SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2609.01147 · September 2026

On the Design Fundamentals of Pixel Text Representation Learning

Yu Rong, Hong Cheng, Chenghao Xiao, Ruifeng Yuan, Hou Chan, Hou Pong Chan, Chaohao Yuan, Zhuoxu Huang

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; 0 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 1 failure 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/2609.01147.json · badge: /api/ran/2609.01147.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
Pixel-Linguist/Pixel-Linguist-II canonical (classifier label)
classifier basis: census
2 ran · 1 no recorded run
FunctionStatusWhere it lives
clean_text Ran
ran; our contract draft was wrong, not the code
Pixel-Linguist/Pixel-Linguist-II/training/filter_dataset.py
code served (permissive licence) · get_code("b5bb3eecb46f7556")
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 torch.nn as nn

torch.manual_seed(0)

# The function clean_text takes a single parameter: text
# It processes the text by:
# 1. Converting to string if needed
# 2. Filtering out control characters (ord < 32) except \n, \r, \t
# 3. Replacing newlines with spaces
# 4. Stripping whitespace

# Create test inputs for the text parameter
test_text_1 = "Hello World\nThis is a test"
test_text_2 = "Text with\ttabs\rand\x01control\x02chars"
test_text_3 = ""
test_text_4 = "   Normal text   "

# We'll use a simple string as the main test case
ARGS = {
    'text': "Hello World\nThis is a test\twith\x01control\x02chars"
}
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 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
has_ancient_script Ran
ran, violated the contract we drafted
Pixel-Linguist/Pixel-Linguist-II/training/filter_dataset.py
code served (permissive licence) · get_code("7013058c74d3b412")
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 bool, ndim 0, dtype family float, shape () 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 re

ANCIENT_SCRIPTS_REGEX = re.compile('[\\u1680-\\u169F\\u16A0-\\u16FF\\U00010000-\\U0001007F\\U00010330-\\U0001034F\\U00010900-\\U0001091F\\U00012000-\\U000123FF\\U00013000-\\U0001342F\\U00010800-\\U0001083F]', re.UNICODE)

def has_ancient_script(text):
    """Whether the text contains archaic scripts."""
    if ANCIENT_SCRIPTS_REGEX.search(text):
        return True
    return False

# Test case 1: regular ASCII text (no ancient scripts)
test_text_1 = "Hello world"

# Test case 2: text with ancient script (Runic script U+16A0-U+16FF)
test_text_2 = "Hello \u16A0 world"  # Contains Runic character

ARGS = {
    "text": test_text_1
}
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 bool, ndim 0, dtype family float, shape () — shape () 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
fetch Not yet run
unverified (harvested, no recorded run)
Pixel-Linguist/Pixel-Linguist-II/training/download_font_lib.py
code served (permissive licence) · get_code("9a899a58e1d1bcd5")
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

Text-rich visual inputs require models that can read, retrieve, and compress language directly in pixel space, yet existing pixel-text encoders struggle with fixed resolution pretraining, visual shortcut learning, weak visual grounding, and multilingual visual text understanding. In this work, we investigate the fundamental design principles required for robust visual text representation learning. Through systematic controlled ablations, we identify four critical components: variable image resolutions and rendered font sizes provide spatial proxies for high-resolution document generalization; natural image-text pairs are indispensable for grounding and prevent text-only collapse; layout-aware rendering helps prevent pixel-level shortcuts; and a two-stage multilingual curriculum enables effective cross-lingual alignment. By integrating these principles into a scalable training recipe, we train Pixel Linguist II, a native-resolution vision encoder trained with on-the-fly rendering, unified contrastive grounding, and a multilingual curriculum over 280M training examples. Pixel Linguist II sets new state-of-the-art results on English, cross-lingual, and multilingual Visual STS and ViDoRe, while also enabling better MLLM downstream evaluation. Notably, Pixel Linguist II remains robust under 80\% visual token compression, showing great promise for optical context compression. Our code and resources are available at https://github.com/Pixel-Linguist/Pixel-Linguist-II.

For agents

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

get_harvested_code_for_paper("2609.01147")
get_code_for_paper("2609.01147")
have("2609.01147")

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

curl https://syntology.ai/api/ran/2609.01147.json

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

[![Syntology run record](https://syntology.ai/api/ran/2609.01147.svg)](https://syntology.ai/paper/2609.01147)

Connect an agent — have() is free.