SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2009.02010 · 2020

ConfuciuX: Autonomous Hardware Resource Assignment for DNN Accelerators using Reinforcement Learning

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; 1 honoured the contract we drafted, 0 violated it and 5 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/2009.02010.json · badge: /api/ran/2009.02010.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
maestro-project/confuciux canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
crossover Ran
ran on a synthesized input
maestro-project/confuciux/src/ConfX/ga_confx.py
code served (permissive licence) · get_code("eca45758227c3dbe")
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 (5, 4, 2) 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 numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create sample parents array
# Shape: (num_parents, num_layers, 2)
# Using 3 parents with 4 layers and 2 values per layer
parents = np.random.randn(3, 4, 2).astype(np.float32)

# Define offspring_size
offspring_size = 5

# Define num_layers
num_layers = 4

# Define eps
eps = 1

ARGS = {
    "parents": parents,
    "offspring_size": offspring_size,
    "num_layers": num_layers,
    "eps": eps,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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 (5, 4, 2) — shape (5, 4, 2) 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_action_space Ran
ran on a synthesized input
maestro-project/confuciux/src/utils/get_action_space.py
code served (permissive licence) · get_code("7b839c5ef9b57e9b")
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 numpy as np

def get_action_space(use_default=True):
    if use_default:
        action_size = 12
        action_space = [np.array([1, 2, 4, 8, 12, 16, 24, 32, 48, 64, 96, 128]), np.array([i + 1 for i in range(action_size)])]
    else:
        raise NameError('Please define the customized action space.')
    action_bound = [np.max(a) for a in action_space]
    action_bottom = [np.min(a) for a in action_space]
    action_space = [action_space[i] / action_bound[i] for i in range(len(action_space))]
    return (action_space, action_bound, action_bottom)

ARGS = {
    "use_default": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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_population_fitness Ran
ran on a synthesized input
maestro-project/confuciux/src/ConfX/ga_confx.py
code served (permissive licence) · get_code("11a8a4f8188bf669")
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 (2,) 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 numpy as np

# Create equation_inputs: shape (4,) - coefficients for the equation
equation_inputs = np.array([1.0, 2.0, 3.0, 4.0])

# Create pop: shape (2, 4) - population of 2 individuals with 4 genes each
pop = np.array([
    [0.5, 1.0, 1.5, 2.0],
    [2.5, 3.0, 3.5, 4.0]
], dtype=np.float32)

ARGS = {
    "equation_inputs": equation_inputs,
    "pop": pop
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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 (2,) — shape (2,) 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
hidden_init Ran
ran, honoured the contract we drafted
maestro-project/confuciux/src/ConfX/rl_confx.py
code served (permissive licence) · get_code("24ee38e114b38bce")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# The function accesses layer.weight.data.size()[0]
# So we need a layer with a weight attribute that has a .data attribute
# and .size() method that returns something indexable

# Create a simple linear layer
layer = nn.Linear(in_features=5, out_features=3)

ARGS = {
    'layer': layer
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy, torch — 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
mutation Ran
ran on a synthesized input
maestro-project/confuciux/src/Other_Opts/other_opt_env.py
code served (permissive licence) · get_code("66506442dd31d0e2")
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 int, 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 torch
import numpy as np
import random

# Set seed for reproducibility
random.seed(42)
torch.manual_seed(0)

# The function expects offsprings to be a 3D array-like object
# with shape (num_offsprings, num_layers, num_parameters)
# It iterates through each offspring, layer, and parameter
# and randomly mutates values to random integers in range [0, 11]

# Create a small 3D numpy array for offsprings
# Shape: (2 offsprings, 3 layers, 4 parameters)
offsprings = np.random.randint(0, 12, size=(2, 3, 4), dtype=np.int32)

# num_layers is used in the loop but seems redundant since it's inferred from shape
num_layers = 3

# rate is the mutation probability (default 0.05)
rate = 0.05

ARGS = {
    "offsprings": offsprings,
    "num_layers": num_layers,
    "rate": rate
}
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 ndarray, ndim 3, dtype family int, 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
select_parents Ran
ran on a synthesized input
maestro-project/confuciux/src/ConfX/ga_confx.py
code served (permissive licence) · get_code("d5fe4c86b97bcd9b")
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 (2,) 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 numpy as np
import copy
import random

# Set seed for reproducibility
np.random.seed(0)
random.seed(0)

# Create a small population of "individuals"
# Each individual is a simple list or dict representing a neural network
pop = [
    {'weights': np.random.randn(2, 3)},
    {'weights': np.random.randn(2, 3)},
    {'weights': np.random.randn(2, 3)},
    {'weights': np.random.randn(2, 3)},
]

# Convert to numpy array so it can be indexed/reordered
pop = np.array(pop)

# Create fitness scores for each individual
# Shape: (num_individuals, 1) - the function accesses fitness[:, 0]
fitness = np.array([
    [0.8],
    [0.5],
    [0.9],
    [0.6],
])

# Number of parents to select
num_parents = 2

# Number of layers (used in the function signature but not in the body)
num_layers = 3

ARGS = {
    'pop': pop,
    'fitness': fitness,
    'num_parents': num_parents,
    'num_layers': num_layers,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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 (2,) — shape (2,) 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
crossover Not yet run
unverified (harvested, no recorded run)
maestro-project/confuciux/src/Other_Opts/other_opt_env.py
code served (permissive licence) · get_code("504c7d524f5c0f44")
How it ran

no invocation recorded.

select_parents Not yet run
unverified (harvested, no recorded run)
maestro-project/confuciux/src/Other_Opts/other_opt_env.py
code served (permissive licence) · get_code("fd2141ce08b3d8ff")
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, 1 are also present in other papers whose code we hold (1 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
hidden_init 11 other papers
6 dated earlier, 5 later · 5 only as an assembly brick
1509.02971 (2015-09)
Continuous control with deep reinforcement learning
1602.01783 (2016-02)
Asynchronous Methods for Deep Reinforcement Learning
1706.02275 (2017-06)
Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments
and 8 more not named here

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

DNN accelerators provide efficiency by leveraging reuse of activations/weights/outputs during the DNN computations to reduce data movement from DRAM to the chip. The reuse is captured by the accelerator's dataflow. While there has been significant prior work in exploring and comparing various dataflows, the strategy for assigning on-chip hardware resources (i.e., compute and memory) given a dataflow that can optimize for performance/energy while meeting platform constraints of area/power for DNN(s) of interest is still relatively unexplored. The design-space of choices for balancing compute and memory explodes combinatorially, as we show in this work (e.g., as large as O(10^(72)) choices for running \mobilenet), making it infeasible to do manual-tuning via exhaustive searches. It is also difficult to come up with a specific heuristic given that different DNNs and layer types exhibit different amounts of reuse. In this paper, we propose an autonomous strategy called ConfuciuX to find optimized HW resource assignments for a given model and dataflow style. ConfuciuX leverages a reinforcement learning method, REINFORCE, to guide the search process, leveraging a detailed HW performance cost model within the training loop to estimate rewards. We also augment the RL approach with a genetic algorithm for further fine-tuning. ConfuciuX demonstrates the highest sample-efficiency for training compared to other techniques such as Bayesian optimization, genetic algorithm, simulated annealing, and other RL methods. It converges to the optimized hardware configuration 4.7 to 24 times faster than alternate techniques.

For agents

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

get_harvested_code_for_paper("2009.02010")
get_code_for_paper("2009.02010")
have("2009.02010")

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

curl https://syntology.ai/api/ran/2009.02010.json

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

[![Syntology run record](https://syntology.ai/api/ran/2009.02010.svg)](https://syntology.ai/paper/2009.02010)

Connect an agent — have() is free.