SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2209.06302 · 2022

Optimization without Backpropagation

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 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 6 have no contract verdict; 0 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/2209.06302.json · badge: /api/ran/2209.06302.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
gbelouze/forward-gradient canonical (classifier label)
classifier basis not recorded on the link
6 ran · 2 no recorded run
FunctionStatusWhere it lives
accuracy_profile Ran
ran on a synthesized input
gbelouze/forward-gradient/benchmark/figures.py
code served (permissive licence) · get_code("04ab04ae8695b9d9")
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 numpy as np
import pandas as pd

# Create a sample DataFrame with the required columns
np.random.seed(0)

# Generate sample data
n_rows = 12
data = {
    'expID': ['exp1', 'exp1', 'exp1', 'exp1', 'exp2', 'exp2', 'exp2', 'exp2', 'exp3', 'exp3', 'exp3', 'exp3'],
    'optimizer': ['adam', 'adam', 'sgd', 'sgd', 'adam', 'adam', 'sgd', 'sgd', 'adam', 'adam', 'sgd', 'sgd'],
    'objective': ['func_a', 'func_b', 'func_a', 'func_b', 'func_a', 'func_b', 'func_a', 'func_b', 'func_a', 'func_b', 'func_a', 'func_b'],
    'loss': np.random.uniform(0.001, 0.1, n_rows),
    'loss0': np.random.uniform(0.5, 1.0, n_rows),
}

df = pd.DataFrame(data)

ARGS = {
    'df': df
}
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 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
circle Ran
ran on a synthesized input
gbelouze/forward-gradient/src/forward/samplers.py
code served (permissive licence) · get_code("d6321e8f96e01110")
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 ndarray, ndim 1, dtype family float, shape (5,) 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 autograd.numpy as np

# The function 'circle' takes a 'size' parameter which can be:
# - an int (single dimension)
# - an Iterable[int] (multiple dimensions)
# 
# The function uses:
# - np.random.normal(size=size)
# - np.sqrt(size) - expects size to be convertible to a number
# - np.linalg.norm(ret) - expects an array
#
# When size is an int, np.sqrt(size) works directly
# When size is an iterable, we need to handle it carefully
# Looking at the code: np.sqrt(size) suggests size should be numeric
# The function appears to expect size as an int or will compute on the product

# Test with an integer size
ARGS = {
    "size": 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
autograd — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (5,) 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
df Ran
ran on a synthesized input
gbelouze/forward-gradient/src/forward/grad.py
code served (permissive licence) · get_code("fb638feb15a956da")
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 function. 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 autograd.numpy as np
from autograd import make_jvp, make_vjp

# Define a simple optimizable function that takes an ndarray and returns a float
def simple_quadratic(x):
    """A simple quadratic function: sum of squares"""
    return np.sum(x**2)

# Another example: a more complex function
def rosenbrock_like(x):
    """A Rosenbrock-like function"""
    return (1 - x[0])**2 + 100 * (x[1] - x[0]**2)**2 + np.sum(x[2:]**2)

# Create a small test input
test_input = np.array([1.5, 2.5, 0.5, -0.5])

ARGS = {
    "f": simple_quadratic
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
autograd — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
gaussian Ran
ran on a synthesized input
gbelouze/forward-gradient/src/forward/samplers.py
code served (permissive licence) · get_code("f0a5202e373d9d3e")
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 ndarray, ndim 3, dtype family float, shape (2, 3, 4) 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 autograd.numpy as np

# The function takes a 'size' parameter which can be either:
# - an int (single dimension)
# - an iterable of ints (multiple dimensions)
# We'll test with both cases

# Test case 1: size as an int
size_int = 5

# Test case 2: size as an iterable (tuple)
size_tuple = (3, 4)

# We'll use the tuple case as it exercises the more general path
ARGS = {
    "size": (2, 3, 4)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
autograd — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 4) 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
rademacher Ran
ran on a synthesized input
gbelouze/forward-gradient/src/forward/samplers.py
code served (permissive licence) · get_code("9efbf539b61ba248")
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 ndarray, ndim 2, dtype family float, shape (2, 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 autograd.numpy as np

np.random.seed(0)

# The function takes a 'size' parameter which can be int or Iterable[int]
# It uses np.random.randint to generate random values
# We'll provide both test cases: a single int and a tuple of ints

ARGS = {
    "size": (2, 3)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
autograd — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 3) — shape (2, 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
sample_in_domain Ran
ran on a synthesized input
gbelouze/forward-gradient/benchmark/make.py
code served (permissive licence) · get_code("abaf02bbe4183369")
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 ndarray, ndim 1, dtype family float, shape (4,) 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 autograd.numpy as np

# Create a domain array
# domain has shape (d, 2) where d is the number of dimensions
# domain[:, 0] are the lower bounds
# domain[:, 1] are the upper bounds
d = 4
domain = np.array([
    [0.0, 1.0],      # dimension 0: bounds [0, 1]
    [-1.0, 1.0],     # dimension 1: bounds [-1, 1]
    [10.0, 20.0],    # dimension 2: bounds [10, 20]
    [-5.0, 5.0]      # dimension 3: bounds [-5, 5]
])

ARGS = {
    "domain": domain
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
autograd — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (4,) — shape (4,) 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_experiment_files Not yet run
unverified (harvested, no recorded run)
gbelouze/forward-gradient/benchmark/figures.py
code served (permissive licence) · get_code("d34dd42dbb5c0649")
How it ran

no invocation recorded.

read_experiment_files Not yet run
unverified (harvested, no recorded run)
gbelouze/forward-gradient/benchmark/figures.py
code served (permissive licence) · get_code("71265864f5120481")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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

Forward gradients have been recently introduced to bypass backpropagation in autodifferentiation, while retaining unbiased estimators of true gradients. We derive an optimality condition to obtain best approximating forward gradients, which leads us to mathematical insights that suggest optimization in high dimension is challenging with forward gradients. Our extensive experiments on test functions support this claim.

For agents

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

get_harvested_code_for_paper("2209.06302")
get_code_for_paper("2209.06302")
have("2209.06302")

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

curl https://syntology.ai/api/ran/2209.06302.json

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

[![Syntology run record](https://syntology.ai/api/ran/2209.06302.svg)](https://syntology.ai/paper/2209.06302)

Connect an agent — have() is free.