SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.08473 · 2024

Strategies for Pretraining Neural Operators

arXiv · PDF · Open in the Atlas

Code that ran

8 of 10 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 7 failures of our instrument; 2 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/2406.08473.json · badge: /api/ran/2406.08473.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
anthonyzhou-1/pretraining_pdes canonical (classifier label)
classifier basis: census
8 ran · 2 no recorded run
FunctionStatusWhere it lives
Downsample Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/models/unet.py
code served (permissive licence) · get_code("2c6f83aa983db012")
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 Conv2d. 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 = {
    'dim': 64
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
Upsample Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/models/unet.py
code served (permissive licence) · get_code("937f75fe76f87b5d")
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 ConvTranspose2d. 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 = {
    "dim": 64
}
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 ConvTranspose2d — a ConvTranspose2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ConvTranspose2d 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
dict2tensor Ran
ran, honoured the contract we drafted
anthonyzhou-1/pretraining_pdes/common/utils.py
code served (permissive licence) · get_code("892313d7e1615391")
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 2, dtype family float, shape (4, 3) 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)

# The function expects a dictionary where each value is a tensor
# It unsqueezes each tensor (adds a dimension at position 0)
# then concatenates them along dim 0, and transposes the result

# Create a dictionary with tensor values
# Each tensor should be 1D or multi-dimensional so unsqueeze(0) makes sense
d = {
    'key1': torch.randn(4),
    'key2': torch.randn(4),
    'key3': torch.randn(4),
}

ARGS = {
    'd': d
}
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 2, dtype family float, shape (4, 3) — shape (4, 3) 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
get_negative_sample Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/loss/spatiotemporal_losses.py
code served (permissive licence) · get_code("552ce5cf4cb55924")
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 (5, 4, 3) 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
import random

torch.manual_seed(0)
random.seed(0)

# Create a small tensor with shape [time_history, x, (y)]
# Using [time_history=5, x=4, y=3] to have distinct dimensions
u = torch.randn(5, 4, 3)

# difficulty parameter ranges from 0.0 to 1.0
difficulty = 0.5

ARGS = {
    "u": u,
    "difficulty": difficulty,
}
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 (5, 4, 3) — shape (5, 4, 3) 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
load_args Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/common/utils.py
code served (permissive licence) · get_code("82f3c79752f554e8")
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 Namespace. 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 argparse
import yaml
import json
import tempfile
import os

# Create temporary config files for testing
temp_dir = tempfile.mkdtemp()

# Create a YAML config file
yaml_config_path = os.path.join(temp_dir, "config.yaml")
yaml_content = {
    "learning_rate": 0.001,
    "batch_size": 32,
    "epochs": 10,
    "model_name": "test_model"
}
with open(yaml_config_path, 'w') as f:
    yaml.dump(yaml_content, f)

# Create a JSON config file
json_config_path = os.path.join(temp_dir, "config.json")
json_content = {
    "learning_rate": 0.002,
    "batch_size": 64,
    "epochs": 20,
    "model_name": "json_model"
}
with open(json_config_path, 'w') as f:
    json.dump(json_content, f)

# Create parser
parser = argparse.ArgumentParser()
parser.add_argument('--config', type=str, default=None)
parser.add_argument('--learning_rate', type=float, default=0.01)
parser.add_argument('--batch_size', type=int, default=16)
parser.add_argument('--epochs', type=int, default=5)
parser.add_argument('--model_name', type=str, default='default_model')

# Create args namespace with config file
args = argparse.Namespace(config=yaml_config_path)

ARGS = {
    "args": args,
    "parser": parser
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Namespace — a Namespace output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Namespace 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
norm_coeffs Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/loss/statistical_losses.py
code served (permissive licence) · get_code("cef23dd47d82c863")
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 torch

torch.manual_seed(0)

# Construct coefficients tensor as a dictionary-like object
# The function accesses coeffs[key], so we need a dict that supports item access
coeffs = {
    'alpha': torch.tensor(3.5),
    'beta': torch.tensor(0.8),
    'gamma': torch.tensor(2.0),
    'nu': torch.tensor(0.05),
    'ax': torch.tensor(5.0),
    'ay': torch.tensor(7.5),
    'cx': torch.tensor(2.0),
    'cy': torch.tensor(3.0)
}

# PDE identifier
pde = 'heat_adv_burgers'

ARGS = {
    'coeffs': coeffs,
    'pde': pde
}
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 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
process_dict Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/common/utils.py
code served (permissive licence) · get_code("55c5fc2140b8cefa")
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 torch
import numpy as np
from collections import OrderedDict

torch.manual_seed(0)

# Create a sample state_dict with OrderedDict
# The function partitions keys by a prefix and removes it
state_dict = OrderedDict([
    ('module.layer1.weight', torch.randn(4, 3)),
    ('module.layer1.bias', torch.randn(4)),
    ('module.layer2.weight', torch.randn(5, 4)),
    ('other.param', torch.randn(2, 2)),
])

prefix = 'module'

ARGS = {
    'state_dict': state_dict,
    'prefix': prefix,
}
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 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
shuffle_odd Ran
ran; our contract draft was wrong, not the code
anthonyzhou-1/pretraining_pdes/loss/spatiotemporal_losses.py
code served (permissive licence) · get_code("bc813d36201f1617")
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 torch
import random

torch.manual_seed(0)
random.seed(0)

# u is a tensor of shape [batch, time_history, x, (y)]
# Looking at the function, it processes u[b] which should be [time_history, x, (y)]
# The get_negative_sample function expects [time_history, x, (y)]
# Let's create a batch of such tensors with distinct sizes

batch_size = 2
time_history = 5
x_dim = 4
y_dim = 3

u = torch.randn(batch_size, time_history, x_dim, y_dim)

ARGS = {
    "u": u,
    "difficulty": 0.5,
    "weighted_sample": 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 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_decoder Not yet run
unverified (harvested, no recorded run)
anthonyzhou-1/pretraining_pdes/loss/spatiotemporal_losses.py
code served (permissive licence) · get_code("b98e86c5c3097f56")
How it ran

no invocation recorded.

vis_2d_plot Not yet run
unverified (harvested, no recorded run)
anthonyzhou-1/pretraining_pdes/common/plotting.py
code served (permissive licence) · get_code("225f951811962879")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 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, 1 later) and 7 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
Upsample 3 other papers
2 dated earlier, 1 later
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2306.06874 (2023-06)
VillanDiffusion: A Unified Backdoor Attack Framework for Diffusion Models
Downsample 1 other paper
1 dated earlier
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
process_dict 1 other paper
1 dated earlier
2403.17728 (2024-03)
Masked Autoencoders are PDE Learners

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

Pretraining for partial differential equation (PDE) modeling has recently shown promise in scaling neural operators across datasets to improve generalizability and performance. Despite these advances, our understanding of how pretraining affects neural operators is still limited; studies generally propose tailored architectures and datasets that make it challenging to compare or examine different pretraining frameworks. To address this, we compare various pretraining methods without optimizing architecture choices to characterize pretraining dynamics on different models and datasets as well as to understand its scaling and generalization behavior. We find that pretraining is highly dependent on model and dataset choices, but in general transfer learning or physics-based pretraining strategies work best. In addition, pretraining performance can be further improved by using data augmentations. Lastly, pretraining can be additionally beneficial when fine-tuning in scarce data regimes or when generalizing to downstream data similar to the pretraining distribution. Through providing insights into pretraining neural operators for physics prediction, we hope to motivate future work in developing and evaluating pretraining methods for PDEs.

For agents

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

get_harvested_code_for_paper("2406.08473")
get_code_for_paper("2406.08473")
have("2406.08473")

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

curl https://syntology.ai/api/ran/2406.08473.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.08473.svg)](https://syntology.ai/paper/2406.08473)

Connect an agent — have() is free.