SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.16964 · NeurIPS · 2024

Are Language Models Actually Useful for Time Series Forecasting?

Thomas Hartvigsen, Tim Althoff, Mingtian Tan, Mike Merrill, Vinayak Gupta

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 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/2406.16964.json · badge: /api/ran/2406.16964.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
thuml/AutoTimes canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
bennytmt/llmsfortimeseries canonical (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
bennytmt/ts_models canonical (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
FunctionStatusWhere it lives
MAPE Ran
ran, honoured the contract we drafted
thuml/AutoTimes/exp/exp_in_context_forecasting.py
code served (permissive licence) · get_code("9d0ccaea639499f4")
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
import torch

torch.manual_seed(0)

# MAPE takes pred and true as arguments
# Looking at the function body:
# - pred and true are used in arithmetic operations: 100 * (pred - true) / (true + 1e-08)
# - np.abs() is applied to the result
# - np.mean() is applied to that
# This means pred and true should be numpy arrays or array-like objects

# Create small numpy arrays with distinct shapes to allow dimension probing
pred = np.array([[1.2, 2.3, 3.1, 4.5],
                 [0.8, 1.9, 2.7, 3.2]])

true = np.array([[1.0, 2.0, 3.0, 4.0],
                 [1.0, 2.0, 3.0, 4.0]])

ARGS = {
    "pred": pred,
    "true": true
}
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 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
SMAPE Ran
ran, honoured the contract we drafted
thuml/AutoTimes/exp/exp_in_context_forecasting.py
code served (permissive licence) · get_code("a7f97eecdd50384f")
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
import torch

torch.manual_seed(0)

# SMAPE takes two arguments: pred and true
# Both are used in numpy operations: subtraction, absolute value, division, mean
# They should be numpy arrays or array-like objects

# Create small numpy arrays for predictions and true values
pred = np.array([1.0, 2.0, 3.0, 4.0])
true = np.array([1.1, 2.2, 2.9, 4.1])

ARGS = {
    "pred": pred,
    "true": true,
}
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 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
divide_no_nan Ran
ran, honoured the contract we drafted
bennytmt/llmsfortimeseries/CALF/utils/losses.py
pointer only (licence: NONE); the stored copy get_code serves, from seungyoon-Choi/CoSPOT, is pointer only (licence: NONE) · get_code("bdc787c888ab83a2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, bennytmt/llmsfortimeseries 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.

l2norm Ran
ran; our contract draft was wrong, not the code
bennytmt/ts_models/OFA/models/PatchTST.py
pointer only (licence: NONE); the stored copy get_code serves, from tmlresearchgroup-cas/clipure, is pointer only (licence: NONE) · get_code("3eb58dae620046d3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, bennytmt/ts_models 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.

url_file_name Ran
ran; our contract draft was wrong, not the code
thuml/AutoTimes/data_provider/m4.py
code served (permissive licence) · get_code("25a40de96ff65a01")
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 torch

ARGS = {
    "url": "https://example.com/path/to/filename.txt"
}
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

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 later) and 2 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
l2norm 16 other papers
12 dated earlier, 4 later · 4 only as an assembly brick
2112.05682 (2021-12)
Self-attention Does Not Need $O(n^2)$ Memory
2202.04200 (2022-02)
MaskGIT: Masked Generative Image Transformer
2203.07852 (2022-03)
Block-Recurrent Transformers
and 13 more not named here
divide_no_nan 14 other papers
5 dated earlier, 9 later
2401.09093 (2024-01)
RWKV-TS: Beyond Traditional Recurrent Neural Network for Time Series Tasks
2402.04852 (2024-02)
Multi-Patch Prediction: Adapting LLMs for Time Series Representation Learning
2403.01493 (2024-03)
ConvTimeNet: A Deep Hierarchical Fully Convolutional Model for Multivariate Time Series Analysis
and 11 more not named here
url_file_name 2 other papers
1 dated earlier, 1 later
2403.07300 (2024-03)
CALF: Aligning LLMs for Time Series Forecasting via Cross-modal Fine-Tuning
2412.17176 (2024-12)
WPMixer: Efficient Multi-Resolution Mixing for Long-Term Time Series Forecasting

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

Large language models (LLMs) are being applied to time series forecasting. But are language models actually useful for time series? In a series of ablation studies on three recent and popular LLM-based time series forecasting methods, we find that removing the LLM component or replacing it with a basic attention layer does not degrade forecasting performance-in most cases, the results even improve! We also find that despite their significant computational cost, pretrained LLMs do no better than models trained from scratch, do not represent the sequential dependencies in time series, and do not assist in few-shot settings. Additionally, we explore time series encoders and find that patching and attention structures perform similarly to LLM-based forecasters. 1

For agents

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

get_harvested_code_for_paper("2406.16964")
get_code_for_paper("2406.16964")
have("2406.16964")

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

curl https://syntology.ai/api/ran/2406.16964.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.16964.svg)](https://syntology.ai/paper/2406.16964)

Connect an agent — have() is free.