SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2207.03084 · 2022

Pre-training helps Bayesian optimization too

arXiv · PDF · Open in the Atlas

Code that ran

7 of 12 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; 0 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 0 failures of our instrument; 5 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/2207.03084.json · badge: /api/ran/2207.03084.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
google-research/hyperbo canonical (classifier label)
classifier basis not recorded on the link
7 ran · 5 no recorded run
FunctionStatusWhere it lives
backtracking_linesearch Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/basics/lbfgs.py
code served (permissive licence) · get_code("96b3667fd2aaf84e")
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, ndim 1, 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 jax
import jax.numpy as jnp
import jax.tree_util

# Define a simple objective function that returns value and gradients
def simple_objective(params, *args):
    """Simple quadratic objective: 0.5 * sum(p^2)"""
    loss = 0.5 * jax.tree_util.tree_reduce(
        jnp.add, 
        jax.tree.map(lambda x: jnp.sum(x**2), params)
    )
    grads = jax.grad(lambda p: 0.5 * jax.tree_util.tree_reduce(
        jnp.add, 
        jax.tree.map(lambda x: jnp.sum(x**2), p)
    ))(params)
    return loss, grads

# Create small test parameters as a dict of arrays
params = {
    'w1': jnp.array([[1.0, 2.0, 3.0, 4.0], 
                     [5.0, 6.0, 7.0, 8.0]]),
    'w2': jnp.array([[0.5, 1.5], 
                     [2.5, 3.5],
                     [4.5, 5.5],
                     [6.5, 7.5]])
}

# Compute initial value and gradients
cur_val, grads = simple_objective(params)

# Define a descent direction (negative gradient direction)
direction = jax.tree.map(lambda g: -g, grads)

# Create a value_and_grad function
val_and_grad_fn = lambda params, *args: simple_objective(params, *args)

ARGS = {
    'val_and_grad_fn': val_and_grad_fn,
    'cur_val': cur_val,
    'params': params,
    'grads': grads,
    'direction': direction,
    'alpha': 1.0,
    'c1': 0.0001,
    'c2': 0.9,
    'tau': 0.5,
    'max_steps': 50,
    'has_aux': False,
    'args': tuple()
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
absl, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
expected_improvement_sub Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/bo_utils/acfun.py
code served (permissive licence) · get_code("2d3ab739783d2452")
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 ArrayImpl, ndim 2, dtype family float, shape (2, 1) 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 jax
import jax.numpy as jnp
import jax.scipy as jsp

# Set seed for reproducibility
key = jax.random.PRNGKey(0)

# Generate small sample data
n = 2  # number of query points
mu = jnp.array([[0.5], [1.2]], dtype=jnp.float32)  # n x 1 posterior means
std = jnp.array([[0.3], [0.4]], dtype=jnp.float32)  # n x 1 posterior standard deviations
target = 2.0  # target value to be improved over

ARGS = {
    "mu": mu,
    "std": std,
    "target": target,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 2, dtype family float, shape (2, 1) — shape (2, 1) 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_best_datapoint Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/bo_utils/bayesopt.py
code served (permissive licence) · get_code("020f1f64d8f9a8de")
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 jax.numpy as jnp

class SubDataset:
    """Stand-in class for SubDataset with x and y attributes."""
    def __init__(self, x, y):
        self.x = x
        self.y = y

# Create small test data
x_data = jnp.array([[1.0, 2.0, 3.0, 4.0],
                     [5.0, 6.0, 7.0, 8.0]])
y_data = jnp.array([0.5, 2.3])

sub_dataset = SubDataset(x=x_data, y=y_data)

ARGS = {
    "sub_dataset": sub_dataset
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — 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
lbfgs_descent_dir_nocedal Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/basics/lbfgs.py
code served (permissive licence) · get_code("7352dd5a69ac440c")
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 jax
import jax.numpy as jnp

# Create sample gradient dict with nested structure
grads = {
    'layer1': jnp.array([1.0, 2.0, 3.0, 4.0]),
    'layer2': jnp.array([0.5, 1.5, 2.5, 3.5])
}

# Create s: list of dicts containing differences in gradients from last M updates
# Using M=2 for a small example
s = [
    {
        'layer1': jnp.array([0.1, 0.2, 0.3, 0.4]),
        'layer2': jnp.array([0.05, 0.15, 0.25, 0.35])
    },
    {
        'layer1': jnp.array([0.2, 0.4, 0.6, 0.8]),
        'layer2': jnp.array([0.1, 0.3, 0.5, 0.7])
    }
]

# Create y: list of dicts containing differences in parameters from last M updates
y = [
    {
        'layer1': jnp.array([0.01, 0.02, 0.03, 0.04]),
        'layer2': jnp.array([0.005, 0.015, 0.025, 0.035])
    },
    {
        'layer1': jnp.array([0.02, 0.04, 0.06, 0.08]),
        'layer2': jnp.array([0.01, 0.03, 0.05, 0.07])
    }
]

ARGS = {
    'grads': grads,
    's': s,
    'y': y
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — 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
random_search Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/bo_utils/acfun.py
code served (permissive licence) · get_code("ed6ff476d6215244")
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 ArrayImpl, ndim 2, dtype family float, shape (2, 1) 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 jax.random as jrd
import jax.numpy as jnp

class MockModel:
    def __init__(self):
        self.rng = jrd.PRNGKey(42)

model = MockModel()
x_queries = jnp.ones((2, 4))

ARGS = {
    "model": model,
    "x_queries": x_queries,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 2, dtype family float, shape (2, 1) — shape (2, 1) 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
sample_dataframe Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/bo_utils/data.py
code served (permissive licence) · get_code("fcb5788c1b680975")
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 DataFrame. 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 numpy as np
import pandas as pd

# Create a small dataframe with some sample data
df = pd.DataFrame({
    'col1': [1, 2, 3, 4, 5],
    'col2': ['a', 'b', 'c', 'd', 'e'],
    'col3': [10.5, 20.3, 30.1, 40.2, 50.4]
})

# Create a key (random state seed)
key = np.array([42])

# Define arguments for the function
ARGS = {
    'key': key,
    'df': df,
    'p_remove': 0.2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
to_list_or_float Ran
ran on a synthesized input
google-research/hyperbo/hyperbo/basics/params_utils.py
code served (permissive licence) · get_code("4d555f42b48f0b81")
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 list, ndim 2, 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 jax.numpy as jnp
import numpy as np

# Create test cases for different input types
x_numpy_array = np.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
x_jax_array = jnp.array([[1.0, 2.0], [3.0, 4.0]])
x_numpy_float32 = np.float32(3.14)
x_numpy_float64 = np.float64(2.71)
x_python_float = 1.5
x_python_int = 42
x_python_string = "test"

# The function takes a single parameter 'x'
# We'll test with a numpy array as the primary case
ARGS = {
    "x": x_numpy_array
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family float — a list output has no array shape of its own; none is stated
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
bfgs Not yet run
unverified (harvested, no recorded run)
google-research/hyperbo/hyperbo/basics/bfgs.py
code served (permissive licence) · get_code("725cd249591d5b64")
How it ran

no invocation recorded.

lbfgs Not yet run
unverified (harvested, no recorded run)
google-research/hyperbo/hyperbo/basics/lbfgs.py
code served (permissive licence) · get_code("b268783ba1900dfd")
How it ran

no invocation recorded.

load_from_file Not yet run
unverified (harvested, no recorded run)
google-research/hyperbo/hyperbo/basics/params_utils.py
code served (permissive licence) · get_code("0612e549e9cc4d08")
How it ran

no invocation recorded.

log_dataset Not yet run
unverified (harvested, no recorded run)
google-research/hyperbo/hyperbo/basics/data_utils.py
code served (permissive licence) · get_code("c932d5b4b4bd1859")
How it ran

no invocation recorded.

solve_linear_system Not yet run
unverified (harvested, no recorded run)
google-research/hyperbo/hyperbo/basics/linalg.py
code served (permissive licence) · get_code("0a6dd267fa7a490c")
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
expected_improvement_sub 1 other paper
1 dated earlier
2109.08215 (2021-09)
Pre-trained Gaussian Processes for Bayesian Optimization
sample_dataframe 1 other paper
1 dated earlier
2109.08215 (2021-09)
Pre-trained Gaussian Processes for Bayesian Optimization
solve_linear_system 1 other paper
1 dated earlier
2109.08215 (2021-09)
Pre-trained Gaussian Processes for Bayesian Optimization

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

Bayesian optimization (BO) has become a popular strategy for global optimization of many expensive real-world functions. Contrary to a common belief that BO is suited to optimizing black-box functions, it actually requires domain knowledge on characteristics of those functions to deploy BO successfully. Such domain knowledge often manifests in Gaussian process priors that specify initial beliefs on functions. However, even with expert knowledge, it is not an easy task to select a prior. This is especially true for hyperparameter tuning problems on complex machine learning models, where landscapes of tuning objectives are often difficult to comprehend. We seek an alternative practice for setting these functional priors. In particular, we consider the scenario where we have data from similar functions that allow us to pre-train a tighter distribution a priori. To verify our approach in realistic model training setups, we collected a large multi-task hyperparameter tuning dataset by training tens of thousands of configurations of near-state-of-the-art models on popular image and text datasets, as well as a protein sequence dataset. Our results show that on average, our method is able to locate good hyperparameters at least 3 times more efficiently than the best competing methods.

For agents

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

get_harvested_code_for_paper("2207.03084")
get_code_for_paper("2207.03084")
have("2207.03084")

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

curl https://syntology.ai/api/ran/2207.03084.json

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

[![Syntology run record](https://syntology.ai/api/ran/2207.03084.svg)](https://syntology.ai/paper/2207.03084)

Connect an agent — have() is free.