SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.17685 · 2026

Flow Matching Policy Optimization with Mirror Descent and Entropy Constraints

Nan Lin, Serge Hoogendoorn, Winnie Daamen, Stavros Orfanoudakis, Ting Gao, Elvin Isufi

arXiv · PDF · Open in the Atlas

Code that ran

23 of 29 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (5 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 7 have no contract verdict; 13 failures of our instrument; 6 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/2603.17685.json · badge: /api/ran/2603.17685.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
lzqw/FLAME canonical (classifier label)
classifier basis: census
15 ran · 0 no recorded run
wadx2019/qvpo community
Papers with Code archive
8 ran · 3 no recorded run
Elessar123/SAC-FLOW community
Papers with Code archive
0 ran · 3 no recorded run
FunctionStatusWhere it lives
Critic Ran
ran on a synthesized input · constructed an object; no result was computed
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("e3e7ea9dea530ccd")
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, Critic: 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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

ARGS = {
    'state_dim': 4,
    'action_dim': 3,
    'hidden_dim': 256,
}
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 Critic — 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 Critic 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
CriticQR Ran
ran on a synthesized input · constructed an object; no result was computed
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("56ea6cda0fdbbddd")
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, CriticQR: 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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

ARGS = {
    "state_dim": 4,
    "action_dim": 3,
    "hidden_dim": 256,
    "n_quantiles": 25,
    "n_nets": 5,
}
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 CriticQR — 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 CriticQR 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
Diffusion Ran
ran on a synthesized input · constructed an object; no result was computed
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("0c81dda4b2c07710")
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, Diffusion: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 torch.nn as nn
import math

torch.manual_seed(0)

class SinusoidalPosEmb(nn.Module):
    def __init__(self, dim):
        super().__init__()
        self.dim = dim
        half_dim = dim // 2
        scale = math.log(10000) / (half_dim - 1)
        self.register_buffer('frequencies', torch.exp(torch.arange(half_dim, dtype=torch.float32) * -scale), persistent=False)

    def forward(self, x):
        emb = x[:, None] * self.frequencies[None, :]
        emb = torch.cat((emb.sin(), emb.cos()), dim=-1)
        return emb

def init_weights(m):
    def truncated_normal_init(t, mean=0.0, std=0.01):
        torch.nn.init.normal_(t, mean=mean, std=std)
        while True:
            cond = torch.logical_or(t < mean - 2 * std, t > mean + 2 * std)
            if not torch.sum(cond):
                break
            t = torch.where(cond, torch.nn.init.normal_(torch.ones(t.shape), mean=mean, std=std), t)
        return t
    if type(m) == nn.Linear:
        input_dim = m.in_features
        truncated_normal_init(m.weight, std=1 / (2 * math.sqrt(input_dim)))
        m.bias.data.fill_(0.0)

class Model(nn.Module):
    def __init__(self, state_dim, action_dim, hidden_size=256, time_dim=32):
        super(Model, self).__init__()
        self.time_mlp = nn.Sequential(SinusoidalPosEmb(time_dim), nn.Linear(time_dim, hidden_size), nn.Mish(), nn.Linear(hidden_size, time_dim))
        input_dim = state_dim + action_dim + time_dim
        self.layer = nn.Sequential(nn.Linear(input_dim, hidden_size), nn.Mish(), nn.Linear(hidden_size, hidden_size), nn.Mish(), nn.Linear(hidden_size, hidden_size), nn.Mish(), nn.Linear(hidden_size, action_dim))
        self.apply(init_weights)

    def forward(self, x, time, state):
        t = self.time_mlp(time)
        out = torch.cat([x, t, state], dim=-1)
        out = self.layer(out)
        return out

ARGS = {
    'state_dim': 4,
    'action_dim': 3,
    'noise_ratio': 0.5,
    'beta_schedule': 'vp',
    'n_timesteps': 100,
    'loss_type': 'l2',
    'clip_denoised': True,
    'predict_epsilon': True,
    'behavior_sample': 16,
    'eval_sample': 512,
    'deterministic': False,
}
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 Diffusion — 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 Diffusion 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
Model Ran
ran on a synthesized input · constructed an object; no result was computed
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("10276c719fb6f855")
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, Model: 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 torch
import torch.nn as nn
import math

torch.manual_seed(0)

ARGS = {
    "state_dim": 4,
    "action_dim": 3,
    "hidden_size": 256,
    "time_dim": 32,
}
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 Model — 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 Model 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
SinusoidalPosEmb Ran
ran on a synthesized input · constructed an object; no result was computed
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("52e215849471bdfc")
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, SinusoidalPosEmb: 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 torch
import torch.nn as nn
import math

torch.manual_seed(0)

# The SinusoidalPosEmb class takes a single parameter: dim
# It expects dim to be an integer representing the embedding dimension
# In forward(), it takes x which should be a 1D tensor of positions

# Create a small valid argument set
ARGS = {
    'dim': 4
}
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 SinusoidalPosEmb — 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 SinusoidalPosEmb 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
WeightedL1 Ran
ran on a synthesized input
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("20545bf1089ab44e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, 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 torch
import torch.nn as nn

class WeightedLoss(nn.Module):

    def __init__(self):
        super().__init__()

    def forward(self, pred, targ, weights=1.0):
        """
            pred, targ : tensor [ batch_size x action_dim ]
        """
        loss = self._loss(pred, targ)
        weighted_loss = (loss * weights).mean()
        return weighted_loss

class WeightedL1(WeightedLoss):

    def _loss(self, pred, targ):
        return torch.abs(pred - targ)


INIT_ARGS = {}

ARGS = {
    'pred': torch.randn(2, 4),
    'targ': torch.randn(2, 4),
    'weights': 1.0
}

if __name__ == '__main__':
    model = WeightedL1(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output: {output}")
    print(f"Output shape: {output.shape}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
WeightedL2 Ran
ran on a synthesized input
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("3b2b8b659d67b0af")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, 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 torch
import torch.nn as nn
import torch.nn.functional as F

class WeightedLoss(nn.Module):

    def __init__(self):
        super().__init__()

    def forward(self, pred, targ, weights=1.0):
        """
            pred, targ : tensor [ batch_size x action_dim ]
        """
        loss = self._loss(pred, targ)
        weighted_loss = (loss * weights).mean()
        return weighted_loss

class WeightedL2(WeightedLoss):

    def _loss(self, pred, targ):
        return F.mse_loss(pred, targ, reduction='none')


INIT_ARGS = {}

ARGS = {
    'pred': torch.randn(2, 4),
    'targ': torch.randn(2, 4),
    'weights': 1.0
}

if __name__ == '__main__':
    model = WeightedL2(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output value: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
build_table Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/make_exp37_tables.py
pointer only (licence: NONE) · get_code("b98ba81a4ea31a61")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

classify Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_exp37_main_figures.py
pointer only (licence: NONE) · get_code("8f52d69d86b15b42")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

entropy2d Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lzqw/FLAME/analysis/plot_action_space_entropy.py
pointer only (licence: NONE) · get_code("1d2ce563da9d7683")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

flatten_rollouts Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_exp37_diagnostics.py
pointer only (licence: NONE) · get_code("54adb6b865c1eab7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

fmt_stat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lzqw/FLAME/analysis/make_exp37_tables.py
pointer only (licence: NONE) · get_code("0a19232d2baba9b2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_json Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_three_obstacle_pullback_results.py
pointer only (licence: NONE) · get_code("5d7dd03a436d96f8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_npz Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_three_obstacle_pullback_results.py
pointer only (licence: NONE) · get_code("0385ed9d7ed29835")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

parse_meta Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/collect_exp37_results.py
pointer only (licence: NONE) · get_code("4fbcb926264f78c1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

parse_path_meta Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_exp37_diagnostics.py
pointer only (licence: NONE) · get_code("502742af75564dca")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

pearson_corr Ran
ran, violated the contract we drafted
lzqw/FLAME/analysis/plot_projection_critic_heatmaps.py
pointer only (licence: NONE) · get_code("abbd0c4947647c0e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

percentile_clip Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lzqw/FLAME/analysis/plot_projection_critic_heatmaps.py
pointer only (licence: NONE) · get_code("1df489cbc7af1f28")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

quantile_huber_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("3e12dfe5d2e56c8d")
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)

# Create small tensors with distinct dimensions
batch_size = 2
num_quantiles = 4
num_samples = 3

# quantiles: shape (batch_size, num_quantiles_groups, num_quantiles)
# The function uses quantiles.shape[2] to get n_quantiles
quantiles = torch.randn(batch_size, 1, num_quantiles)

# samples: shape (batch_size, num_samples)
# Used in broadcasting: samples[:, None, None, :] creates (batch_size, 1, 1, num_samples)
samples = torch.randn(batch_size, num_samples)

ARGS = {
    "quantiles": quantiles,
    "samples": samples,
}
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
require_path Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_double_integrator_pullback_results.py
pointer only (licence: NONE) · get_code("c48ae4d3213b89a9")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

run_rollouts Ran
ran; our contract draft was wrong, not the code
lzqw/FLAME/analysis/plot_curvature_lqr.py
pointer only (licence: NONE) · get_code("e54daa8eee4256d4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

safe_eigvals Ran
ran, honoured the contract we drafted
lzqw/FLAME/analysis/plot_curvature_lqr.py
pointer only (licence: NONE) · get_code("c43c9056d3743694")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

spearman_corr Ran
ran, violated the contract we drafted
lzqw/FLAME/analysis/plot_projection_critic_heatmaps.py
pointer only (licence: NONE) · get_code("7d29cbd982dfe045")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lzqw/FLAME when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Actor Not yet run
unverified (harvested, no recorded run)
Elessar123/SAC-FLOW/from_scratch_code/SAC_flow_gru_jax.py
pointer only (licence: NONE) · get_code("cbe4b2a64b8a77d3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Elessar123/SAC-FLOW when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

BatchRenorm Not yet run
unverified (harvested, no recorded run)
Elessar123/SAC-FLOW/from_scratch_code/SAC_flow_gru_jax.py
pointer only (licence: NONE) · get_code("c7dd064a50cc4845")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Elessar123/SAC-FLOW when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

QVPO Not yet run
unverified (harvested, no recorded run)
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("fe3538c8acd17840")
How it ran

no invocation recorded.

SinusoidalPosEmb Not yet run
unverified (harvested, no recorded run)
Elessar123/SAC-FLOW/from_scratch_code/SAC_flow_gru_jax.py
pointer only (licence: NONE) · get_code("aab29e00b150b7b2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Elessar123/SAC-FLOW when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

WeightedLoss Not yet run
unverified (harvested, no recorded run)
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("728e142be8c63b8b")
How it ran

no invocation recorded.

init_weights Not yet run
unverified (harvested, no recorded run)
wadx2019/qvpo/agent/qvpo.py
code served (permissive licence) · get_code("6f97de74414878c5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 29 samples lifted from this paper's linked repositories, 29 carry a function identity and 0 do not. Of the 29 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 25 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
WeightedL1
brick of this paper's assembly: helper
8 other papers
8 dated earlier · 2 only as an assembly brick
2208.06193 (2022-08)
Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning
2210.15629 (2022-10)
Language Control Diffusion: Efficiently Scaling through Space, Time, and Tasks
2305.18459 (2023-05)
Diffusion Model is an Effective Planner and Data Synthesizer for Multi-Task Reinforcement Learning
and 5 more not named here
WeightedL2
brick of this paper's assembly: helper
8 other papers
8 dated earlier · 2 only as an assembly brick
2208.06193 (2022-08)
Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning
2210.15629 (2022-10)
Language Control Diffusion: Efficiently Scaling through Space, Time, and Tasks
2305.18459 (2023-05)
Diffusion Model is an Effective Planner and Data Synthesizer for Multi-Task Reinforcement Learning
and 5 more not named here
WeightedLoss
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
2208.06193 (2022-08)
Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning
2310.05333 (2023-10)
DiffCPS: Diffusion-based Constrained Policy Search for Offline Reinforcement Learning
init_weights
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2305.13122 (2023-05)
Policy Representation via Diffusion Probability Model for Reinforcement Learning

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

Balancing policy expressiveness with the exploration-exploitation trade-off is a core challenge in online Reinforcement Learning (RL). While Stochastic Differential Equation (SDE)-based diffusion policies can represent complex, multimodal action distributions, they suffer from two critical limitations: their stochastic reverse processes render entropy intractable (necessitating heuristic exploration), and computing policy gradients through long denoising chains is expensive and unstable. In this work, we show that ODE-based flow matching inherently resolves these issues by enabling both simulation-free policy optimization and tractable entropy computation. Building on this, we introduce Flow Matching Policy Optimization with Mirror Descent and Entropy Constraints (FMER). Our framework exploits this insight in three ways. First, we theoretically establish that minimizing an advantage-weighted conditional flow matching loss acts as a simulation-free surrogate for policy mirror descent. This steers the velocity field toward high-value regions while entirely avoiding backpropagation through the ODE solver. Second, we derive an analytic entropy objective that corrects for the density distortion caused by the tanh transformation (mapping an unbounded latent space to bounded actions), thereby facilitating principled maximum-entropy optimization. Finally, we dynamically tune the mirror descent temperature based on the effective sample size to enforce a robust trust region during training. Empirical evaluations demonstrate that FMER achieves superior performance on the challenging sparse-reward FrankaKitchen environment, while maintaining competitive results across standard dense-reward MuJoCo benchmarks. Code is available at https://anonymous.4open.science/r/submission-25E2/.

For agents

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

get_harvested_code_for_paper("2603.17685")
get_code_for_paper("2603.17685")
have("2603.17685")

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

curl https://syntology.ai/api/ran/2603.17685.json

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

[![Syntology run record](https://syntology.ai/api/ran/2603.17685.svg)](https://syntology.ai/paper/2603.17685)

Connect an agent — have() is free.