SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2111.10367 · 2021

SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech

arXiv · PDF · Open in the Atlas

Code that ran

11 of 12 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, 0 violated it and 11 have no contract verdict; 0 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/2111.10367.json · badge: /api/ran/2111.10367.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
asappresearch/slue-toolkit canonical (classifier label)
classifier basis not recorded on the link
11 ran · 1 no recorded run
FunctionStatusWhere it lives
convert Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_nel.py
code served (permissive licence) · get_code("0302aff920740fbc")
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 int, ndim 0, dtype family int, 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 torch

torch.manual_seed(0)

ARGS = {
    "sec": 5.5
}
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 int, ndim 0, dtype family int, 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
eval_ctc_model Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_w2v.py
code served (permissive licence) · get_code("3b9788e17612ec5c")
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 os
import tempfile
import shutil

# Create temporary directories for the setup
temp_dir = tempfile.mkdtemp()
model_dir = os.path.join(temp_dir, 'save', 'asr', 'w2v2-base-vc')
os.makedirs(os.path.join(model_dir, 'checkpoints'), exist_ok=True)
os.makedirs(os.path.join(model_dir, 'decode'), exist_ok=True)

# Create a minimal checkpoint file
checkpoint_path = os.path.join(model_dir, 'checkpoints', 'checkpoint_best.pt')
with open(checkpoint_path, 'w') as f:
    f.write('dummy checkpoint')

# Create user_dir
user_dir = os.path.join(temp_dir, 'slue_toolkit', 'fairseq_addon')
os.makedirs(user_dir, exist_ok=True)

# Create data manifest directory
data_dir = os.path.join(temp_dir, 'manifest', 'slue-voxceleb')
os.makedirs(data_dir, exist_ok=True)

# Create a dummy manifest file
with open(os.path.join(data_dir, 'dev.tsv'), 'w') as f:
    f.write('root\n')
    f.write('audio.wav\t100\n')

ARGS = {
    'model': model_dir,
    'lm': 'nolm-argmax',
    'beam_size': 500,
    'lm_weight': 2.0,
    'word_score': -1.0,
    'subset': 'dev',
    'data': os.path.join(temp_dir, 'manifest', 'slue-voxceleb'),
    'upsample': 1.0,
    'save_results': False,
    'dump_emissions': False,
    'csv_log_file': None,
    'fp16': False,
    'batch_size': -1,
    'quiet': True,
    'use_bpe': False,
    'user_dir': user_dir,
    'dry_run': True,
    'max_tokens': 4000000,
}
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 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
evaluate Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_nel.py
code served (permissive licence) · get_code("c8831cc7145df3ab")
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
import numpy as np

torch.manual_seed(0)

# The function evaluate takes three integer parameters representing
# confusion matrix components: true_pos, false_neg, false_pos
# It performs arithmetic operations on them to compute precision, recall, and f1

ARGS = {
    "true_pos": 42,
    "false_neg": 8,
    "false_pos": 5
}
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 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
get_ner_scores Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_ner.py
code served (permissive licence) · get_code("7df2656f9761c553")
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 numpy as np

# Build valid arguments for get_ner_scores
# The function expects:
# - all_gt: List of lists of tuples (label, phrase, identifier)
# - all_predictions: List of lists of tuples (label, phrase, identifier)

# Create ground truth data - list of sentences, each sentence is a list of (label, phrase, identifier) tuples
all_gt = [
    [
        ("PER", "John", 0),
        ("LOC", "New York", 0),
        ("ORG", "Apple", 0),
    ],
    [
        ("PER", "Jane", 0),
        ("LOC", "London", 0),
    ],
]

# Create predictions - some correct, some incorrect
all_predictions = [
    [
        ("PER", "John", 0),  # correct
        ("LOC", "New York", 0),  # correct
        ("ORG", "Microsoft", 0),  # incorrect label
    ],
    [
        ("PER", "Jane", 0),  # correct
        ("LOC", "Paris", 0),  # incorrect entity
    ],
]

ARGS = {
    "all_gt": all_gt,
    "all_predictions": all_predictions,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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_ner_stats Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_ner.py
code served (permissive licence) · get_code("7f746dc6c2a62bb2")
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
from collections import defaultdict

# Define ARGS with example NER ground truth and predictions
# Each entry is a tuple of (type_name, entity_info1, entity_info2)

all_gt = [
    [
        ("PERSON", 0, 5),
        ("LOCATION", 10, 15),
        ("ORG", 20, 25),
    ],
    [
        ("PERSON", 0, 3),
        ("LOCATION", 8, 12),
    ]
]

all_predictions = [
    [
        ("PERSON", 0, 5),
        ("LOCATION", 10, 16),  # slightly different end
        ("ORG", 20, 25),
    ],
    [
        ("PERSON", 0, 3),
        ("LOCATION", 8, 12),
        ("MISC", 15, 18),
    ]
]

ARGS = {
    "all_gt": all_gt,
    "all_predictions": all_predictions,
}
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
load_dct Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/generic_utils.py
code served (permissive licence) · get_code("656dc5c1fcd38ff8")
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 pickle as pkl
import tempfile
import os

# Create temporary directory for test files
tmpdir = tempfile.mkdtemp()

# Create a test JSON file
json_file = os.path.join(tmpdir, "test_data.json")
test_json_data = {"key": "value", "number": 42}
with open(json_file, 'w') as f:
    json.dump(test_json_data, f)

# Create a test pickle file
pkl_file = os.path.join(tmpdir, "test_data.pkl")
test_pkl_data = {"pickle_key": "pickle_value", "list": [1, 2, 3]}
with open(pkl_file, 'wb') as f:
    pkl.dump(test_pkl_data, f)

# Define ARGS for load_dct function
# The function takes one parameter: fname (filename)
# It checks if '.pkl' or '.json' is in the filename and loads accordingly
ARGS = {
    "fname": json_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 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_json Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/generic_utils.py
code served (permissive licence) · get_code("16afefff56a55570")
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 tempfile
import os

# Create a temporary JSON file
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.json")

# Write valid JSON data to the file
test_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "object"}
}

with open(temp_file, "w") as f:
    json.dump(test_data, f)

ARGS = {
    "fname": 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 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_pkl Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/generic_utils.py
code served (permissive licence) · get_code("e29d3eb7f31a6d05")
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 pickle as pkl
import tempfile
import os

# Create a temporary pickle file with some test data
temp_dir = tempfile.mkdtemp()
test_file = os.path.join(temp_dir, "test.pkl")

# Create test data and pickle it
test_data = {"key": "value", "list": [1, 2, 3], "nested": {"a": 1}}
with open(test_file, 'wb') as f:
    pkl.dump(test_data, f)

# Define ARGS with valid arguments for load_pkl
ARGS = {
    "fname": test_file,
    "encdng": None
}
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
make_distinct Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_w2v_ner.py
code served (permissive licence) · get_code("5b6c17ce2ecdeb80")
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 torch

# Analyze the function:
# - Takes label_lst as parameter
# - Iterates through label_lst items
# - Each item is unpacked as (tag, wrd) - so it's a tuple of 2 elements
# - Returns a list of tuples with 3 elements (tag, wrd, count)

# Create a simple label_lst with tuples of (tag, word)
label_lst = [
    ("NOUN", "cat"),
    ("VERB", "run"),
    ("NOUN", "dog"),
    ("NOUN", "cat"),
]

ARGS = {
    "label_lst": label_lst,
}
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 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
safe_divide Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_ner.py
code served (permissive licence) · get_code("815aaded7a9eae16")
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 ndarray, ndim 1, dtype family float, shape (4,) 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 numpy as np

# Construct valid arguments for safe_divide
numerator_val = [1.0, 2.0, 3.0, 0.0]
denominator_val = [2.0, 0.0, 3.0, 4.0]

ARGS = {
    "numerator": numerator_val,
    "denominator": denominator_val,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (4,) — shape (4,) 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
undetected_indices Ran
ran on a synthesized input
asappresearch/slue-toolkit/slue_toolkit/eval/eval_utils_nel.py
code served (permissive licence) · get_code("c6afd1a15980c103")
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 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

torch.manual_seed(0)

# The function takes two dictionaries where keys are tuples
# It finds keys that exist in gt_tuple_dct but not in pred_tuple_dct
# So we need two dicts with tuple keys, where gt has some keys pred doesn't have

gt_tuple_dct = {
    (0, 1): "ground_truth_1",
    (1, 2): "ground_truth_2",
    (2, 3): "ground_truth_3",
    (3, 4): "ground_truth_4",
}

pred_tuple_dct = {
    (0, 1): "prediction_1",
    (1, 2): "prediction_2",
}

ARGS = {
    "gt_tuple_dct": gt_tuple_dct,
    "pred_tuple_dct": pred_tuple_dct,
}
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 2, 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
parse_result Not yet run
unverified (harvested, no recorded run)
asappresearch/slue-toolkit/slue_toolkit/eval/eval_w2v.py
code served (permissive licence) · get_code("400008f29f6e7364")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 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 12 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

Progress in speech processing has been facilitated by shared datasets and benchmarks. Historically these have focused on automatic speech recognition (ASR), speaker identification, or other lower-level tasks. Interest has been growing in higher-level spoken language understanding tasks, including using end-to-end models, but there are fewer annotated datasets for such tasks. At the same time, recent work shows the possibility of pre-training generic representations and then fine-tuning for several tasks using relatively little labeled data. We propose to create a suite of benchmark tasks for Spoken Language Understanding Evaluation (SLUE) consisting of limited-size labeled training sets and corresponding evaluation sets. This resource would allow the research community to track progress, evaluate pre-trained representations for higher-level tasks, and study open questions such as the utility of pipeline versus end-to-end approaches. We present the first phase of the SLUE benchmark suite, consisting of named entity recognition, sentiment analysis, and ASR on the corresponding datasets. We focus on naturally produced (not read or synthesized) speech, and freely available datasets. We provide new transcriptions and annotations on subsets of the VoxCeleb and VoxPopuli datasets, evaluation metrics and results for baseline models, and an open-source toolkit to reproduce the baselines and evaluate new models.

For agents

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

get_harvested_code_for_paper("2111.10367")
get_code_for_paper("2111.10367")
have("2111.10367")

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

curl https://syntology.ai/api/ran/2111.10367.json

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

[![Syntology run record](https://syntology.ai/api/ran/2111.10367.svg)](https://syntology.ai/paper/2111.10367)

Connect an agent — have() is free.