnp.float32)

# n: number of negatives to consider (optional)
n = 3

# beta: temperature parameter for weighting hard negatives
beta = 0.5

# masking: optional masking array, same shape as logits
# Use -1000000 to mask out positions, any other value to keep them
masking = jnp.array([
    [-1000000, 0.0, 0.0, 0.0],
    [0.0, -1000000, 0.0, 0.0]
], dtype=jnp.float32)

ARGS = {
    "logits": logits,
    "labels": labels,
    "n": n,
    "beta": beta,
    "masking": masking
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, 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
verb_hard_neg_nce Not yet run
unverified (harvested, no recorded run)
google-research/scenic/scenic/projects/verbs_in_action/losses.py
code served (permissive licence) · get_code("f9dbdd085cd8749d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

two brown horses eating grass LLM Original caption Video Add prompt two brown horses running on the grass two brown horses fighting on the grass two brown horses lying on the grass eating grass Verb phrase Hard verb negative captions two brown horses sleeping on the grass two brown horses playing on the grass eating grass cleaning camera standing squatting Verb phrase loss two brown horses eating grass two brown horses lying on the grass a woman standing in a post office person squatting at the gym a man cleaning his camera Hard negative loss Generated hard negative caption Negative captions in the batch Negative verb phrases in the batch Batch Figure 1. Verb-Focused Contrastive (VFC) learning: (Left): Given a video and its corresponding caption, we leverage a Large Language Model (LLM) to output (1) hard negative captions, where only the verb has been changed while keeping the remaining context, and (2) verb phrases which succinctly describe the action in the video. (Right): To encourage better verb reasoning, we subsequently enforce (1) a calibrated hard negative loss, using our generated hard negative captions and other captions in the batch, and (2) a fine-grained, verb phrase loss. We show that VFC improves verb understanding of video-language models compared to the standard contrastive loss.

For agents

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

get_harvested_code_for_paper("2304.06708")
get_code_for_paper("2304.06708")
have("2304.06708")

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

curl https://syntology.ai/api/ran/2304.06708.json

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

[![Syntology run record](https://syntology.ai/api/ran/2304.06708.svg)](https://syntology.ai/paper/2304.06708)

Connect an agent — have() is free.