SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.01047 · 2026

Evaluating GFlowNet from partial episodes for stable and flexible policy-based training

Xiaoning Qian, Puhua Niu, Shili Wu

arXiv · PDF · Open in the Atlas

Code that ran

4 of 7 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; 3 failures of our instrument; 3 no recorded run), measured 2026-11-22; 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/2603.01047.json · badge: /api/ran/2603.01047.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
niupuhua1234/Sub-EB canonical (classifier label)
classifier basis: census
4 ran · 3 no recorded run
FunctionStatusWhere it lives
expected_edges Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
niupuhua1234/Sub-EB/gfn-subeb/data_dag/metrics.py
code served (permissive licence) · get_code("2d3fc4a1e259fd6d")
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 float64, ndim 0, 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 numpy as np

# Create a small posterior approximation array
# Shape: (B=2, N=4, N=4)
# B=2: number of sample graphs
# N=4: number of variables
# Values should be between 0 and 1 (representing edge probabilities)
posterior = np.random.rand(2, 4, 4)

ARGS = {
    "posterior": posterior
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
ix_ Ran
ran; our contract draft was wrong, not the code
niupuhua1234/Sub-EB/gfn-subeb/data_dag/scores/base.py
code served (permissive licence) · get_code("a2b19bef0331f930")
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 numpy as np

torch.manual_seed(0)

# The function ix_ takes an array and returns a tuple of two views
# It adds dimensions to enable broadcasting: (array[..., :, None], array[..., None, :])
# The array can be a numpy array or torch tensor

# Create a small test array with distinct dimensions for probing
array = np.arange(12).reshape(2, 3, 2)

ARGS = {
    "array": array
}
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 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
logdet Ran
ran, honoured the contract we drafted
niupuhua1234/Sub-EB/gfn-subeb/data_dag/scores/base.py
code served (permissive licence) · get_code("0cb1419ec492cafd")
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 float64, ndim 0, 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 numpy as np

# The function logdet takes an array and computes the log determinant
# using np.linalg.slogdet. We need to provide a square matrix.

# Create a small square matrix that is invertible
array = np.array([
    [2.0, 0.5],
    [0.3, 1.5]
], dtype=np.float64)

ARGS = {
    "array": array
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
logdet Ran
ran; our contract draft was wrong, not the code
niupuhua1234/Sub-EB/gfn-subeb/data_dag/scores/base_torch.py
code served (permissive licence) · get_code("52ac2620e1dc4258")
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 0, dtype family float, shape () 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 logdet expects a square matrix (2D tensor) or batch of square matrices
# torch.slogdet computes the sign and log-determinant of a matrix
# We'll create a small square tensor that is invertible

array = torch.randn(4, 4)

ARGS = {
    "array": array
}
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 0, dtype family float, shape () — shape () 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
download Not yet run
unverified (harvested, no recorded run)
niupuhua1234/Sub-EB/gfn-subeb/data_dag/factories.py
code served (permissive licence) · get_code("874cb43146e2c04f")
How it ran

no invocation recorded.

expected_shd Not yet run
unverified (harvested, no recorded run)
niupuhua1234/Sub-EB/gfn-subeb/data_dag/metrics.py
code served (permissive licence) · get_code("89736f5ba822ec0a")
How it ran

no invocation recorded.

threshold_metrics Not yet run
unverified (harvested, no recorded run)
niupuhua1234/Sub-EB/gfn-subeb/data_dag/metrics.py
code served (permissive licence) · get_code("0fca2cf84bdf5862")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 0 later) and 3 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
logdet 3 other papers
3 dated earlier
2202.13903 (2022-02)
Bayesian Structure Learning with Generative Flow Networks
2408.05885 (2024-08)
GFlowNet Training by Policy Gradients
2410.00965 (2024-10)
Causal Discovery in Astrophysics: Unraveling Supermassive Black Hole and Galaxy Coevolution
download 2 other papers
2 dated earlier
2202.13903 (2022-02)
Bayesian Structure Learning with Generative Flow Networks
2408.05885 (2024-08)
GFlowNet Training by Policy Gradients
ix_ 1 other paper
1 dated earlier
2408.05885 (2024-08)
GFlowNet Training by Policy Gradients
logdet 1 other paper
1 dated earlier
2408.05885 (2024-08)
GFlowNet Training by Policy Gradients

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

Generative Flow Networks (GFlowNets) were developed to learn policies for efficiently sampling combinatorial candidates by interpreting their generative processes as trajectories in directed acyclic graphs. In the value-based training workflow, the objective is to enforce the balance over partial episodes between the flows of the learned policy and the estimated flows of the desired policy, implicitly encouraging policy divergence minimization. The policy-based strategy alternates between estimating the policy divergence and updating the policy, but reliable estimation of the divergence under directed acyclic graphs remains a major challenge. This work bridges the two perspectives by showing that flow balance also yields a principled policy evaluator that measures the divergence, and an evaluation balance objective over partial episodes is proposed for learning the evaluator. As demonstrated on both synthetic and real-world tasks, evaluation balance not only strengthens the reliability of policy-based training but also broadens its flexibility by seamlessly supporting parameterized backward policies and enabling the integration of offline data-collection techniques. Our code is available at github.com/niupuhua1234/Sub-EB.

For agents

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

get_harvested_code_for_paper("2603.01047")
get_code_for_paper("2603.01047")
have("2603.01047")

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

curl https://syntology.ai/api/ran/2603.01047.json

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

[![Syntology run record](https://syntology.ai/api/ran/2603.01047.svg)](https://syntology.ai/paper/2603.01047)

Connect an agent — have() is free.