SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.04402 · 2019

Polysemous Visual-Semantic Embedding for Cross-Modal Retrieval

arXiv · PDF · Open in the Atlas

Code that ran

12 of 12 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 12 have no contract verdict; 0 failures of our instrument; 0 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/1906.04402.json · badge: /api/ran/1906.04402.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
yalesong/pvse community (classifier label)
classifier basis not recorded on the link
12 ran · 0 no recorded run
FunctionStatusWhere it lives
collate_fn Ran
ran on a synthesized input
yalesong/pvse/data.py
code served (permissive licence) · get_code("067569ab2ca0cb91")
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. 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)

# The function expects data to be a list of tuples
# Each tuple contains: (image, sentence, id, img_id)
# Based on the function body:
# - image: torch tensor of shape (3, 256, 256)
# - sentence: torch tensor of shape (variable_length,) with integer tokens
# - id: some identifier
# - img_id: some image identifier

# Create sample data
image1 = torch.randn(3, 256, 256)
image2 = torch.randn(3, 256, 256)

sentence1 = torch.tensor([1, 2, 3, 4, 5])
sentence2 = torch.tensor([6, 7, 8])

id1 = 0
id2 = 1

img_id1 = "img_0"
img_id2 = "img_1"

data = [
    (image1, sentence1, id1, img_id1),
    (image2, sentence2, id2, img_id2),
]

ARGS = {
    "data": data
}
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 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: none by the harness; any seed is in the setup block
cosine_sim Ran
ran on a synthesized input
yalesong/pvse/loss.py
code served (permissive licence) · get_code("e42286bdac3eb647")
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, 3) 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
import torch.nn as nn

torch.manual_seed(0)

# The function expects x and y to be 2D tensors (matrices) that support .mm() and .t() operations
# x.mm(y.t()) performs matrix multiplication: (batch_x, dim) @ (dim, batch_y).T = (batch_x, batch_y)
# Both tensors should be L2 normalized according to the docstring

# Create two L2-normalized tensors
# x: (2, 4) - represents 2 images in 4-dimensional space
x = torch.randn(2, 4)
x = torch.nn.functional.normalize(x, p=2, dim=1)

# y: (3, 4) - represents 3 sentences in 4-dimensional space
y = torch.randn(3, 4)
y = torch.nn.functional.normalize(y, p=2, dim=1)

ARGS = {
    "x": x,
    "y": y,
}
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, 3) — shape (2, 3) 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
from_mrw_json Ran
ran on a synthesized input
yalesong/pvse/vocab.py
code served (permissive licence) · get_code("bea183ae94fb8b9d")
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 list, ndim 1, 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 json
import tempfile
import os

# Create a temporary JSON file with the expected structure
temp_dir = tempfile.mkdtemp()
json_path = os.path.join(temp_dir, "test_data.json")

# Create sample data matching what the function expects
sample_data = [
    {
        "sentence": "This is a test/r/ caption with r/ some /u/ noise u/ and.. special-characters"
    },
    {
        "sentence": "Another/r/ example-sentence/u/ with replacements"
    }
]

with open(json_path, 'w') as f:
    json.dump(sample_data, f)

ARGS = {
    "path": json_path
}
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 list, ndim 1, 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
from_tgif_tsv Ran
ran on a synthesized input
yalesong/pvse/vocab.py
code served (permissive licence) · get_code("fd9b502b79464525")
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 list, ndim 1, 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 tempfile
import os

# Create a temporary TSV file with the expected format
# The function expects lines with at least 2 tab-separated fields,
# where the second field (index 1) is the caption
temp_dir = tempfile.mkdtemp()
test_file = os.path.join(temp_dir, "test.tsv")

with open(test_file, 'w') as f:
    f.write("id1\tcaption one\textra_field\n")
    f.write("id2\tcaption two\textra_field\n")
    f.write("id3\tcaption three\textra_field\n")

ARGS = {
    'path': test_file
}
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 list, ndim 1, 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
get_cnn Ran
ran on a synthesized input
yalesong/pvse/model.py
code served (permissive licence) · get_code("2a091499b862ef78")
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 ResNet. 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 torchvision
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "arch": "resnet18",
    "pretrained": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResNet — a ResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ResNet 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_pad_mask Ran
ran on a synthesized input
yalesong/pvse/model.py
code served (permissive licence) · get_code("38b1696acd2ad684")
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, 5) 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
from torch.autograd import Variable

torch.manual_seed(0)

# Create tiny tensors for testing
# max_length: scalar or int representing maximum sequence length
max_length = 5

# lengths: 1D tensor of actual sequence lengths for each item in batch
# Using batch size 2 with different lengths
lengths = torch.tensor([3, 4], dtype=torch.long)

# set_pad_to_one: boolean flag
set_pad_to_one = True

ARGS = {
    "max_length": max_length,
    "lengths": lengths,
    "set_pad_to_one": set_pad_to_one,
}
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, 5) — shape (2, 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_paths Ran
ran on a synthesized input
yalesong/pvse/data.py
code served (permissive licence) · get_code("9919713948f52bff")
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 1, 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 os
import tempfile
import numpy as np

# Create a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()

# Create subdirectories for COCO dataset
os.makedirs(os.path.join(temp_dir, 'images', 'train2014'), exist_ok=True)
os.makedirs(os.path.join(temp_dir, 'images', 'val2014'), exist_ok=True)
os.makedirs(os.path.join(temp_dir, 'annotations'), exist_ok=True)

# Create dummy numpy files for COCO
np.save(os.path.join(temp_dir, 'annotations', 'coco_train_ids.npy'), np.array([1, 2, 3, 4, 5]))
np.save(os.path.join(temp_dir, 'annotations', 'coco_dev_ids.npy'), np.array([6, 7, 8, 9, 10]))
np.save(os.path.join(temp_dir, 'annotations', 'coco_test_ids.npy'), np.array([11, 12, 13, 14, 15]))
np.save(os.path.join(temp_dir, 'annotations', 'coco_restval_ids.npy'), np.array([16, 17, 18, 19, 20]))

# Create dummy annotation files
with open(os.path.join(temp_dir, 'annotations', 'captions_train2014.json'), 'w') as f:
    f.write('{}')
with open(os.path.join(temp_dir, 'annotations', 'captions_val2014.json'), 'w') as f:
    f.write('{}')

ARGS = {
    'path': temp_dir,
    'name': 'coco',
    'use_restval': True
}
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 tuple, ndim 1, 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
get_uid_tgif Ran
ran on a synthesized input
yalesong/pvse/data.py
code served (permissive licence) · get_code("41c61aa3138d76aa")
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 numpy as np

torch.manual_seed(0)

def get_uid_tgif(url):
    return url.strip().split('/')[-1].replace('.gif', '')

# The function takes a url string, strips whitespace, splits by '/',
# takes the last element, and removes '.gif' extension
# We need to provide a valid URL string as argument

ARGS = {
    'url': 'https://media.giphy.com/media/example123.gif'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, 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
l2norm Ran
ran on a synthesized input
yalesong/pvse/loss.py
code served (permissive licence) · get_code("7db07517e868f2bd")
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. 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 l2norm takes a tensor x and L2-normalizes its columns
# It performs operations: pow, sum, sqrt, div
# These are all standard tensor operations that work on any shape

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

ARGS = {
    "x": x
}
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
l2norm Ran
ran on a synthesized input
yalesong/pvse/model.py
code served (permissive licence) · get_code("0e45ef681f2e9391")
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. 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.nn.init

torch.manual_seed(0)

# The function l2norm takes a tensor x and L2-normalizes its columns
# It performs operations: pow, sum, sqrt, div on the tensor
# The tensor should have at least 2 dimensions (it uses dim=-1 for sum)

# Create a small tensor with shape (2, 4)
# This allows us to test the L2 normalization across columns
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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
order_sim Ran
ran on a synthesized input
yalesong/pvse/loss.py
code served (permissive licence) · get_code("27932a44f7af1fe7")
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 (3, 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
import torch.nn as nn

torch.manual_seed(0)

# The function expects two tensors: x and y
# From the body, we can see:
# - y.unsqueeze(1).expand(y.size(0), x.size(0), y.size(1))
# - x.unsqueeze(0).expand(y.size(0), x.size(0), y.size(1))
# This means x and y should be 2D tensors where:
# - x has shape (x.size(0), embedding_dim)
# - y has shape (y.size(0), embedding_dim)
# The embedding dimensions must match

# Create small sample tensors
x = torch.randn(3, 4)  # 3 embeddings of dimension 4
y = torch.randn(2, 4)  # 2 embeddings of dimension 4

ARGS = {
    'x': x,
    'y': y
}
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 (3, 2) — shape (3, 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
verify_input_args Ran
ran on a synthesized input
yalesong/pvse/option.py
code served (permissive licence) · get_code("1a7561ed4bd4cf47")
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 Args. 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
import sys
import tempfile
from pathlib import Path

class Args:
    def __init__(self):
        # Create a temporary directory for logger_name
        self.temp_dir = tempfile.mkdtemp()
        self.log_dir = tempfile.mkdtemp()
        
        # Create a temporary checkpoint file
        self.temp_ckpt = tempfile.NamedTemporaryFile(delete=False, suffix='.pt')
        self.temp_ckpt.close()
        
        self.ckpt = self.temp_ckpt.name
        self.num_embeds = 1
        self.img_attention = True
        self.txt_attention = True
        self.val_metric = 'rsum'
        self.logger_name = self.temp_dir
        self.log_file = os.path.join(self.log_dir, 'log.txt')

args_instance = Args()

ARGS = {
    'args': args_instance
}
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 Args — a Args output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Args 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

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 11 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 11 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

Visual-semantic embedding aims to find a shared latent space where related visual and textual instances are close to each other. Most current methods learn injective embedding functions that map an instance to a single point in the shared space. Unfortunately, injective embedding cannot effectively handle polysemous instances with multiple possible meanings; at best, it would find an average representation of different meanings. This hinders its use in real-world scenarios where individual instances and their cross-modal associations are often ambiguous. In this work, we introduce Polysemous Instance Embedding Networks (PIE-Nets) that compute multiple and diverse representations of an instance by combining global context with locally-guided features via multi-head self-attention and residual learning. To learn visual-semantic embedding, we tie-up two PIE-Nets and optimize them jointly in the multiple instance learning framework. Most existing work on cross-modal retrieval focuses on image-text data. Here, we also tackle a more challenging case of video-text retrieval. To facilitate further research in video-text retrieval, we release a new dataset of 50K video-sentence pairs collected from social media, dubbed MRW (my reaction when). We demonstrate our approach on both image-text and video-text retrieval scenarios using MS-COCO, TGIF, and our new MRW dataset.

For agents

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

get_harvested_code_for_paper("1906.04402")
get_code_for_paper("1906.04402")
have("1906.04402")

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

curl https://syntology.ai/api/ran/1906.04402.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.04402.svg)](https://syntology.ai/paper/1906.04402)

Connect an agent — have() is free.