SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2510.08445 · NeurIPS · 2025

Synthetic Series-Symbol Data Generation for Time Series Foundation Models

Yujian Li, Wenxuan Wang, Kai Wu, Xiaoyu Zhang, Dan Wang

arXiv · PDF

Code that ran

5 of 9 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 1 failure of our instrument; 4 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/2510.08445.json · badge: /api/ran/2510.08445.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
wwhenxuan/SymTime community
Papers with Code archive
5 ran · 4 no recorded run
FunctionStatusWhere it lives
MultiHeadAttention Ran
ran on a synthesized input · constructed an object; no result was computed
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("394f2672bffaba81")
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, MultiHeadAttention: 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
from torch import nn

torch.manual_seed(0)

# Create sample tensors for the forward pass
batch_size = 2
seq_len = 3
d_model = 4
n_heads = 2

# Input tensor q: [batch_size x seq_len x d_model]
q = torch.randn(batch_size, seq_len, d_model)

# Optional: k and v (can be None, will default to q in forward)
k = torch.randn(batch_size, seq_len, d_model)
v = torch.randn(batch_size, seq_len, d_model)

# Optional attention mask: [seq_len] - will be broadcast in forward
attn_mask = torch.randint(0, 2, (seq_len,), dtype=torch.bool)

ARGS = {
    'd_model': 4,
    'n_heads': 2,
    'd_k': None,
    'd_v': None,
    'attn_dropout': 0.0,
    'proj_dropout': 0.0,
    'qkv_bias': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MultiHeadAttention — 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 MultiHeadAttention 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
PositionalEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("98baad7ba2deef88")
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, PositionalEmbedding: 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
from torch import nn
from torch import Tensor
import math

class PositionalEmbedding(nn.Module):
    """Adding the positional encoding to the input for Transformer"""

    def __init__(self, d_model: int, max_len: int=5000) -> None:
        super(PositionalEmbedding, self).__init__()
        pe = torch.zeros(max_len, d_model).float()
        pe.requires_grad = False
        position = torch.arange(0, max_len).float().unsqueeze(1)
        div_term = (torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model)).exp()
        pe[:, 0::2] = torch.sin(position * div_term)
        pe[:, 1::2] = torch.cos(position * div_term)
        pe = pe.unsqueeze(0)
        self.register_buffer('pe', pe)

    def forward(self, x: Tensor) -> Tensor:
        return x + self.pe[:, :x.size(1)]

torch.manual_seed(0)

# Create a PositionalEmbedding instance with d_model=4
pos_embedding = PositionalEmbedding(d_model=4, max_len=100)

# Create a sample input tensor with shape (batch_size=2, sequence_length=10, d_model=4)
x = torch.randn(2, 10, 4)

ARGS = {
    'd_model': 4,
    'max_len': 100
}
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 PositionalEmbedding — 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 PositionalEmbedding 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
TSTEncoderLayer Ran
ran on a synthesized input · constructed an object; no result was computed
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("ca01642b3d8ec47f")
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, TSTEncoderLayer: 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
from torch import nn
import torch.nn.functional as F

torch.manual_seed(0)

# The function signature shows TSTEncoderLayer.__init__ takes these parameters:
# d_model, n_heads, d_k, d_v, d_ff, store_attn, norm, attn_dropout, dropout, bias, activation, pre_norm

ARGS = {
    'd_model': 64,
    'n_heads': 4,
    'd_k': None,
    'd_v': None,
    'd_ff': 256,
    'store_attn': False,
    'norm': 'BatchNorm',
    'attn_dropout': 0.1,
    'dropout': 0.1,
    'bias': True,
    'activation': 'gelu',
    'pre_norm': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TSTEncoderLayer — 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 TSTEncoderLayer 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
_ScaledDotProductAttention Ran
ran on a synthesized input
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("323c9bcb7517c914")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) 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 torch
import numpy as np

# INIT_ARGS for _ScaledDotProductAttention.__init__
INIT_ARGS = {
    'd_model': 12,
    'n_heads': 3,
    'attn_dropout': 0.1,
    'res_attention': False
}

# ARGS for _ScaledDotProductAttention.forward
batch_size = 2
n_heads = 3
num_token = 5
d_k = 4  # d_model // n_heads = 12 // 3 = 4

q = torch.randn(batch_size, n_heads, num_token, d_k)
k = torch.randn(batch_size, n_heads, d_k, num_token)
v = torch.randn(batch_size, n_heads, num_token, d_k)
attn_mask = torch.randn(batch_size, num_token) > 0  # boolean mask

ARGS = {
    'q': q,
    'k': k,
    'v': v,
    'attn_mask': attn_mask
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
einops, numpy, torch — 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
get_activation_fn Ran
ran; our contract draft was wrong, not the code
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("bb11ef32f48c0c09")
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 ReLU. 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 torch import nn
from typing import Union, Callable

torch.manual_seed(0)

# Test case 1: string activation "relu"
ARGS_relu = {
    'activation': 'relu'
}

# Test case 2: string activation "gelu"
ARGS_gelu = {
    'activation': 'gelu'
}

# Test case 3: callable activation function
def custom_activation():
    return nn.Tanh()

ARGS_callable = {
    'activation': custom_activation
}

# Use the first test case as the primary ARGS
ARGS = ARGS_relu
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 ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU 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
LLM Not yet run
unverified (harvested, no recorded run)
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("43a8088d7f81189c")
How it ran

no invocation recorded.

SymTime Not yet run
unverified (harvested, no recorded run)
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("cf8961e48fabac14")
How it ran

no invocation recorded.

TSTEncoder Not yet run
unverified (harvested, no recorded run)
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("b034c725d5790051")
How it ran

no invocation recorded.

Transpose Not yet run
unverified (harvested, no recorded run)
wwhenxuan/SymTime/models/pretrain_model.py
code served (permissive licence) · get_code("b6f4a7655eb0836c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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

Foundation models for time series analysis (TSA) have attracted significant attention. However, challenges such as training data scarcity and imbalance continue to hinder their development. Inspired by complex dynamic system theories, we design a series-symbol data generation mechanism, enabling the unrestricted creation of high-quality time series data paired with corresponding symbolic expressions. To leverage series-symbol data pairs with strong correlations, we develop SymTime, a pre-trained foundation model for enhancing time series representation using symbolic information. SymTime demonstrates competitive performance across five major TSA tasks when fine-tunes with downstream tasks, rivaling foundation models pre-trained on real-world datasets. This approach underscores the potential of series-symbol data generation and pretraining mechanisms in overcoming data scarcity and enhancing task performance. The code is available at https://github.com/wwhenxuan/SymTime.

For agents

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

get_harvested_code_for_paper("2510.08445")
get_code_for_paper("2510.08445")
have("2510.08445")

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

curl https://syntology.ai/api/ran/2510.08445.json

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

[![Syntology run record](https://syntology.ai/api/ran/2510.08445.svg)](https://syntology.ai/paper/2510.08445)

Connect an agent — have() is free.