SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2605.01663 · ICML · 2026

Towards Efficient and Expressive Offline RL via Flow-Anchored Noise-conditioned Q-Learning

Dohyeong Kim, Eshan Balachandar, Zelal Mustafaoglu, Keshav Pingali, Sungyoung Lee

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 failures of our instrument; 1 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/2605.01663.json · badge: /api/ran/2605.01663.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
brianlsy98/FAN community
Papers with Code archive
4 ran · 1 no recorded run
FunctionStatusWhere it lives
ActorVectorField Ran
ran on a synthesized input · constructed an object; no result was computed
brianlsy98/FAN/agents/fan.py
code served (permissive licence) · get_code("c5ada5c7d43ea8f4")
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, ActorVectorField: 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 jax
import jax.numpy as jnp
import flax.linen as nn
from typing import Sequence

def default_init(scale=1.0):
    """Default kernel initializer."""
    return nn.initializers.variance_scaling(scale, 'fan_avg', 'uniform')

class SimpleEncoder(nn.Module):
    """Simple encoder module for testing."""
    output_dim: int
    
    @nn.compact
    def __call__(self, x):
        x = nn.Dense(self.output_dim)(x)
        return x

# Create the ARGS dictionary with all required parameters for ActorVectorField
ARGS = {
    'hidden_dims': (8, 16),  # Sequence of hidden layer dimensions
    'action_dim': 3,          # Action dimension
    'layer_norm': False,      # Whether to apply layer normalization
    'encoder': SimpleEncoder(output_dim=5),  # Optional encoder module
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ActorVectorField — 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 ActorVectorField 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
ImpalaEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
brianlsy98/FAN/agents/fan.py
code served (permissive licence) · get_code("5c2823b42170fed6")
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, ImpalaEncoder: 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 jax
import jax.numpy as jnp
from flax import linen as nn
from typing import Sequence

# Create a small input tensor (batch=2, height=4, width=4, channels=3 for RGB image)
x = jnp.ones((2, 4, 4, 3), dtype=jnp.uint8)

ARGS = {
    'width': 1,
    'stack_sizes': (16, 32, 32),
    'num_blocks': 2,
    'dropout_rate': None,
    'mlp_hidden_dims': (512,),
    'layer_norm': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ImpalaEncoder — 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 ImpalaEncoder 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
MLP Ran
ran on a synthesized input · constructed an object; no result was computed
brianlsy98/FAN/agents/fan.py
code served (permissive licence) · get_code("c10b91c12ad9a306")
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, MLP: 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 jax
import jax.numpy as jnp
import flax.linen as nn

def default_init(scale=1.0):
    """Default kernel initializer."""
    return nn.initializers.variance_scaling(scale, 'fan_avg', 'uniform')

ARGS = {
    'hidden_dims': (8, 16, 4),
    'activations': nn.gelu,
    'activate_final': False,
    'kernel_init': default_init(),
    'layer_norm': False,
    'dropout_rate': 0.1,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MLP — 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 MLP 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
Value Ran
ran on a synthesized input · constructed an object; no result was computed
brianlsy98/FAN/agents/fan.py
code served (permissive licence) · get_code("9235803e8ce9944a")
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, Value: 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 jax
import jax.numpy as jnp
import flax.linen as nn
from typing import Sequence

def default_init(scale=1.0):
    """Default kernel initializer."""
    return nn.initializers.variance_scaling(scale, 'fan_avg', 'uniform')

def ensemblize(cls, num_qs, in_axes=None, out_axes=0, **kwargs):
    """Ensemblize a module."""
    return nn.vmap(cls, variable_axes={'params': 0, 'intermediates': 0}, split_rngs={'params': True}, in_axes=in_axes, out_axes=out_axes, axis_size=num_qs, **kwargs)

class MLP(nn.Module):
    """Multi-layer perceptron.

    Attributes:
        hidden_dims: Hidden layer dimensions.
        activations: Activation function.
        activate_final: Whether to apply activation to the final layer.
        kernel_init: Kernel initializer.
        layer_norm: Whether to apply layer normalization.
    """
    hidden_dims: Sequence[int]
    activations: any = nn.gelu
    activate_final: bool = False
    kernel_init: any = default_init()
    layer_norm: bool = False
    dropout_rate: float = 0.0

    @nn.compact
    def __call__(self, x, training=False):
        for i, size in enumerate(self.hidden_dims):
            x = nn.Dense(size, kernel_init=self.kernel_init)(x)
            if i + 1 < len(self.hidden_dims) or self.activate_final:
                x = self.activations(x)
                if self.layer_norm:
                    x = nn.LayerNorm()(x)
                if self.dropout_rate > 0.0:
                    x = nn.Dropout(rate=self.dropout_rate)(x, deterministic=not training)
            if i == len(self.hidden_dims) - 2:
                self.sow('intermediates', 'feature', x)
        return x

class Value(nn.Module):
    """Value/critic network.

    This module can be used for both value V(s, g) and critic Q(s, a, g) functions.

    Attributes:
        hidden_dims: Hidden layer dimensions.
        value_dim: Value dimension.
        layer_norm: Whether to apply layer normalization.
        num_ensembles: Number of ensemble components.
        encoder: Optional encoder module to encode the inputs.
    """
    hidden_dims: Sequence[int]
    value_dim: int = 1
    layer_norm: bool = True
    num_ensembles: int = 2
    encoder: nn.Module = None

    def setup(self):
        mlp_class = MLP
        if self.num_ensembles > 1:
            mlp_class = ensemblize(mlp_class, self.num_ensembles)
        value_net = mlp_class((*self.hidden_dims, self.value_dim), activate_final=False, layer_norm=self.layer_norm)
        self.value_net = value_net

    def __call__(self, observations, actions=None, noises=None):
        """Return values or critic values.

        Args:
            observations: Observations.
            actions: Actions (optional).
        """
        if self.encoder is not None:
            inputs = [self.encoder(observations)]
        else:
            inputs = [observations]
        if actions is not None:
            inputs.append(actions)
        if noises is not None:
            inputs.append(noises)
        inputs = jnp.concatenate(inputs, axis=-1)
        v = self.value_net(inputs)
        if self.value_dim == 1:
            v = v.squeeze(-1)
        return v

class SimpleEncoder(nn.Module):
    """Simple encoder that projects input."""
    output_dim: int

    @nn.compact
    def __call__(self, x):
        x = nn.Dense(self.output_dim)(x)
        return x

ARGS = {
    'hidden_dims': (8, 6),
    'value_dim': 1,
    'layer_norm': True,
    'num_ensembles': 2,
    'encoder': SimpleEncoder(output_dim=5),
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Value — 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 Value 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
FANAgent Not yet run
unverified (harvested, no recorded run)
brianlsy98/FAN/agents/fan.py
code served (permissive licence) · get_code("2f9a4dc27b5290d0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 1 the same month, 1 later) and 2 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
ImpalaEncoder
brick of this paper's assembly: helper
9 other papers
7 dated earlier, 1 the same month, 1 later · 5 only as an assembly brick
2406.09329 (2024-06)
Is Value Learning Really the Main Bottleneck in Offline RL?
2502.02538 (2025-02)
Flow Q-Learning
2506.07744 (2025-06)
Graph-Assisted Stitching for Offline Hierarchical Reinforcement Learning
and 6 more not named here
ActorVectorField
brick of this paper's assembly: helper
3 other papers
3 dated earlier · 2 only as an assembly brick
2502.02538 (2025-02)
Flow Q-Learning
2602.18015 (2026-02)
Flow Actor-Critic for Offline Reinforcement Learning
MLP
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2510.07650 (2025-10)
Value Flows

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

We propose Flow-Anchored Noise-conditioned Q-Learning (FAN), a highly efficient and highperforming offline reinforcement learning (RL) algorithm. Recent work has shown that expressive flow policies and distributional critics improve offline RL performance, but at a high computational cost. Specifically, flow policies require iterative sampling to produce a single action, and distributional critics require computation over multiple samples (e.g., quantiles) to estimate value. To address these inefficiencies while maintaining high performance, we introduce FAN. Our method employs a behavior regularization technique that uses a single flow policy iteration and requires a single Gaussian noise sample for distributional critics. Our theoretical analysis of convergence and performance bounds demonstrates that these simplifications not only improve efficiency but also lead to superior task performance. Experiments on robotic manipulation and locomotion tasks demonstrate that FAN achieves state-of-theart performance while significantly reducing both training and inference runtimes. We release our code at https://github.com/brianlsy98/FAN.

For agents

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

get_harvested_code_for_paper("2605.01663")
get_code_for_paper("2605.01663")
have("2605.01663")

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

curl https://syntology.ai/api/ran/2605.01663.json

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

[![Syntology run record](https://syntology.ai/api/ran/2605.01663.svg)](https://syntology.ai/paper/2605.01663)

Connect an agent — have() is free.