SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.01289 · 2024

Greed is All You Need: An Evaluation of Tokenizer Inference Methods

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; 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/2403.01289.json · badge: /api/ran/2403.01289.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
melelbgu/tokenizers_intrinsic_benchmark canonical (classifier label)
classifier basis: census
6 ran · 1 no recorded run
FunctionStatusWhere it lives
corpus_to_list Ran
ran; our contract draft was wrong, not the code
melelbgu/tokenizers_intrinsic_benchmark/utils.py
code served (permissive licence) · get_code("09cc2fed4f32c676")
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 file with some test content
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt', encoding='utf-8')
temp_file.write("Line 1\n")
temp_file.write("Line 2\n")
temp_file.write("Line 3\n")
temp_file.close()

ARGS = {
    'file_path': temp_file.name,
    'encoding': 'utf-8'
}
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
encode_corpus Ran
ran; our contract draft was wrong, not the code
melelbgu/tokenizers_intrinsic_benchmark/Intrinsic_measures/static.py
code served (permissive licence) · get_code("179dca114c61f9e8")
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 dict. 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

ARGS = {
    'corpus': [
        'hello world',
        'foo bar baz'
    ],
    'tokenized_corpus': [
        ['hello', 'world'],
        ['foo', 'bar', 'baz']
    ]
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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_boundaries Ran
ran, honoured the contract we drafted
melelbgu/tokenizers_intrinsic_benchmark/Intrinsic_measures/ling_utils.py
code served (permissive licence) · get_code("d1c70365e5bcfc9e")
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 int; 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

ARGS = {
    "tokenization": ["hello", "world", "foo", "bar"]
}
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 int — 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_hf_normalizer Ran
ran; our contract draft was wrong, not the code
melelbgu/tokenizers_intrinsic_benchmark/utils.py
code served (permissive licence) · get_code("28bc7b54a7dfb03c")
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 BertNormalizer. 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 tokenizers import normalizers

ARGS = {
    'normalizer_config': {
        'type': 'BertNormalizer',
        'lowercase': True,
        'accent_map': None,
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tokenizers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type BertNormalizer — a BertNormalizer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BertNormalizer 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_seg_coverage Ran
ran; our contract draft was wrong, not the code
melelbgu/tokenizers_intrinsic_benchmark/Intrinsic_measures/ling_utils.py
code served (permissive licence) · get_code("abfab6454d26977b")
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 dict. 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 pandas as pd
import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a mock tokenizer with necessary methods
class MockTokenizer:
    def __init__(self):
        self.vocab = {
            'hello': 0,
            'world': 1,
            'Ġhello': 2,
            'Ġworld': 3,
            '##world': 4,
            'test': 5,
            'Ġtest': 6,
            '##test': 7,
            'ing': 8,
            '##ing': 9,
        }
    
    def get_vocab(self):
        return self.vocab
    
    def tokenize(self, text):
        # Simple mock tokenization
        if text == "hello world":
            return ['hello', 'world']
        elif text == "testing":
            return ['test', '##ing']
        else:
            return [text]

# Create a mock dataframe
def create_mock_dataframe():
    data = {
        'text': ['hello world', 'testing', 'hello'],
        'other_col': ['a', 'b', 'c']
    }
    return pd.DataFrame(data)

# Create a mock get_gstandard function
def mock_get_gstandard(row):
    text = row['text']
    if text == 'hello world':
        return ['hello', 'world']
    elif text == 'testing':
        return ['test', 'ing']
    else:
        return ['hello']

# Construct the arguments
ARGS = {
    'x': create_mock_dataframe(),
    'tokenizer': MockTokenizer(),
    'key_in_df': 'text',
    'get_gstandard': mock_get_gstandard,
    'special': '##'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
load_tokenizer Ran
ran; our contract draft was wrong, not the code
melelbgu/tokenizers_intrinsic_benchmark/utils.py
code served (permissive licence) · get_code("07940de2cf1e2aa0")
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 dict. 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 json
import os
import tempfile

# Create a temporary directory and config file
temp_dir = tempfile.mkdtemp()
config_filepath = os.path.join(temp_dir, "tokenizer_config.json")

# Create a valid tokenizer config file
tokenizer_config = {
    "vocab_size": 1000,
    "model_type": "bert",
    "max_position_embeddings": 512,
    "hidden_size": 768,
    "num_attention_heads": 12
}

with open(config_filepath, 'w', encoding='utf-8') as f:
    json.dump(tokenizer_config, f)

ARGS = {
    "config_filepath": config_filepath
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
eval_cog Not yet run
unverified (harvested, no recorded run)
melelbgu/tokenizers_intrinsic_benchmark/Intrinsic_measures/human_comp.py
code served (permissive licence) · get_code("a3fbe1b7c4be8a13")
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

While subword tokenizers such as BPE and WordPiece are typically used to build vocabularies for NLP models, the method of decoding text into a sequence of tokens from these vocabularies is often left unspecified, or ill-suited to the method in which they were constructed. We provide a controlled analysis of seven tokenizer inference methods across four different algorithms and three vocabulary sizes, performed on a novel intrinsic evaluation suite we curated for English, combining measures rooted in morphology, cognition, and information theory. We show that for the most commonly used tokenizers, greedy inference performs surprisingly well; and that SaGe, a recently-introduced contextually-informed tokenizer, outperforms all others on morphological alignment.

For agents

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

get_harvested_code_for_paper("2403.01289")
get_code_for_paper("2403.01289")
have("2403.01289")

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

curl https://syntology.ai/api/ran/2403.01289.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.01289.svg)](https://syntology.ai/paper/2403.01289)

Connect an agent — have() is free.