SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2209.13575 · NeurIPS · 2022

Efficient Non-Parametric Optimizer Search for Diverse Tasks

Cho-Jui Hsieh, Ruochen Wang, Minhao Cheng, Yuanhao Xiong

arXiv · PDF · Open in the Atlas

Code that ran

3 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 3 have no contract verdict; 0 failures of our instrument; 6 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/2209.13575.json · badge: /api/ran/2209.13575.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
ruocwang/efficient-optimizer-search marked official in Papers with Code
Papers with Code archive
3 ran · 6 no recorded run
FunctionStatusWhere it lives
ProgramLearningAlgorithm Ran
ran on a synthesized input · constructed an object; no result was computed
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("4465536a110c6ef4")
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, ProgramLearningAlgorithm: 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 numpy as np

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

class SimpleConfig:
    def __init__(self):
        self.learning_rate = 0.001
        self.max_iterations = 100
        self.batch_size = 2

class SimpleDataset:
    def __init__(self):
        self.data = [
            (torch.randn(2, 4), torch.randn(2, 3)),
            (torch.randn(2, 4), torch.randn(2, 3))
        ]
    
    def __iter__(self):
        return iter(self.data)
    
    def __len__(self):
        return len(self.data)

model = SimpleModel()
config = SimpleConfig()
dataset = SimpleDataset()
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)
loss_fn = nn.MSELoss()

ARGS = {
}
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 ProgramLearningAlgorithm — 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 ProgramLearningAlgorithm 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
ProgramNode Ran
ran on a synthesized input · constructed an object; no result was computed
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("8191438f2eae0781")
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, ProgramNode: 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

torch.manual_seed(0)

# Define a simple program representation
program = "sample_program"

# Define a numeric score
score = 0.85

# Define a parent node (None for root)
parent = None

# Define depth in tree
depth = 0

# Define cost value
cost = 1.5

# Define order
order = 0

ARGS = {
    "program": program,
    "score": score,
    "parent": parent,
    "depth": depth,
    "cost": cost,
    "order": order
}
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 ProgramNode — 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 ProgramNode 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
Visit Ran
ran on a synthesized input · constructed an object; no result was computed
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("a07a253ee8057616")
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, Visit: 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 numpy as np
import copy
import logging

logging.basicConfig(level=logging.INFO)

torch.manual_seed(0)

# Create a minimal dsl_func.LibraryFunction stand-in
class MockLibraryFunction:
    def __init__(self, name, submodules=None, has_params=False, parameters=None):
        self.name = name
        self.submodules = submodules or {}
        self.has_params = has_params
        self.parameters = parameters or {}
    
    def reset_memory_state(self):
        pass
    
    def execute_on_batch(self, probe, deter=False):
        # Return a tensor of shape (4,) to match probe dimensions
        return torch.randn(4)
    
    def execute_sym(self):
        return "G + M1"

# Create a minimal Features class
class MockFeatures:
    def __init__(self):
        self.names = ['G', 'W', 'M1', 'M2', 'Gp', 'M1p', 'M2p', 'M3', 'RMSprop', 'Adam', 't', 'T', 'lr', 'L', 'lamb', 'momentum', 'momentum2', 'momentum3']

# Create the features object
features = MockFeatures()

# Create a mock program
program = MockLibraryFunction(
    name="test_program",
    submodules={},
    has_params=False,
    parameters={}
)

# Define the ARGS dictionary
ARGS = {
    'features': features,
    'update': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sympy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Visit — 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 Visit 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
MC_SAMPLING_NOS Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("5f8b23db306d8206")
How it ran

no invocation recorded.

ProgramGraph Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("c5929cfc13a96de4")
How it ran

no invocation recorded.

load_training_task Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("a73f58271e56d4a3")
How it ran

no invocation recorded.

log_and_print Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("ad6c9f7884554a1f")
How it ran

no invocation recorded.

print_program Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("90e91c9cf039b628")
How it ran

no invocation recorded.

print_program_dict Not yet run
unverified (harvested, no recorded run)
ruocwang/efficient-optimizer-search/src/algorithms/nos_mc_sampling.py
code served (permissive licence) · get_code("845bc5b5aadd180c")
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

Efficient and automated design of optimizers plays a crucial role in full-stack AutoML systems. However, prior methods in optimizer search are often limited by their scalability, generability, or sample efficiency. With the goal of democratizing research and application of optimizer search, we present the first efficient, scalable and generalizable framework that can directly search on the tasks of interest. We first observe that optimizer updates are fundamentally mathematical expressions applied to the gradient. Inspired by the innate tree structure of the underlying math expressions, we re-arrange the space of optimizers into a super-tree, where each path encodes an optimizer. This way, optimizer search can be naturally formulated as a path-finding problem, allowing a variety of well-established tree traversal methods to be used as the search algorithm. We adopt an adaptation of the Monte Carlo method to tree search, equipped with rejection sampling and equivalentform detection that leverage the characteristics of optimizer update rules to further boost the sample efficiency. We provide a diverse set of tasks to benchmark our algorithm and demonstrate that, with only 128 evaluations, the proposed framework can discover optimizers that surpass both human-designed counterparts and prior optimizer search methods. Preprint. Under review.

For agents

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

get_harvested_code_for_paper("2209.13575")
get_code_for_paper("2209.13575")
have("2209.13575")

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

curl https://syntology.ai/api/ran/2209.13575.json

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

[![Syntology run record](https://syntology.ai/api/ran/2209.13575.svg)](https://syntology.ai/paper/2209.13575)

Connect an agent — have() is free.