SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.19113 · 2026

Learning from Complexity: Exploring Dynamic Sample Pruning of Spatio-Temporal Training

Wei Chen, Yuxuan Liang, Xiaofang Zhou, Junle Chen, Yuqian Wu

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 (8 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 8 have no contract verdict; 1 failure of our instrument; 3 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/2602.19113.json · badge: /api/ran/2602.19113.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
HKUDS/OpenCity community
Papers with Code archive
8 ran · 3 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
DropPath Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("dbb44455f4f7cf38")
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, DropPath: 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

torch.manual_seed(0)

# Create a small tensor for testing
# batch_size=2, feature_dim=4
x = torch.randn(2, 4)

# Define ARGS dict with all parameters for DropPath.__init__
ARGS = {
    'drop_prob': 0.5
}
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 DropPath — 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 DropPath 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
FeedForward Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("cd2445c057559fdf")
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, FeedForward: 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

torch.manual_seed(0)

ARGS = {
    'hidden_size': 4,
    'intermediate_size': 8,
}
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 FeedForward — 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 FeedForward 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
GCN Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("e7c52aa258e7a375")
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, GCN: 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

torch.manual_seed(0)

class GCN(nn.Module):
    def __init__(self, in_dim, out_dim, prob_drop, alpha):
        super(GCN, self).__init__()
        self.fc1 = nn.Linear(in_dim, out_dim, bias=False)
        self.mlp = nn.Linear(out_dim, out_dim)
        self.dropout = prob_drop
        self.alpha = alpha

    def forward(self, x, adj):
        d = adj.sum(1)
        h = x
        a = adj / d.view(-1, 1)
        gcn_out = self.fc1(torch.einsum('bdkt,nk->bdnt', h, a))
        out = self.alpha * x + (1 - self.alpha) * gcn_out
        ho = self.mlp(out)
        return ho

# Create test inputs
# x shape: [batch=2, d=3, k=4, t=5] for the einsum 'bdkt'
x = torch.randn(2, 3, 4, 5)

# adj shape: [n=4, k=4] for the einsum 'nk' 
# Make sure it's positive for the division operation
adj = torch.abs(torch.randn(4, 4)) + 0.1

# Create the GCN module
gcn_module = GCN(
    in_dim=5,      # matches t dimension from x
    out_dim=6,     # output dimension
    prob_drop=0.5, # dropout probability
    alpha=0.5      # mixing parameter
)

ARGS = {
    'in_dim': 5,
    'out_dim': 6,
    'prob_drop': 0.5,
    'alpha': 0.5
}
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 GCN — 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 GCN 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
LaplacianPE Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("cdf28431d6d0ca21")
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, LaplacianPE: 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

torch.manual_seed(0)

# Construct arguments for LaplacianPE
# The __init__ takes lape_dim and embed_dim
# The forward takes lap_mx which is passed to a Linear layer expecting input of size lape_dim
# The output is reshaped with unsqueeze(0).unsqueeze(0)

lape_dim = 4
embed_dim = 3

# Create a sample Laplacian matrix with shape (batch_size, lape_dim)
# Since forward expects lap_mx to go through nn.Linear(lape_dim, embed_dim)
# lap_mx should have last dimension of lape_dim
lap_mx = torch.randn(2, lape_dim)

ARGS = {
    "lape_dim": lape_dim,
    "embed_dim": embed_dim,
}
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 LaplacianPE — 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 LaplacianPE 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
LlamaRMSNorm Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("3e97aa6dae556601")
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, LlamaRMSNorm: 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

torch.manual_seed(0)

# Construct arguments for LlamaRMSNorm
hidden_size = 4
eps = 1e-06

# Create the module instance
model = nn.Module()

# Create a small hidden_states tensor (batch_size=2, sequence_length=3, hidden_size=4)
hidden_states = torch.randn(2, 3, 4)

ARGS = {
    "hidden_size": hidden_size,
    "eps": eps,
}
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 LlamaRMSNorm — 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 LlamaRMSNorm 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
PatchEmbedding_flow Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("80cbbab1399eaddc")
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, PatchEmbedding_flow: 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 math

torch.manual_seed(0)

# Define the PositionalEncoding class (needed as dependency)
class PositionalEncoding(nn.Module):
    def __init__(self, embed_dim, max_len=100):
        super(PositionalEncoding, self).__init__()
        pe = torch.zeros(max_len, embed_dim).float()
        pe.require_grad = False
        position = torch.arange(0, max_len).float().unsqueeze(1)
        div_term = (torch.arange(0, embed_dim, 2).float() * -(math.log(10000.0) / embed_dim)).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):
        return self.pe[:, :x.size(2)].unsqueeze(1).expand_as(x)

# Set up arguments for PatchEmbedding_flow
d_model = 8
patch_len = 4
stride = 2
padding = 0
his = 16

ARGS = {
    'd_model': d_model,
    'patch_len': patch_len,
    'stride': stride,
    'padding': padding,
    'his': his
}
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 PatchEmbedding_flow — 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 PatchEmbedding_flow 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
PatchEmbedding_time Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("665a9dbf1cad74e7")
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, PatchEmbedding_time: 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

torch.manual_seed(0)

# Analyze the forward method to understand what x should be:
# - x.size() returns (bs, ts, nn, dim)
# - x is permuted and reshaped to (bs, -1, ts)
# - x is unfolded along dimension=-1
# - Later: x_tdh, x_dwh, x_tdp, x_dwp are extracted and used with embeddings
# - The embeddings expect integer indices (for nn.Embedding)
# - x_tdh and x_tdp use daytime_embedding (expects values 0 to minute_size-1 = 0 to 1440)
# - x_dwh and x_dwp use weekday_embedding (expects values 0 to weekday_size-1 = 0 to 7)

# Build a minimal valid input tensor
# We need bs=2, ts (time steps), nn (number of nodes/features), dim=4
# After unfold and reshape operations, the tensor elements will be used as embedding indices
# So we need values in the appropriate ranges

bs = 2
ts = 8  # time steps
nn = 4  # number of nodes
dim = 4  # dimension (must be 4 to have indices for all 4 embeddings)

# Create input with values that are valid embedding indices
# Values should be in range [0, 1440] for daytime and [0, 7] for weekday
x = torch.randint(0, 8, (bs, ts, nn, dim), dtype=torch.long)

ARGS = {
    "d_model": 4,
    "patch_len": 2,
    "stride": 1,
    "padding": 0,
    "his": 8,
}
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 PatchEmbedding_time — 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 PatchEmbedding_time 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
PositionalEncoding Ran
ran on a synthesized input · constructed an object; no result was computed
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("4c1d82c5aee489f3")
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, PositionalEncoding: 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 math
import torch
import torch.nn as nn

torch.manual_seed(0)

# PositionalEncoding.__init__ takes:
# - self (implicit)
# - embed_dim: int
# - max_len: int (default=100)

# PositionalEncoding.forward takes:
# - self (implicit)
# - x: tensor with shape (batch, seq_len, embed_dim) based on usage in forward

ARGS = {
    'embed_dim': 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 PositionalEncoding — 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 PositionalEncoding 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
drop_path Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
this paper's copy was not recorded; identical code first harvested from hkuds/opencity
pointer only · get_code("9bdf2492c6e0a8de")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

OpenCity Not yet run
unverified (harvested, no recorded run)
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("ec0b8f261704414d")
How it ran

no invocation recorded.

STEncoderBlock Not yet run
unverified (harvested, no recorded run)
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("33b9c54c53084309")
How it ran

no invocation recorded.

TemporalSelfAttention Not yet run
unverified (harvested, no recorded run)
HKUDS/OpenCity/model/OpenCity/OpenCity.py
code served (permissive licence) · get_code("5c81ca3888f2a910")
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, 3 are also present in other papers whose code we hold (3 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.

FunctionAlso present inEarliest-datedOthers named
DropPath
brick of this paper's assembly: helper
47 other papers
47 dated earlier · 33 only as an assembly brick
2003.12039 (2020-03)
RAFT: Recurrent All-Pairs Field Transforms for Optical Flow
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
2105.14432 (2021-05)
TransMatcher: Deep Image Matching Through Transformers for Generalizable Person Re-identification
and 44 more not named here
drop_path
brick of this paper's assembly: helper
3 other papers
3 dated earlier · 1 only as an assembly brick
2301.07945 (2023-01)
PDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for Traffic Flow Prediction
2305.05140 (2023-05)
Linguistic More: Taking a Further Step toward Efficient and Accurate Scene Text Recognition
2408.10269 (2024-08)
OpenCity: Open Spatio-Temporal Foundation Models for Traffic Prediction
LaplacianPE
brick of this paper's assembly: model
1 other paper
1 dated earlier · 1 only as an assembly brick
2301.07945 (2023-01)
PDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for Traffic Flow Prediction

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

Spatio-temporal forecasting is fundamental to intelligent systems in transportation, climate science, and urban planning. However, training deep learning models on the massive, often redundant, datasets from these domains presents a significant computational bottleneck. Existing solutions typically focus on optimizing model architectures or optimizers, while overlooking the inherent inefficiency of the training data itself. This conventional approach of iterating over the entire static dataset each epoch wastes considerable resources on easy-to-learn or repetitive samples. In this paper, we explore a novel training-efficiency techniques, namely learning from complexity with dynamic sample pruning, ST-Prune , for spatio-temporal forecasting. Through dynamic sample pruning, we aim to intelligently identify the most informative samples based on the model's real-time learning state, thereby accelerating convergence and improving training efficiency. Extensive experiments conducted on real-world spatio-temporal datasets show that ST-Prune significantly accelerates the training speed while maintaining or even improving the model performance, and it also has scalability and universality.

For agents

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

get_harvested_code_for_paper("2602.19113")
get_code_for_paper("2602.19113")
have("2602.19113")

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

curl https://syntology.ai/api/ran/2602.19113.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.19113.svg)](https://syntology.ai/paper/2602.19113)

Connect an agent — have() is free.