SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.12467 · NeurIPS · 2020

The Depth-to-Width Interplay in Self-Attention

Yoav Levine, Noam Wies, Hofit Bata, Amnon Shashua

arXiv · PDF · Open in the Atlas

Code that ran

16 of 16 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, 1 violated it and 10 have no contract verdict; 4 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/2006.12467.json · badge: /api/ran/2006.12467.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
uf-hobi-informatics-lab/GatorTron community (classifier label)
classifier basis not recorded on the link
16 ran · 0 no recorded run
FunctionStatusWhere it lives
calc Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/re/brat_eval_res.py
code served (permissive licence) · get_code("88094c3b96a13d89")
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 torch

torch.manual_seed(0)

# The function calc takes three numeric parameters: tp, tp_fp, tp_tn
# These appear to be counts (true positives, true positives + false positives, true positives + true negatives)
# The function performs division operations on them, so they should be numeric values

tp = 45
tp_fp = 100
tp_tn = 60

ARGS = {
    "tp": tp,
    "tp_fp": tp_fp,
    "tp_tn": tp_tn
}
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, 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
d2t Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/pretraining/preprocessing/gatortron_process_data.py
code served (permissive licence) · get_code("e390dc7cb8d9947a")
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 json
import torch
import torch.nn as nn

torch.manual_seed(0)

class MockTokenizer:
    def sent_tokenizer(self, txt):
        return f"tokenized: {txt}"

class MockArgs:
    def __init__(self):
        self.col_name = "text_column"

x = {
    "text_column": "This is a sample text for tokenization."
}

tok = MockTokenizer()

args = MockArgs()

ARGS = {
    "x": x,
    "tok": tok,
    "args": args
}
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
exact_match_score Ran
ran, violated the contract we drafted
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-v1.1.py
pointer only (licence: MIT); the stored copy get_code serves, from sunlab-osu/CliniQG4QA, is pointer only (licence: NONE) · get_code("bf75673211903e14")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from sunlab-osu/CliniQG4QA, 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.

f1_score Ran
ran, honoured the contract we drafted
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-v1.1.py
pointer only (licence: MIT); the stored copy get_code serves, from sunlab-osu/CliniQG4QA, is pointer only (licence: NONE) · get_code("cdf4b8c0fec3da02")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from sunlab-osu/CliniQG4QA, 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.

ff Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/generate_conll_compatible_eval.py
code served (permissive licence) · get_code("82542482c691f03a")
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. 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

def ff(data):
    nsents = []
    for sent in data.split('\n\n'):
        nsent = []
        for each in sent.split('\n'):
            nsent.append(each.split(' '))
        nsents.append(nsent)
    return nsents

# Construct a valid argument for the function
# The function expects a string with sentences separated by '\n\n'
# and words within sentences separated by '\n' and ' '
data = "hello world\ngoodbye moon\n\nfoo bar baz\nqux quux"

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 list — a list 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
get_tokens Ran
ran; our contract draft was wrong, not the code
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-squad-v2.0.py
pointer only (licence: MIT); the stored copy get_code serves, from saptarshi059/cdqa-project, is pointer only (licence: NONE) · get_code("aa6bc32d9c17a2f1")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from saptarshi059/cdqa-project, 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.

load_bio_file_into_sents Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/new_bio_eval.py
code served (permissive licence) · get_code("ffebb9d76088e5ad")
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. 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

# Create a temporary BIO format file
with tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt') as f:
    bio_content = """word1 tag1
word2 tag2

word3 tag3
word4 tag4
word5 tag5

word6 tag6"""
    f.write(bio_content)
    bio_file_path = f.name

try:
    ARGS = {
        'bio_file': bio_file_path,
        'word_sep': ' ',
        'do_lower': False,
    }
finally:
    # Clean up temp file after defining ARGS
    pass
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 — a list 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
load_gs_labels Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/ner_expr.py
code served (permissive licence) · get_code("736dd0792c0ef1ce")
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 2, 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 label file with valid content
temp_dir = tempfile.mkdtemp()
label_file = os.path.join(temp_dir, "labels.txt")

with open(label_file, 'w') as f:
    f.write("class1 attribute1 value1\n")
    f.write("class2 attribute2 value2\n")
    f.write("\n")
    f.write("class3 attribute3 value3\n")

ARGS = {
    "label_file": label_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 2, 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
make_qid_to_has_ans Ran
ran; our contract draft was wrong, not the code
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-squad-v2.0.py
code served (permissive licence) · get_code("e91c44f5680827e2")
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) 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
import numpy as np

# Construct a minimal dataset matching the expected structure
dataset = [
    {
        'paragraphs': [
            {
                'qas': [
                    {
                        'id': 'q1',
                        'answers': [{'text': 'answer1', 'answer_start': 0}]
                    },
                    {
                        'id': 'q2',
                        'answers': []
                    }
                ]
            },
            {
                'qas': [
                    {
                        'id': 'q3',
                        'answers': [{'text': 'answer3', 'answer_start': 5}]
                    }
                ]
            }
        ]
    },
    {
        'paragraphs': [
            {
                'qas': [
                    {
                        'id': 'q4',
                        'answers': []
                    }
                ]
            }
        ]
    }
]

ARGS = {
    'dataset': dataset
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, 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
measure_prf Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/re/brat_eval_res.py
code served (permissive licence) · get_code("ead79428fb588ca6")
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.

from collections import defaultdict

mappings = {'Duration': 'Duration-Drug', 'Route': 'Route-Drug', 'Strength': 'Strength-Drug', 'ADE': 'ADE-Drug', 'Form': 'Form-Drug', 'Reason': 'Reason-Drug', 'Dosage': 'Dosage-Drug', 'Frequency': 'Frequency-Drug'}

class PRF:

    def __init__(self):
        self.tp = 0
        self.fp = 0

    def __repr__(self):
        return f'tp: {self.tp}; fp: {self.fp}'

def calc(tp, tp_fp, tp_tn):
    if tp_fp != 0:
        pre = tp / tp_fp
    else:
        pre = 0
    if tp_tn == 0:
        rec = 0
    else:
        rec = tp / tp_tn
    if pre == 0 and rec == 0:
        f1 = 0
    else:
        f1 = 2 * pre * rec / (pre + rec)
    return (pre, rec, f1)

def measure_prf(gss, preds):
    tp_tn_dict = defaultdict(lambda: 0)
    for each in gss:
        tp_tn_dict[each[0]] += 1
    tp_fp_cnt = defaultdict(PRF)
    total_tp, total_tp_fp, total_tp_tn = (0, 0, 0)
    res = dict()
    for each in preds:
        rel_type = each[0]
        for_eval = (each[0], *each[3:])
        if rel_type == 'NonRel':
            rel_type = mappings[each[1]]
        if for_eval in gss:
            tp_fp_cnt[rel_type].tp += 1
        else:
            tp_fp_cnt[rel_type].fp += 1
    for k, v in tp_fp_cnt.items():
        tp_tn = tp_tn_dict[k]
        tp = v.tp
        fp = v.fp
        tp_fp = tp + fp
        pre, rec, f1 = calc(tp, tp_fp, tp_tn)
        res[k] = (pre, rec, f1)
        total_tp += tp
        total_tp_fp += tp_fp
        total_tp_tn += tp_tn
    res['micro_avg'] = calc(total_tp, total_tp_fp, total_tp_tn)
    return res

gss = [
    ('Duration-Drug', 'entity1', 'entity2', 'arg1', 'arg2'),
    ('Duration-Drug', 'entity3', 'entity4', 'arg3', 'arg4'),
    ('Route-Drug', 'entity5', 'entity6', 'arg5', 'arg6'),
    ('ADE-Drug', 'entity7', 'entity8', 'arg7', 'arg8'),
]

preds = [
    ('Duration-Drug', 'Duration', 'other_info', 'entity1', 'entity2', 'arg1', 'arg2'),
    ('Duration-Drug', 'Duration', 'other_info', 'entity3', 'entity4', 'arg3', 'arg4'),
    ('NonRel', 'Route', 'other_info', 'entity5', 'entity6', 'arg5', 'arg6'),
    ('ADE-Drug', 'ADE', 'other_info', 'entity9', 'entity10', 'arg9', 'arg10'),
    ('Route-Drug', 'Route', 'other_info', 'entity11', 'entity12', 'arg11', 'arg12'),
]

ARGS = {
    'gss': gss,
    'preds': preds,
}
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
normalize_answer Ran
ran; our contract draft was wrong, not the code
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-v1.1.py
pointer only (licence: MIT); the stored copy get_code serves, from borealisai/radar-decoding, is pointer only (licence: LGPL-3.0) · get_code("e7e75981cb464788")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from borealisai/radar-decoding, and LGPL-3.0, as GitHub detects it on that repository, is not in the permissive set Syntology serves inline. 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.

normalize_answer Ran
ran; our contract draft was wrong, not the code
uf-hobi-informatics-lab/GatorTron/finetuning/qa/evaluate-squad-v2.0.py
pointer only (licence: MIT); the stored copy get_code serves, from saptarshi059/cdqa-project, is pointer only (licence: NONE) · get_code("6a96435eba311b08")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from saptarshi059/cdqa-project, 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.

predict_in_sent Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/ner_expr.py
code served (permissive licence) · get_code("7caf28c5dc24bdc1")
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
from tempfile import NamedTemporaryFile
import os

torch.manual_seed(0)

# Create a minimal config class with label_ids
class MinimalConfig:
    def __init__(self):
        self.label_ids = {
            'O': 0,
            'B-PER': 1,
            'I-PER': 2,
            'B-LOC': 3,
            'I-LOC': 4
        }

# Create a minimal NER model
class MinimalNERModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.cfg = MinimalConfig()
        self.linear = nn.Linear(10, 5)  # 5 label classes
    
    def _infer(self, lines, batch_size):
        """
        Simulates inference: returns a flat list of predicted label IDs.
        Each line gets a prediction for each word.
        """
        predictions = []
        for line in lines:
            words = line.split()
            # Simulate predictions: cycle through label IDs
            for i, word in enumerate(words):
                pred_id = i % 5  # 5 labels (0-4)
                predictions.append(pred_id)
        return predictions

# Create a temporary text file with sample sentences
text_file_content = "Hello world\nThis is a test\n"
temp_file = NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write(text_file_content)
temp_file.close()
text_file_path = temp_file.name

try:
    # Construct all arguments
    model_ner = MinimalNERModel()
    
    ARGS = {
        'text_file': text_file_path,
        'model_ner': model_ner,
        'bz': 32,
        'combine': True
    }
    
finally:
    # Clean up temp file after defining ARGS
    pass
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
read_from_file Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/new_bio_eval.py
code served (permissive licence) · get_code("532d336bacaef9b6")
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 tempfile
import os

# Create a temporary file with some text content
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("Hello, this is test content for the file reading function.")
temp_file.close()

# Define arguments for read_from_file
ARGS = {
    'ifn': temp_file.name
}
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 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
read_text Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/nemo2bio.py
code served (permissive licence) · get_code("fbf1f4b2112b56ad")
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 tempfile
import os

# Create a temporary text file
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.txt")
with open(temp_file, 'w') as f:
    f.write("This is a test file with some sample text.\nLine 2\nLine 3")

# Define arguments for the read_text function
ARGS = {
    "fn": temp_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 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
t2l Ran
ran on a synthesized input
uf-hobi-informatics-lab/GatorTron/finetuning/nemo_ner/nemo2bio.py
code served (permissive licence) · get_code("47e420d0a8dd35ec")
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. 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

def t2l(text):
    return [each.split(' ') for each in text.strip().split('\n')]

ARGS = {
    'text': 'hello world\nfoo bar baz\ntest'
}
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 — a list 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

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 6 later) and 10 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
normalize_answer 133 other papers
23 dated earlier, 109 later, 1 undated · 1 only as an assembly brick
1508.06615 (2015-08)
Character-Aware Neural Language Models
1606.05250 (2016-06)
SQuAD: 100,000+ Questions for Machine Comprehension of Text
1608.07905 (2016-08)
Machine Comprehension Using Match-LSTM and Answer Pointer
and 130 more not named here
exact_match_score 56 other papers
20 dated earlier, 36 later
1508.06615 (2015-08)
Character-Aware Neural Language Models
1606.05250 (2016-06)
SQuAD: 100,000+ Questions for Machine Comprehension of Text
1608.07905 (2016-08)
Machine Comprehension Using Match-LSTM and Answer Pointer
and 53 more not named here
f1_score 41 other papers
19 dated earlier, 22 later
1508.06615 (2015-08)
Character-Aware Neural Language Models
1606.05250 (2016-06)
SQuAD: 100,000+ Questions for Machine Comprehension of Text
1608.07905 (2016-08)
Machine Comprehension Using Match-LSTM and Answer Pointer
and 38 more not named here
normalize_answer 37 other papers
10 dated earlier, 27 later · 1 only as an assembly brick
1808.07042 (2018-08)
CoQA: A Conversational Question Answering Challenge
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1902.01718 (2019-02)
End-to-End Open-Domain Question Answering with BERTserini
and 34 more not named here
get_tokens 25 other papers
6 dated earlier, 19 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1902.01718 (2019-02)
End-to-End Open-Domain Question Answering with BERTserini
1908.08962 (2019-08)
Well-Read Students Learn Better: On the Importance of Pre-training Compact Models
and 22 more not named here
make_qid_to_has_ans 9 other papers
4 dated earlier, 5 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1908.08962 (2019-08)
Well-Read Students Learn Better: On the Importance of Pre-training Compact Models
1910.01244 (2019-10)
Linking artificial and human neural representations of language
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

Self-attention architectures, which are rapidly pushing the frontier in natural language processing, demonstrate a surprising depth-inefficient behavior: previous works indicate that increasing the internal representation (network width) is just as useful as increasing the number of self-attention layers (network depth). We theoretically predict a width-dependent transition between depth-efficiency and depth-inefficiency in self-attention. We conduct systematic empirical ablations on networks of depths 6 to 48 that clearly reveal the theoretically predicted behaviors, and provide explicit quantitative suggestions regarding the optimal depth-to-width allocation for a given self-attention network size. The race towards beyond 1-Trillion parameter language models renders informed guidelines for increasing self-attention depth and width in tandem an essential ingredient. Our guidelines elucidate the depth-to-width trade-off in self-attention networks of sizes up to the scale of GPT3 (which we project to be too deep for its size), and beyond, marking an unprecedented width of 30K as optimal for a 1-Trillion parameter network. 34th Conference on Neural Information Processing Systems (NeurIPS 2020),

For agents

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

get_harvested_code_for_paper("2006.12467")
get_code_for_paper("2006.12467")
have("2006.12467")

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

curl https://syntology.ai/api/ran/2006.12467.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.12467.svg)](https://syntology.ai/paper/2006.12467)

Connect an agent — have() is free.