SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.09888 · 2022

Benchmarking the Linear Algebra Awareness of TensorFlow and PyTorch

arXiv · PDF · Open in the Atlas

Code that ran

14 of 19 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 14 have no contract verdict; 0 failures of our instrument; 5 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/2202.09888.json · badge: /api/ran/2202.09888.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
as641651/linearalgebra-awareness-benchmark canonical (classifier label)
classifier basis not recorded on the link
14 ran · 4 no recorded run
chrispsa/lamp_benchmark community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
FunctionStatusWhere it lives
format_cutoff_results_md Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/prepare_report.py
code served (permissive licence) · get_code("22a6a516d56403ba")
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 dict. 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

torch.manual_seed(0)

# Analyzing the function: it iterates over a dict of dicts,
# checking if values are True/False and replacing them with emoji strings.
# The structure is: cutoff_results[exp][test] = True/False
# We need to create a nested dict with experiment names as keys,
# and test results as values (dicts of test_name -> bool)

cutoff_results = {
    "experiment_1": {
        "test_a": True,
        "test_b": False,
        "test_c": True,
    },
    "experiment_2": {
        "test_x": False,
        "test_y": True,
    },
}

ARGS = {
    "cutoff_results": cutoff_results,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
format_floats_recursive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/prepare_report.py
code served (permissive licence) · get_code("126ee0390b236079")
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 dict. 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

# Construct test data with nested dictionaries and float values
data = {
    "metrics": {
        "accuracy": 0.95432,
        "loss": 0.12345,
        "nested_metrics": {
            "precision": 0.98765,
            "recall": 0.91234
        }
    },
    "learning_rate": 0.001,
    "threshold": 0.5,
    "config": {
        "param1": 1.23456,
        "param2": 9.87654
    }
}

ARGS = {
    "data": data,
    "precision": 2
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
linalg_matmul Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_blocked.py
code served (permissive licence) · get_code("c5b298b754b9fb33")
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 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 torch

torch.manual_seed(0)

# Create two small tensors for matrix multiplication
# A: shape (2, 4, 3) - batch of 2, 4x3 matrices
# B: shape (2, 3, 5) - batch of 2, 3x5 matrices
# Result will be (2, 4, 5)
A = torch.randn(2, 4, 3)
B = torch.randn(2, 3, 5)

ARGS = {
    "A": A,
    "B": B,
}
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 3, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
operator Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_blocked.py
code served (permissive licence) · get_code("227aac16367e955a")
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 2, 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

torch.manual_seed(0)

# The operator function performs matrix multiplication: A @ B
# Both A and B must be tensors that can be multiplied together
# We'll create two small 2D tensors with compatible dimensions

A = torch.randn(2, 4)  # Shape: (2, 4)
B = torch.randn(4, 3)  # Shape: (4, 3) - compatible for A @ B

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_negative Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_distributivity2.py
code served (permissive licence) · get_code("2d2a55983a86dae7")
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 2, 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

torch.manual_seed(0)

# Analyze the function:
# - A: matrix, used in subtraction with tmp @ x
# - H: matrix, used in torch.t(H) @ H
# - x: vector, used in matrix multiplication
# The function computes: (A - H^T @ H) @ x

# Choose dimensions:
# Let H be 4x3 (so H^T @ H is 3x3)
# Let A be 3x3 (to match H^T @ H for subtraction)
# Let x be 3x2 (vector/matrix to multiply with result)

A = torch.randn(3, 3)
H = torch.randn(4, 3)
x = torch.randn(3, 2)

ARGS = {
    "A": A,
    "H": H,
    "x": x,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_negative Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_transpose.py
code served (permissive licence) · get_code("5d516c141faae762")
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 2, 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

torch.manual_seed(0)

# ref_negative takes two tensor arguments A and B
# It computes: torch.t(B) @ A @ torch.t(A) @ B
# For this to be valid:
# - A must be a 2D tensor (square matrix for the multiplication chain to work)
# - B must be a 2D tensor where dimensions are compatible
# - torch.t(A) transposes A
# - torch.t(B) transposes B
# - The multiplications must be dimensionally compatible

# Let's construct:
# A: shape (4, 4) - a square matrix
# B: shape (4, 3) - compatible for the chain of operations

A = torch.randn(4, 4)
B = torch.randn(4, 3)

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_negative Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cm_partial_op_prod.py
code served (permissive licence) · get_code("953e2ef4ffaf77b2")
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 2, 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

torch.manual_seed(0)

# ref_negative performs matrix multiplication: A @ B
# A and B must be 2D tensors (or compatible for @)
# Using distinct sizes: A is 2x4, B is 4x3

A = torch.randn(2, 4)
B = torch.randn(4, 3)

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_negative Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cm_partial_op_sum.py
code served (permissive licence) · get_code("11d2511028f02081")
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 2, 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

torch.manual_seed(0)

A = torch.randn(2, 4)
B = torch.randn(2, 4)

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_blocked.py
code served (permissive licence) · get_code("36b39fcbf80f56e4")
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 2, 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

torch.manual_seed(0)

# Analyze the function:
# - A1 @ B1: matrix multiplication, so A1 and B1 must be compatible
# - A2 @ B2: matrix multiplication, so A2 and B2 must be compatible
# - torch.cat((A1 @ B1, A2 @ B2), dim=0): concatenate along dim 0
# This means (A1 @ B1) and (A2 @ B2) must have compatible shapes for concatenation

# Let's use:
# A1: (2, 4) @ B1: (4, 3) = (2, 3)
# A2: (2, 4) @ B2: (4, 5) = (2, 5)
# After cat on dim=0, we'd get (4, ?) - but shapes don't match on dim 1

# Better approach:
# A1: (2, 4) @ B1: (4, 3) = (2, 3)
# A2: (3, 4) @ B2: (4, 3) = (3, 3)
# After cat on dim=0: (5, 3)

A1 = torch.randn(2, 4)
B1 = torch.randn(4, 3)
A2 = torch.randn(3, 4)
B2 = torch.randn(4, 3)

ARGS = {
    'A1': A1,
    'A2': A2,
    'B1': B1,
    'B2': B2,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_distributivity1.py
code served (permissive licence) · get_code("1529084dacf58e68")
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 2, 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

torch.manual_seed(0)

# The function ref_positive takes three parameters: A, B, C
# It performs matrix multiplication: A @ (B + C)
# So A, B, C should be tensors where this operation is valid

# For matrix multiplication to work:
# - A should be shape (m, n)
# - (B + C) should be shape (n, p)
# - B and C must have the same shape for addition

# Using distinct sizes: A is 2x4, B and C are 4x3
A = torch.randn(2, 4)
B = torch.randn(4, 3)
C = torch.randn(4, 3)

ARGS = {
    "A": A,
    "B": B,
    "C": C
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_transpose.py
code served (permissive licence) · get_code("54d5a84e0d761703")
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 2, 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

torch.manual_seed(0)

# Construct small tensors for the function
# ref_positive takes two tensors A and B and computes:
# S = A^T @ B
# ret = S^T @ S

# A: shape (batch=2, dim1=4)
# B: shape (batch=2, dim2=3)
# S will be (dim1=4, dim2=3)
# ret will be (dim2=3, dim2=3)

A = torch.randn(2, 4)
B = torch.randn(2, 3)

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cm_partial_op_prod.py
code served (permissive licence) · get_code("d792bafba67d2899")
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; 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

torch.manual_seed(0)

# Construct tensors for ref_positive
# A[2] accesses index 2, so A must be at least 3D or indexable to get a 1D tensor
# B[:, 2] accesses column 2, so B must be 2D with at least 3 columns
# torch.dot(A[2], B[:, 2]) computes dot product of two 1D tensors

A = torch.randn(5, 4)  # 2D tensor, A[2] will be shape (4,)
B = torch.randn(2, 3)  # 2D tensor, B[:, 2] will be shape (2,)

# For torch.dot to work, both arguments must have the same length
# A[2] has length 4, B[:, 2] has length 2 - mismatch
# Let me fix this: make B have first dimension matching A's second dimension

A = torch.randn(5, 4)  # A[2] is shape (4,)
B = torch.randn(4, 3)  # B[:, 2] is shape (4,)

ARGS = {
    "A": A,
    "B": B,
}
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 — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cm_partial_op_sum.py
code served (permissive licence) · get_code("191ebcca1eb39dd6")
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; 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

torch.manual_seed(0)

# The function indexes A[2, 2] and B[2, 2], so both need to be at least 3x3 tensors
A = torch.randn(4, 5)
B = torch.randn(3, 3)

ARGS = {
    "A": A,
    "B": B,
}
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 — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
ref_positive Ran
ran on a synthesized input
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cse_addition.py
code served (permissive licence) · get_code("ef0d2ee2886a35a8")
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 2, 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

torch.manual_seed(0)

# The function takes two tensors A and B
# It computes: transpose(A) @ B, then multiplies by 2
# A and B need compatible dimensions for matrix multiplication
# If A is (m, n), then transpose(A) is (n, m)
# For transpose(A) @ B to work, B must be (m, k)
# Result will be (n, k)

A = torch.randn(2, 4)  # shape (2, 4)
B = torch.randn(2, 3)  # shape (2, 3)

ARGS = {
    "A": A,
    "B": B,
}
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 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
benchmark Not yet run
unverified (harvested, no recorded run)
chrispsa/lamp_benchmark/python/numpy/benchmarker.py
code served (permissive licence) · get_code("5f6adb6e03177270")
How it ran

no invocation recorded.

ref_negative Not yet run
unverified (harvested, no recorded run)
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_distributivity1.py
code served (permissive licence) · get_code("16e19eafa21d1903")
How it ran

no invocation recorded.

ref_positive Not yet run
unverified (harvested, no recorded run)
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/am_distributivity2.py
code served (permissive licence) · get_code("c73ed459f46b94af")
How it ran

no invocation recorded.

ref_positive Not yet run
unverified (harvested, no recorded run)
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cm_loops.py
code served (permissive licence) · get_code("40c15acbfeef17da")
How it ran

no invocation recorded.

ref_positive Not yet run
unverified (harvested, no recorded run)
as641651/linearalgebra-awareness-benchmark/laab_python/src/PyTorch/v2-cpu/cse_matmul_no_paranthesis.py
code served (permissive licence) · get_code("bf7659c80bf0ef0b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 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 19 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

Linear algebra operations, which are ubiquitous in machine learning, form major performance bottlenecks. The High-Performance Computing community invests significant effort in the development of architecture-specific optimized kernels, such as those provided by the BLAS and LAPACK libraries, to speed up linear algebra operations. However, end users are progressively less likely to go through the error prone and time-consuming process of directly using said kernels; instead, frameworks such as TensorFlow (TF) and PyTorch (PyT), which facilitate the development of machine learning applications, are becoming more and more popular. Although such frameworks link to BLAS and LAPACK, it is not clear whether or not they make use of linear algebra knowledge to speed up computations. For this reason, in this paper we develop benchmarks to investigate the linear algebra optimization capabilities of TF and PyT. Our analyses reveal that a number of linear algebra optimizations are still missing; for instance, reducing the number of scalar operations by applying the distributive law, and automatically identifying the optimal parenthesization of a matrix chain. In this work, we focus on linear algebra computations in TF and PyT; we both expose opportunities for performance enhancement to the benefit of the developers of the frameworks and provide end users with guidelines on how to achieve performance gains.

For agents

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

get_harvested_code_for_paper("2202.09888")
get_code_for_paper("2202.09888")
have("2202.09888")

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

curl https://syntology.ai/api/ran/2202.09888.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.09888.svg)](https://syntology.ai/paper/2202.09888)

Connect an agent — have() is free.