SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.04882 · ICLR · 2024

LMUFormer: Low Complexity Yet Powerful Spiking Model With Legendre Memory Units

Peter Beerel, Anni Li, Zeyu Liu, Gourav Datta

arXiv · PDF · Open in the Atlas

Code that ran

9 of 12 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 7 failures of our instrument; 3 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/2402.04882.json · badge: /api/ran/2402.04882.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
zeyuliu1037/LMUFormer canonical (classifier label)
classifier basis: census
8 ran · 1 no recorded run
zeyuliu1037/lmuformer canonical (classifier label)
classifier basis: classifier_2025
1 ran · 2 no recorded run
FunctionStatusWhere it lives
LMUFFTCell Ran
ran on a synthesized input · constructed an object; no result was computed
zeyuliu1037/lmuformer/src/blocks/lmu.py
code served (permissive licence) · get_code("6b8fea6e27713ec3")
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 an instance of its own class, LMUFFTCell: the run constructed the object and did not call it, so no result was computed. 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
import numpy as np

torch.manual_seed(0)

# Build arguments for LMUFFTCell.__init__
input_size = 3
hidden_size = 4
memory_size = 5
seq_len = 6
theta = np.pi

ARGS = {
    'input_size': input_size,
    'hidden_size': hidden_size,
    'memory_size': memory_size,
    'seq_len': seq_len,
    'theta': theta,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LMUFFTCell — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a LMUFFTCell 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
calculate_stats Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zeyuliu1037/LMUFormer/src/utilities/stats.py
code served (permissive licence) · get_code("571ae4f8e76e6bc8")
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 numpy as np
from scipy import stats
from sklearn import metrics

np.random.seed(0)

# Create sample data for the function
# output: 2d array (samples_num, classes_num)
# target: 2d array (samples_num, classes_num)

samples_num = 10
classes_num = 4

# Create target as one-hot encoded labels
target = np.zeros((samples_num, classes_num))
for i in range(samples_num):
    target[i, i % classes_num] = 1

# Create output as probability-like scores (values between 0 and 1)
output = np.random.rand(samples_num, classes_num)
# Normalize to make them more like probabilities
output = output / output.sum(axis=1, keepdims=True)

ARGS = {
    'output': output,
    'target': target
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, sklearn — 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
computeMatchmap Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/utilities/util.py
code served (permissive licence) · get_code("5ee11f4af0d2fc1e")
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 3, dtype family float, shape (3, 5, 2) 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)

# I should be a 3D tensor with shape (D, H, W)
# A should be a 2D tensor with shape (D, T)
# The function reshapes I to (H*W, D), transposes to (H*W, D),
# then multiplies with A (D, T) to get (H*W, T), then reshapes to (H, W, T)

D = 4  # feature dimension
H = 3  # height
W = 5  # width
T = 2  # output channels

I = torch.randn(D, H, W)
A = torch.randn(D, T)

ARGS = {
    "I": I,
    "A": A
}
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 3, dtype family float, shape (3, 5, 2) — shape (3, 5, 2) 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
d_prime Ran
ran, honoured the contract we drafted
zeyuliu1037/LMUFormer/src/utilities/stats.py
code served (permissive licence) · get_code("e469cef4384ff73a")
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 float64, ndim 0, 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 numpy as np
from scipy import stats

torch_seed = 0

ARGS = {
    "auc": 0.75
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
lookup_list Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/dataloader.py
code served (permissive licence) · get_code("c15efdd12442aeb4")
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 csv
import os

# Create a temporary CSV file with the required structure
temp_dir = tempfile.mkdtemp()
label_csv_path = os.path.join(temp_dir, "labels.csv")

# Write a sample CSV file with 'index' and 'display_name' columns
with open(label_csv_path, 'w', newline='') as f:
    writer = csv.DictWriter(f, fieldnames=['index', 'display_name'])
    writer.writeheader()
    writer.writerow({'index': '0', 'display_name': 'cat'})
    writer.writerow({'index': '1', 'display_name': 'dog'})
    writer.writerow({'index': '2', 'display_name': 'bird'})

# Define the arguments
ARGS = {
    'index_list': ['0', '1', '2'],
    'label_csv': label_csv_path
}
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
make_index_dict Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/dataloader.py
code served (permissive licence) · get_code("a8fb3f62e2a901d7")
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 csv
import tempfile
import os

# Create a temporary CSV file with the expected structure
temp_dir = tempfile.mkdtemp()
label_csv_path = os.path.join(temp_dir, "labels.csv")

# Write a sample CSV file with 'mid' and 'index' columns
with open(label_csv_path, 'w', newline='') as f:
    writer = csv.DictWriter(f, fieldnames=['mid', 'index'])
    writer.writeheader()
    writer.writerow({'mid': 'mid_001', 'index': '0'})
    writer.writerow({'mid': 'mid_002', 'index': '1'})
    writer.writerow({'mid': 'mid_003', 'index': '2'})

ARGS = {
    'label_csv': label_csv_path
}
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_name_dict Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/dataloader.py
code served (permissive licence) · get_code("cbf115a314bcd23e")
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 csv
import tempfile
import os

# Create a temporary CSV file with the expected structure
temp_dir = tempfile.mkdtemp()
label_csv_path = os.path.join(temp_dir, "labels.csv")

with open(label_csv_path, 'w', newline='') as f:
    writer = csv.DictWriter(f, fieldnames=['index', 'display_name'])
    writer.writeheader()
    writer.writerow({'index': '0', 'display_name': 'cat'})
    writer.writerow({'index': '1', 'display_name': 'dog'})
    writer.writerow({'index': '2', 'display_name': 'bird'})

ARGS = {
    'label_csv': label_csv_path
}
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
matchmapSim Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/utilities/util.py
code served (permissive licence) · get_code("546678fcbd2854db")
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 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 torch

torch.manual_seed(0)

# Create a 3D tensor: (batch=2, height=3, width=4)
# The function calls .max(0), .max(2), and .mean() on it
M = torch.randn(2, 3, 4)

# simtype can be one of: 'SISA', 'MISA', or 'SIMA'
simtype = 'SISA'

ARGS = {
    'M': M,
    'simtype': simtype
}
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 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
validate_ensemble Ran
ran; our contract draft was wrong, not the code
zeyuliu1037/LMUFormer/src/traintest.py
code served (permissive licence) · get_code("ecb5ae1f3a358b96")
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 os
import numpy as np
import tempfile
import shutil

# Create a temporary directory structure for the experiment
temp_dir = tempfile.mkdtemp()
exp_dir = temp_dir
predictions_dir = os.path.join(exp_dir, 'predictions')
os.makedirs(predictions_dir, exist_ok=True)

# Create dummy target data
target_data = np.array([[0.1, 0.2], [0.3, 0.4], [0.5, 0.6], [0.7, 0.8]])
np.savetxt(os.path.join(predictions_dir, 'target.csv'), target_data, delimiter=',')

# Create dummy predictions for epoch 1
predictions_1 = np.array([[0.15, 0.25], [0.35, 0.45], [0.55, 0.65], [0.75, 0.85]])
np.savetxt(os.path.join(predictions_dir, 'predictions_1.csv'), predictions_1, delimiter=',')

# Create dummy predictions for epoch 2
predictions_2 = np.array([[0.12, 0.22], [0.32, 0.42], [0.52, 0.62], [0.72, 0.82]])
np.savetxt(os.path.join(predictions_dir, 'predictions_2.csv'), predictions_2, delimiter=',')

# Create cumulative predictions for epoch 2
cum_predictions = predictions_1
np.savetxt(os.path.join(predictions_dir, 'cum_predictions.csv'), cum_predictions, delimiter=',')

# Define a minimal calculate_stats function that the main function calls
def calculate_stats(predictions, target):
    """Dummy stats calculation function"""
    return {
        'mse': float(np.mean((predictions - target) ** 2)),
        'mae': float(np.mean(np.abs(predictions - target)))
    }

# Make calculate_stats available in the global scope
import builtins
builtins.calculate_stats = calculate_stats

# Create a simple args class
class Args:
    def __init__(self, exp_dir):
        self.exp_dir = exp_dir

ARGS = {
    'args': Args(exp_dir),
    'epoch': 1
}
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
SLMU Not yet run
unverified (harvested, no recorded run)
zeyuliu1037/lmuformer/src/blocks/lmu.py
code served (permissive licence) · get_code("f4bcb5b173cf2e51")
How it ran

no invocation recorded.

SpikingLMUFFTCell Not yet run
unverified (harvested, no recorded run)
zeyuliu1037/lmuformer/src/blocks/lmu.py
code served (permissive licence) · get_code("72efcd381a467f79")
How it ran

no invocation recorded.

calc_recalls Not yet run
unverified (harvested, no recorded run)
zeyuliu1037/LMUFormer/src/utilities/util.py
code served (permissive licence) · get_code("0bed13389f791924")
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, 9 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 0 the same month, 5 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.

FunctionAlso present inEarliest-datedOthers named
lookup_list 7 other papers
3 dated earlier, 4 later
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2208.00061 (2022-08)
UAVM: Towards Unifying Audio and Visual Models
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
and 4 more not named here
make_index_dict 7 other papers
3 dated earlier, 4 later
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2208.00061 (2022-08)
UAVM: Towards Unifying Audio and Visual Models
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
and 4 more not named here
make_name_dict 7 other papers
3 dated earlier, 4 later
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2208.00061 (2022-08)
UAVM: Towards Unifying Audio and Visual Models
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
and 4 more not named here
d_prime 4 other papers
3 dated earlier, 1 later
1903.00765 (2019-03)
Weakly Labelled AudioSet Tagging with Attention Neural Networks
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
and 1 more not named here
validate_ensemble 3 other papers
1 dated earlier, 2 later
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2405.11831 (2024-05)
SSAMBA: Self-Supervised Audio Representation Learning with Mamba State Space Model
2406.03344 (2024-06)
Audio Mamba: Bidirectional State Space Model for Audio Representation Learning
calc_recalls 2 other papers
2 dated earlier
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
computeMatchmap 2 other papers
2 dated earlier
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
matchmapSim 2 other papers
2 dated earlier
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer
2307.03183 (2023-07)
Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggers
calculate_stats 1 other paper
1 dated earlier
2110.09784 (2021-10)
SSAST: Self-Supervised Audio Spectrogram Transformer

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

Transformer models have demonstrated high accuracy in numerous applications but have high complexity and lack sequential processing capability making them ill-suited for many streaming applications at the edge where devices are heavily resource-constrained. Thus motivated, many researchers have proposed reformulating the transformer models as RNN modules which modify the self-attention computation with explicit states. However, these approaches often incur significant performance degradation. The ultimate goal is to develop a model that has the following properties: parallel training, streaming and low-cost inference, and state-of-the-art (SOTA) performance. In this paper, we propose a new direction to achieve this goal. We show how architectural modifications to a fully-sequential recurrent model can help push its performance toward Transformer models while retaining its sequential processing capability. Specifically, inspired by the recent success of Legendre Memory Units (LMU) in sequence learning tasks, we propose LMUFormer, which augments the LMU with convolutional patch embedding and convolutional channel mixer. Moreover, we present a spiking version of this architecture, which introduces the benefit of states within the patch embedding and channel mixer modules while simultaneously reducing the computing complexity. We evaluated our architectures on multiple sequence datasets. Of particular note is our performance on the Speech Commands V2 dataset (35 classes). In comparison to SOTA transformer-based models within the ANN domain, our LMUFormer demonstrates comparable performance while necessitating a remarkable 53× reduction in parameters and a substantial 65× decrement in FLOPs. Furthermore, when benchmarked against extant low-complexity SNN variants, our model establishes a new SOTA with an accuracy of 96.12%. Additionally, owing to our model's proficiency in real-time data processing, we are able to achieve a 32.03% reduction in sequence length, all while incurring an inconsequential decline in performance. Our code is publicly available at https://github.com/zeyuliu1037/LMUFormer.git

For agents

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

get_harvested_code_for_paper("2402.04882")
get_code_for_paper("2402.04882")
have("2402.04882")

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

curl https://syntology.ai/api/ran/2402.04882.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.04882.svg)](https://syntology.ai/paper/2402.04882)

Connect an agent — have() is free.