SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.03704 · 2019

SVRG for Policy Evaluation with Fewer Gradient Evaluations

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 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 3 have no contract verdict; 0 failures of our instrument; 1 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/1906.03704.json · badge: /api/ran/1906.03704.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
zilunpeng/svrg_for_policy_evaluation_with_fewer_gradients community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
calc_mspbe_torch Ran
ran on a synthesized input
zilunpeng/svrg_for_policy_evaluation_with_fewer_gradients/mspbe.py
code served (permissive licence) · get_code("cb5c7fe6e8682e01")
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)

class Experiment:
    def __init__(self):
        # A is a matrix used in torch.mv(exp.A, exp.theta)
        # theta is a vector, so A should be compatible for matrix-vector multiplication
        self.A = torch.randn(3, 4)
        
        # theta is a vector
        self.theta = torch.randn(4)
        
        # b is a vector that matches the output size of torch.mv(exp.A, exp.theta)
        self.b = torch.randn(3)
        
        # C_inv is used in torch.mv(exp.C_inv, A_theta_minus_b)
        # A_theta_minus_b has size 3, so C_inv should be (3, 3)
        self.C_inv = torch.randn(3, 3)
        
        # rho_omega is a scalar multiplier
        self.rho_omega = torch.tensor(0.5)
        
        # omega is a vector that matches the size of theta for the dot product
        self.omega = torch.randn(4)

exp = Experiment()
rho = 0.1

ARGS = {
    "exp": exp,
    "rho": rho
}
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
mspbe_grad_omega Ran
ran on a synthesized input
zilunpeng/svrg_for_policy_evaluation_with_fewer_gradients/mspbe.py
code served (permissive licence) · get_code("2767cc415ecf7032")
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 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 torch

torch.manual_seed(0)

# Analyze the function:
# - theta: used in torch.mv(A, theta), so needs to be a 1D tensor
# - omega: used in torch.mv(C, omega) and as scalar multiplication (omega * rho_omega), so 1D tensor
# - A: used in torch.mv(A, theta), so needs to be a 2D tensor (matrix)
# - b: subtracted from result of torch.mv, so 1D tensor
# - C: used in torch.mv(C, omega), so 2D tensor (matrix)
# - rho_omega: scalar multiplied with omega, so scalar or 0D tensor

# torch.mv(A, theta) expects A to be shape (m, n) and theta to be shape (n,)
# torch.mv returns shape (m,)
# For distinct dimensions: let's use m=4, n=3

# theta: shape (3,)
theta = torch.randn(3)

# omega: shape (4,) - same dimension as output of torch.mv(A, theta)
omega = torch.randn(4)

# A: shape (4, 3) for torch.mv(A, theta) where theta is (3,)
A = torch.randn(4, 3)

# b: shape (4,) - subtracted from result of torch.mv(A, theta)
b = torch.randn(4)

# C: shape (4, 4) for torch.mv(C, omega) where omega is (4,)
C = torch.randn(4, 4)

# rho_omega: scalar
rho_omega = torch.tensor(0.5)

ARGS = {
    "theta": theta,
    "omega": omega,
    "A": A,
    "b": b,
    "C": C,
    "rho_omega": rho_omega,
}
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 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
mspbe_grad_theta Ran
ran on a synthesized input
zilunpeng/svrg_for_policy_evaluation_with_fewer_gradients/mspbe.py
code served (permissive licence) · get_code("d5eb28340e57ce2c")
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 1, dtype family float, shape (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 torch

torch.manual_seed(0)

# theta: scalar or tensor, multiplied by rho
# omega: vector, used in torch.mv with transposed A
# A: matrix, transposed and used in torch.mv with omega
# rho: scalar or tensor, multiplied with theta

theta = torch.tensor(2.0)
omega = torch.randn(4)
A = torch.randn(4, 3)
rho = torch.tensor(0.5)

ARGS = {
    "theta": theta,
    "omega": omega,
    "A": A,
    "rho": rho,
}
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 1, dtype family float, shape (3,) — shape (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
get_pi_env Not yet run
unverified (harvested, no recorded run)
zilunpeng/svrg_for_policy_evaluation_with_fewer_gradients/agent_env.py
code served (permissive licence) · get_code("3031d42074b687f2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Stochastic variance-reduced gradient (SVRG) is an optimization method originally designed for tackling machine learning problems with a finite sum structure. SVRG was later shown to work for policy evaluation, a problem in reinforcement learning in which one aims to estimate the value function of a given policy. SVRG makes use of gradient estimates at two scales. At the slower scale, SVRG computes a full gradient over the whole dataset, which could lead to prohibitive computation costs. In this work, we show that two variants of SVRG for policy evaluation could significantly diminish the number of gradient calculations while preserving a linear convergence speed. More importantly, our theoretical result implies that one does not need to use the entire dataset in every epoch of SVRG when it is applied to policy evaluation with linear function approximation. Our experiments demonstrate large computational savings provided by the proposed methods.

For agents

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

get_harvested_code_for_paper("1906.03704")
get_code_for_paper("1906.03704")
have("1906.03704")

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

curl https://syntology.ai/api/ran/1906.03704.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.03704.svg)](https://syntology.ai/paper/1906.03704)

Connect an agent — have() is free.