SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.10911 · 2026

Tuning the burn-in phase in training recurrent neural networks improves their performance

Victor Lopez, Matthias Müller, Julian Schiller, Malte Heinrich

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 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; 2 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; 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/2602.10911.json · badge: /api/ran/2602.10911.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
jdschille/rnn-training canonical (classifier label)
classifier basis: census
7 ran · 1 no recorded run
FunctionStatusWhere it lives
create_bound Ran
ran, honoured the contract we drafted
jdschille/rnn-training/02_public_datasets/code/check_mdl_TSF.py
code served (permissive licence) · get_code("6972b1409b230e0a")
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 2, dtype family float, shape (10, 1) 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

C = 50
lam = 0.9999
seq_len = 10

ARGS = {
    "C": C,
    "lam": lam,
    "seq_len": seq_len
}
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 2, dtype family float, shape (10, 1) — shape (10, 1) 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
create_sequences Ran
ran; our contract draft was wrong, not the code
jdschille/rnn-training/02_public_datasets/code/check_mdl_TSF.py
code served (permissive licence) · get_code("9490846531c3f8f5")
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. 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

# Analyze the function: create_sequences takes parameter 'x'
# The function:
# - Iterates over range(len(x))
# - Accesses x[i:] (slicing)
# - Calls np.size(x, axis=1) to get second dimension
# - Concatenates arrays
# - x must be a 2D numpy array

# Create a small 2D numpy array with shape (4, 3)
# Using distinct dimensions: 4 rows, 3 columns
x = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0],
    [10.0, 11.0, 12.0]
], dtype=np.float32)

ARGS = {
    'x': x
}
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 tuple — a tuple 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
create_sequences_io_many_to_many_shift_by_length_mapping Ran
ran; our contract draft was wrong, not the code
jdschille/rnn-training/02_public_datasets/code/SYSID_functions.py
code served (permissive licence) · get_code("f8c935dffb57df0a")
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 3, 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

# Create simple numpy arrays for x and y
x = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12], dtype=np.float32)
y = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120], dtype=np.float32)

# Set length to create sequences
length = 3

ARGS = {
    "x": x,
    "y": y,
    "length": length
}
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 tuple, ndim 3, 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
create_sequences_io_many_to_many_shift_by_one_mapping Ran
ran; our contract draft was wrong, not the code
jdschille/rnn-training/02_public_datasets/code/SYSID_functions.py
code served (permissive licence) · get_code("749c6902a08b3c7e")
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. 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

# Create simple numpy arrays for x and y
# x and y should be 1D or 2D arrays that can be sliced
x = np.array([1, 2, 3, 4, 5, 6, 7, 8])
y = np.array([10, 20, 30, 40, 50, 60, 70, 80])

# Set length to create sequences
length = 3

ARGS = {
    "x": x,
    "y": y,
    "length": length,
}
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 tuple — a tuple 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
create_sequences_io_many_to_one_mapping Ran
ran; our contract draft was wrong, not the code
jdschille/rnn-training/02_public_datasets/code/SYSID_functions.py
code served (permissive licence) · get_code("1fc39074f3ff0e1d")
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. 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

# Create small sample data for x and y
x = np.array([1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0])
y = np.array([10.0, 20.0, 30.0, 40.0, 50.0, 60.0, 70.0, 80.0])

# Define the length parameter
length = 3

ARGS = {
    "x": x,
    "y": y,
    "length": length
}
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 tuple — a tuple 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
create_val_data Ran
ran; our contract draft was wrong, not the code
jdschille/rnn-training/02_public_datasets/code/TSF_functions.py
code served (permissive licence) · get_code("410e86672fc0a99b")
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. 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

# Create a small time series array
# The function slices z in two ways:
# X = z[:-forecasting_length_H]
# Y = z[lookback_length_L:]
# So we need z to be long enough to support both slices

lookback_length_L = 3
forecasting_length_H = 2

# z needs to be longer than both lookback_length_L and forecasting_length_H
# Let's make it length 10 to have enough data
z = np.array([1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0])

ARGS = {
    "z": z,
    "lookback_length_L": lookback_length_L,
    "forecasting_length_H": forecasting_length_H,
}
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 tuple — a tuple 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_data_length Ran
ran, honoured the contract we drafted
jdschille/rnn-training/02_public_datasets/code/TSF_functions.py
code served (permissive licence) · get_code("fe4cf68d59a9f77e")
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 pandas as pd
import tempfile
import os

# Create a temporary CSV file with test data
temp_dir = tempfile.mkdtemp()
csv_path = os.path.join(temp_dir, "test_data.csv")

# Create a simple CSV with 5 columns (representing a time series with 5 features)
df = pd.DataFrame({
    'feature_1': [1.0, 2.0, 3.0],
    'feature_2': [4.0, 5.0, 6.0],
    'feature_3': [7.0, 8.0, 9.0],
    'feature_4': [10.0, 11.0, 12.0],
    'feature_5': [13.0, 14.0, 15.0]
})
df.to_csv(csv_path, index=False)

ARGS = {
    'path': csv_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — 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
create_seq2seq_mapping Not yet run
unverified (harvested, no recorded run)
jdschille/rnn-training/02_public_datasets/code/TSF_functions.py
code served (permissive licence) · get_code("5393540beba283bb")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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

Training recurrent neural networks (RNNs) with standard backpropagation through time (BPTT) can be challenging, especially in the presence of long input sequences. A practical alternative to reduce computational and memory overhead is to perform BPTT repeatedly over shorter segments of the training data set, corresponding to truncated BPTT. In this paper, we examine the training of RNNs when using such a truncated learning approach for time series tasks. Specifically, we establish theoretical bounds on the accuracy and performance loss when optimizing over subsequences instead of the full data sequence. This reveals that the burn-in phase of the RNN is an important tuning knob in its training, with significant impact on the performance guarantees. We validate our theoretical results through experiments on standard benchmarks from the fields of system identification and time series forecasting. In all experiments, we observe a strong influence of the burn-in phase on the training process, and proper tuning can lead to a reduction of the prediction error on the training and test data of more than 60% in some cases.

For agents

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

get_harvested_code_for_paper("2602.10911")
get_code_for_paper("2602.10911")
have("2602.10911")

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

curl https://syntology.ai/api/ran/2602.10911.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.10911.svg)](https://syntology.ai/paper/2602.10911)

Connect an agent — have() is free.