SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.07063 · NeurIPS · 2023

Bootstrapping Vision-Language Learning with Decoupled Language Pre-training

Chongyang Gao, Soroush Vosoughi, Yiren Jian

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 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; 6 failures of our instrument; 2 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/2307.07063.json · badge: /api/ran/2307.07063.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
yiren-jian/BLIText canonical (classifier label)
classifier basis not recorded on the link
7 ran · 2 no recorded run
FunctionStatusWhere it lives
generate_caption Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yiren-jian/BLIText/app/caption.py
code served (permissive licence) · get_code("56f02812d66a0d11")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class MockCaptionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 8)
    
    def generate(self, samples, use_nucleus_sampling=False, num_beams=3, max_length=40, min_length=5, top_p=0.9):
        # Mock generate method that returns a list of captions
        # Each caption is a string
        batch_size = samples['image'].shape[0]
        captions = ["a cat on a mat"] * batch_size
        return captions

# Create a small image tensor (batch_size=2, channels=3, height=4, width=4)
image = torch.randn(2, 3, 4, 4)

# Create the model
model = MockCaptionModel()

ARGS = {
    'model': model,
    'image': image,
    'use_nucleus_sampling': False,
    'num_beams': 3,
    'max_length': 40,
    'min_length': 5
}
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 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_concat_v Ran
ran; our contract draft was wrong, not the code
yiren-jian/BLIText/app/dataset_browser.py
code served (permissive licence) · get_code("92f2e16ee0d3a24d")
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 Image. 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.

from PIL import Image

# Create two small PIL images for testing
im1 = Image.new('RGB', (100, 50), color='Red')
im2 = Image.new('RGB', (80, 60), color='Blue')

ARGS = {
    'im1': im1,
    'im2': im2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int 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
read_img Ran
ran, honoured the contract we drafted
yiren-jian/BLIText/app/calculate_coco_features.py
code served (permissive licence) · get_code("13a43a7d815937e6")
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 Image. 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 tempfile
import os
from PIL import Image

# Create a temporary image file for testing
temp_dir = tempfile.mkdtemp()
temp_image_path = os.path.join(temp_dir, "test_image.png")

# Create a simple RGB image and save it
test_image = Image.new('RGB', (4, 3), color=(255, 0, 0))
test_image.save(temp_image_path)

ARGS = {
    'filepath': temp_image_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int 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
resize_img Ran
ran; our contract draft was wrong, not the code
yiren-jian/BLIText/app/utils.py
code served (permissive licence) · get_code("926bb38f979a61ef")
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 Image. 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.

from PIL import Image
import torch

torch.manual_seed(0)

# Create a small PIL Image object
# The function calls .size on it and .resize()
# We need an image with width > 240 so the scaling factor makes sense
raw_img = Image.new('RGB', (480, 360))

ARGS = {
    'raw_img': raw_img
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int 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
resize_img_w Ran
ran; our contract draft was wrong, not the code
yiren-jian/BLIText/app/dataset_browser.py
code served (permissive licence) · get_code("f809f2ecf34c1c99")
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 Image. 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.

from PIL import Image

# Create a small test image
def create_test_image(width=100, height=80):
    return Image.new('RGB', (width, height), color='red')

# Test case 1: single image
single_image = create_test_image(100, 80)

# Test case 2: list of images
image_list = [create_test_image(100, 80), create_test_image(120, 90), create_test_image(110, 85)]

ARGS = {
    "raw_img": single_image,
    "new_w": 224
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int 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
sample_dataset Ran
ran; our contract draft was wrong, not the code
yiren-jian/BLIText/app/dataset_browser.py
code served (permissive licence) · get_code("3d4bbb5ca53fd24a")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class MockDataset:
    def __init__(self):
        self.data = {
            0: {"id": 0, "value": torch.randn(4)},
            1: {"id": 1, "value": torch.randn(4)},
            2: {"id": 2, "value": torch.randn(4)},
        }
    
    def displ_item(self, idx):
        return self.data[idx]

dataset = MockDataset()
indices = [0, 1, 2]

ARGS = {
    "dataset": dataset,
    "indices": indices,
}
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 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
tile Ran
ran; our contract draft was wrong, not the code
yiren-jian/BLIText/lavis/models/base_model.py
pointer only (licence: BSD-3-Clause); the stored copy get_code serves, from MengzhouLu/RetroMPA, is pointer only (licence: NONE) · get_code("cb33571427334815")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (BSD-3-Clause), but get_code serves the stored copy, first harvested from MengzhouLu/RetroMPA, 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.

all_gather_with_grad Not yet run
unverified (harvested, no recorded run)
yiren-jian/BLIText/lavis/models/base_model.py
code served (permissive licence) · get_code("0ec9fc2025c16f65")
How it ran

no invocation recorded.

compute_gradcam_batch Not yet run
unverified (harvested, no recorded run)
yiren-jian/BLIText/app/multimodal_search.py
code served (permissive licence) · get_code("3e12ea8c09674aa8")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 9 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 2 the same month, 9 later) and 0 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
all_gather_with_grad 91 other papers
8 dated earlier, 3 the same month, 80 later · 2 only as an assembly brick
2206.02967 (2022-06)
Masked Unsupervised Self-training for Label-free Image Classification
2210.08773 (2022-10)
Plug-and-Play VQA: Zero-shot VQA by Conjoining Large Pretrained Models with Zero Training
2302.06605 (2023-02)
UniAdapter: Unified Parameter-Efficient Transfer Learning for Cross-modal Modeling
and 88 more not named here
tile 84 other papers
11 dated earlier, 3 the same month, 70 later
2206.07699 (2022-06)
Write and Paint: Generative Vision-Language Models are Unified Modal Learners
2210.07795 (2022-10)
EfficientVLM: Fast and Accurate Vision-Language Models via Knowledge Distillation and Modal-adaptive Pruning
2210.08773 (2022-10)
Plug-and-Play VQA: Zero-shot VQA by Conjoining Large Pretrained Models with Zero Training
and 81 more not named here
generate_caption 12 other papers
2 dated earlier, 10 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 9 more not named here
read_img 12 other papers
2 dated earlier, 10 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 9 more not named here
compute_gradcam_batch 10 other papers
2 dated earlier, 8 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 7 more not named here
get_concat_v 10 other papers
2 dated earlier, 8 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 7 more not named here
resize_img_w 10 other papers
2 dated earlier, 8 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 7 more not named here
sample_dataset 10 other papers
2 dated earlier, 8 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 7 more not named here
resize_img 9 other papers
2 dated earlier, 7 later
2305.01278 (2023-05)
VPGTrans: Transfer Visual Prompt Generator across LLMs
2305.06988 (2023-05)
Self-Chained Image-Language Model for Video Localization and Question Answering
2310.17419 (2023-10)
AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors
and 6 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

We present a novel methodology aimed at optimizing the application of frozen large language models (LLMs) for resource-intensive vision-language (VL) pre-training. The current paradigm uses visual features as prompts to guide language models, with a focus on determining the most relevant visual features for corresponding text. Our approach diverges by concentrating on the language component, specifically identifying the optimal prompts to align with visual features. We introduce the Prompt-Transformer (P-Former), a model that predicts these ideal prompts, which is trained exclusively on linguistic data, bypassing the need for image-text pairings. This strategy subtly bifurcates the end-to-end VL training process into an additional, separate stage. Our experiments reveal that our framework significantly enhances the performance of a robust image-to-text baseline (BLIP-2), and effectively narrows the performance gap between models trained with either 4M or 129M image-text pairs. Importantly, our framework is modality-agnostic and flexible in terms of architectural design, as validated by its successful application in a video learning task using varied base modules. The code will be made available at https://github.com/yiren-jian/BLIText.

For agents

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

get_harvested_code_for_paper("2307.07063")
get_code_for_paper("2307.07063")
have("2307.07063")

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

curl https://syntology.ai/api/ran/2307.07063.json

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

[![Syntology run record](https://syntology.ai/api/ran/2307.07063.svg)](https://syntology.ai/paper/2307.07063)

Connect an agent — have() is free.