SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.23588 · 2026

Hyperdimensional Cross-Modal Alignment of Frozen Language and Image Models for Efficient Image Captioning

Vasant Honavar, Abhishek Dalvi

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 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; 5 failures of our instrument; 1 no recorded run), measured 2026-11-22; 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/2602.23588.json · badge: /api/ran/2602.23588.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
Abhishek-Dalvi410/HDFLIM canonical (classifier label)
classifier basis: census
6 ran · 1 no recorded run
FunctionStatusWhere it lives
LLM_get_next_token_logits Ran
ran; our contract draft was wrong, not the code
Abhishek-Dalvi410/HDFLIM/HD_eval.py
code served (permissive licence) · get_code("fa1226027a5d85a4")
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 1, dtype family float, shape (10,) 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)

class MockTokenizer:
    def encode(self, text, return_tensors=None):
        # Return a simple token sequence
        tokens = torch.tensor([[1, 2, 3, 4]])
        return tokens

class MockLLM(nn.Module):
    def __init__(self, vocab_size=10, hidden_dim=4):
        super().__init__()
        self.vocab_size = vocab_size
        self.hidden_dim = hidden_dim
        self.embedding = nn.Embedding(vocab_size, hidden_dim)
        self.linear = nn.Linear(hidden_dim, vocab_size)
    
    def forward(self, tokens):
        # tokens shape: (batch, seq_len)
        embedded = self.embedding(tokens)  # (batch, seq_len, hidden_dim)
        logits = self.linear(embedded)  # (batch, seq_len, vocab_size)
        
        # Create output object with logits attribute
        class Output:
            pass
        
        output = Output()
        output.logits = logits
        return output

model = MockLLM(vocab_size=10, hidden_dim=4)
model.eval()

tokenizer = MockTokenizer()

tokens = [1, 2, 3]

device = 'cpu'

ARGS = {
    'model': model,
    'tokenizer': tokenizer,
    'tokens': tokens,
    'device': device
}
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 1, dtype family float, shape (10,) — shape (10,) 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
collate_batch Ran
ran; our contract draft was wrong, not the code
Abhishek-Dalvi410/HDFLIM/train_scripts/train_dataloader.py
code served (permissive licence) · get_code("3fce7eb3771cbc0d")
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 a batch parameter which is an iterable of (image, caption) tuples
# Images should be tensors that can be stacked
# Captions should be strings

batch = [
    (torch.randn(3, 4, 4), "a dog running in the park. the dog is happy"),
    (torch.randn(3, 4, 4), "a cat sleeping. the cat looks peaceful")
]

ARGS = {
    'batch': batch
}
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
collate_batch_with_url Ran
ran; our contract draft was wrong, not the code
Abhishek-Dalvi410/HDFLIM/train_scripts/train_dataloader.py
code served (permissive licence) · get_code("5e81330d181a2824")
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 a batch which is a list of tuples (image, caption, url)
# Each image should be a tensor, caption a string, url a string

batch = [
    (torch.randn(3, 4, 4), "hello world. this is a test", "http://example.com/1"),
    (torch.randn(3, 4, 4), "another caption. with multiple sentences", "http://example.com/2"),
]

ARGS = {
    "batch": batch,
}
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
make_transform Ran
ran; our contract draft was wrong, not the code
Abhishek-Dalvi410/HDFLIM/transform_inference.py
code served (permissive licence) · get_code("0f6d4cce31d2e3a5")
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 Compose. 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.transforms.v2 as Tv2

torch.manual_seed(0)

ARGS = {
    "resize_size": 512,
    "crop_size": 512,
    "aspect_ratio_threshold": 1.4,
}
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 Compose — a Compose output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Compose 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
make_transform Ran
ran; our contract draft was wrong, not the code
Abhishek-Dalvi410/HDFLIM/train_scripts/train_dataloader.py
code served (permissive licence) · get_code("47c74dfbe51ef957")
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 Compose. 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.transforms.v2 as Tv2

torch.manual_seed(0)

ARGS = {
    "resize_size": 512,
    "crop_size": 512,
    "aspect_ratio_threshold": 1.4,
}
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 Compose — a Compose output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Compose 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
pack_boolean_tensor Ran
ran, honoured the contract we drafted
Abhishek-Dalvi410/HDFLIM/HD_eval.py
code served (permissive licence) · get_code("a51de12f664b0ee2")
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 int, shape (2, 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
import numpy as np

torch.manual_seed(0)

# Create a small boolean tensor
# Shape: (2, 4) - batch of 2, dimension of 4
token_hd_vec = torch.randint(0, 2, (2, 4), dtype=torch.bool)

ARGS = {
    "token_hd_vec": token_hd_vec
}
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 Tensor, ndim 2, dtype family int, shape (2, 1) — shape (2, 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
download_image_with_retry Not yet run
unverified (harvested, no recorded run)
Abhishek-Dalvi410/HDFLIM/HD_eval.py
code served (permissive licence) · get_code("c729b1b75e07464c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 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 7 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

Large unimodal foundation models for vision and language encode rich semantic structures, yet aligning them typically requires computationally intensive multimodal fine-tuning. Such approaches depend on large-scale parameter updates, are resource-intensive, and can perturb pretrained representations. Emerging evidence suggests, however, that independently trained foundation models may already exhibit latent semantic compatibility, reflecting shared structures in the data they model. This raises a fundamental question: can cross-modal alignment be achieved without modifying the models themselves? Here we introduce HDFLIM (HyperDimensional computing with Frozen Language and Image Models), a framework that establishes cross-modal mappings while keeping pretrained vision and language models fully frozen. HDFLIM projects unimodal embeddings into a shared hyperdimensional space and leverages lightweight symbolic operations -binding, bundling, and similarity-based retrieval -to construct associative cross-modal representations in a single pass over the data. Caption generation emerges from high-dimensional memory retrieval rather than iterative gradient-based optimization. We show that HDFLIM achieves performance comparable to end-to-end vision-language training methods and produces captions that are more semantically grounded than zero-shot baselines. By decoupling alignment from parameter tuning, our results suggest that semantic mapping across foundation models can be realized through symbolic operations on hyperdimensional encodings of the respective embeddings. More broadly, this work points toward an alternative paradigm for foundation model alignment in which frozen models are integrated through structured representational mappings rather than through large-scale retraining. The codebase for our implementation can be found at https://github.com/Abhishek-Dalvi410/HDFLIM Keywords Hyperdimensional Computing • Image Captioning • Frozen Models • Symbolic Mapping Recent work points to an emergent alignment between the internal representations learned by large-scale vision and language models [42,27,41]. Remarkably, even when trained in isolation, unimodal models can exhibit substantial semantic compatibility, suggesting that both modalities may converge toward shared latent structure in their representa-

For agents

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

get_harvested_code_for_paper("2602.23588")
get_code_for_paper("2602.23588")
have("2602.23588")

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

curl https://syntology.ai/api/ran/2602.23588.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.23588.svg)](https://syntology.ai/paper/2602.23588)

Connect an agent — have() is free.