SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1911.00002 · 2019

Continual Multi-task Gaussian Processes

arXiv · PDF · Open in the Atlas

Code that ran

7 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; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 4 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/1911.00002.json · badge: /api/ran/1911.00002.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
pmorenoz/ContinualGP canonical (classifier label)
classifier basis: classifier_2025
4 ran · 0 no recorded run
DennisReddyhoff/ContinualGP community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
data_streaming Ran
ran; our contract draft was wrong, not the code
pmorenoz/ContinualGP/experiments/single_output.py
code served (permissive licence) · get_code("e1afc46f94e2fe56")
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

# Create small numpy arrays for the function
# Y_output: shape (N, output_dim) - output data
# X_input: shape (N, input_dim) - input data
# T_sections: number of time sections to split into

N = 2000
output_dim = 5
input_dim = 3

Y_output = np.random.randn(N, output_dim).astype(np.float32)
X_input = np.random.randn(N, input_dim).astype(np.float32)
T_sections = 4

ARGS = {
    "Y_output": Y_output,
    "X_input": X_input,
    "T_sections": T_sections
}
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
data_streaming Ran
ran on a synthesized input
DennisReddyhoff/ContinualGP/experiments/banana.py
code served (permissive licence) · get_code("acfe6b6a4de22c6b")
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

# Create small numpy arrays for testing
np.random.seed(0)

# Y_output: shape (N, output_dim) - will be split into T_sections
N = 8
output_dim = 3
Y_output = np.random.randn(N, output_dim)

# X_input: shape (N, input_dim) - will be split into T_sections
input_dim = 4
X_input = np.random.randn(N, input_dim)

# T_sections: number of sections to split into
T_sections = 2

ARGS = {
    "Y_output": Y_output,
    "X_input": X_input,
    "T_sections": T_sections
}
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_batch_scales Ran
ran on a synthesized input
DennisReddyhoff/ContinualGP/continualgp/util.py
code served (permissive licence) · get_code("0f5bc927b86a68d8")
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 list, 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 numpy as np

torch.manual_seed(0)

# The function iterates over X_all and X in parallel,
# accessing .shape[0] on each element to compute a ratio.
# X_all and X should be lists of tensors or array-like objects.

# Create sample data:
# X_all: list of tensors with different batch sizes (larger batches)
# X: list of tensors with smaller batch sizes (for computing the ratio)

X_all = [
    torch.randn(8, 4),   # batch size 8
    torch.randn(6, 4),   # batch size 6
    torch.randn(10, 4),  # batch size 10
]

X = [
    torch.randn(2, 4),   # batch size 2
    torch.randn(3, 4),   # batch size 3
    torch.randn(5, 4),   # batch size 5
]

ARGS = {
    "X_all": X_all,
    "X": X,
}
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 list, ndim 1, dtype family float — a list 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
true_f Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pmorenoz/ContinualGP/experiments/single_output.py
code served (permissive licence) · get_code("5710ef8624a56244")
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 list, ndim 3, 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

# Analyze the function: true_f takes X_input (a numpy array)
# and performs operations on it, returning a list containing one array

# X_input needs to be a 2D numpy array since it uses X_input.shape[0]
# and indexing with X_input in trigonometric operations

# Create a small sample input array
# batch size = 2, feature dim = 1 (or can be larger, but 1D is typical for univariate function)
X_input = np.random.randn(2, 1)

ARGS = {
    "X_input": X_input
}
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 list, ndim 3, dtype family float — a list 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
true_f_functions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pmorenoz/ContinualGP/experiments/multi_output.py
code served (permissive licence) · get_code("09d22ebfe0d73e9f")
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

# Analyze the function to understand what parameters are needed
# true_f_functions(true_u, X_list)
# - true_u: appears to be indexed as true_u[d][:, q, None] where d ranges 0-1, q ranges 0-1
#   This suggests true_u is a list of 2 arrays, each with shape (batch, 2)
# - X_list: appears to be indexed as X_list[d].shape[0] where d ranges 0-1
#   This suggests X_list is a list of 2 arrays where we need their first dimension (batch size)

# Construct true_u: list of 2 arrays, each with shape (batch_size, 2)
# Using batch_size=3 to have a small but non-trivial size
batch_size = 3
true_u = [
    np.random.randn(batch_size, 2),  # d=0: shape (3, 2)
    np.random.randn(batch_size, 2),  # d=1: shape (3, 2)
]

# Construct X_list: list of 2 arrays with different batch dimensions
# We only need X_list[d].shape[0], so we can make simple arrays
X_list = [
    np.random.randn(batch_size, 4),  # d=0: shape (3, 4)
    np.random.randn(batch_size, 5),  # d=1: shape (3, 5)
]

ARGS = {
    "true_u": true_u,
    "X_list": X_list,
}
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
true_u_functions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pmorenoz/ContinualGP/experiments/multi_output.py
code served (permissive licence) · get_code("87c9940b3d5c99d1")
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 list, ndim 3, 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

N = 2000
max_X = 2.0

X1 = np.sort(max_X * np.random.rand(N))[:, None]
X2 = np.sort(max_X * np.random.rand(N))[:, None]

X_list = [X1, X2]

ARGS = {
    "X_list": X_list
}
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 list, ndim 3, dtype family float — a list 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
true_u_functions Ran
ran on a synthesized input
DennisReddyhoff/ContinualGP/continualgp/util.py
code served (permissive licence) · get_code("1e686d574aa04e8d")
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 list. 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

np.random.seed(0)

# X_list: a list of numpy arrays, each representing spatial points
# The function iterates over X_list and for each X, creates u_task with shape (X.shape[0], Q)
# X should be 1D or have a first dimension that represents number of points

X1 = np.linspace(0, 1, 5).reshape(-1, 1)  # shape (5, 1)
X2 = np.linspace(0, 1, 3).reshape(-1, 1)  # shape (3, 1)
X_list = [X1, X2]

# Q: number of tasks/outputs
Q = 2

ARGS = {
    "X_list": X_list,
    "Q": Q
}
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 list — a list 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
true_f_functions Not yet run
unverified (harvested, no recorded run)
DennisReddyhoff/ContinualGP/continualgp/util.py
code served (permissive licence) · get_code("e1c0f6a2a17c4ddb")
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 7 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 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.

FunctionAlso present inEarliest-datedOthers named
get_batch_scales 2 other papers
2 dated earlier
1805.07633 (2018-05)
Heterogeneous Multi-output Gaussian Process Prediction
1906.09412 (2019-06)
Multi-task Learning for Aggregated Data using Gaussian Processes
true_f_functions 2 other papers
2 dated earlier
1805.07633 (2018-05)
Heterogeneous Multi-output Gaussian Process Prediction
1906.09412 (2019-06)
Multi-task Learning for Aggregated Data using Gaussian Processes
true_u_functions 2 other papers
2 dated earlier
1805.07633 (2018-05)
Heterogeneous Multi-output Gaussian Process Prediction
1906.09412 (2019-06)
Multi-task Learning for Aggregated Data using Gaussian Processes

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 address the problem of continual learning in multi-task Gaussian process (GP) models for handling sequential input-output observations. Our approach extends the existing prior-posterior recursion of online Bayesian inference, i.e.\ past posterior discoveries become future prior beliefs, to the infinite functional space setting of GP. For a reason of scalability, we introduce variational inference together with an sparse approximation based on inducing inputs. As a consequence, we obtain tractable continual lower-bounds where two novel Kullback-Leibler (KL) divergences intervene in a natural way. The key technical property of our method is the recursive reconstruction of conditional GP priors conditioned on the variational parameters learned so far. To achieve this goal, we introduce a novel factorization of past variational distributions, where the predictive GP equation propagates the posterior uncertainty forward. We then demonstrate that it is possible to derive GP models over many types of sequential observations, either discrete or continuous and amenable to stochastic optimization. The continual inference approach is also applicable to scenarios where potential multi-channel or heterogeneous observations might appear. Extensive experiments demonstrate that the method is fully scalable, shows a reliable performance and is robust to uncertainty error propagation over a plenty of synthetic and real-world datasets.

For agents

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

get_harvested_code_for_paper("1911.00002")
get_code_for_paper("1911.00002")
have("1911.00002")

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

curl https://syntology.ai/api/ran/1911.00002.json

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

[![Syntology run record](https://syntology.ai/api/ran/1911.00002.svg)](https://syntology.ai/paper/1911.00002)

Connect an agent — have() is free.