SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.00091 · 2019

Deep Learning Recommendation Model for Personalization and Recommendation Systems

arXiv · PDF · Open in the Atlas

Code that ran

19 of 28 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; 1 honoured the contract we drafted, 0 violated it and 15 have no contract verdict; 3 failures of our instrument; 9 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/1906.00091.json · badge: /api/ran/1906.00091.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
alibaba/easyrec community (classifier label)
classifier basis not recorded on the link
10 ran · 6 no recorded run
ravikucb/dnas community (classifier label)
classifier basis not recorded on the link
5 ran · 1 no recorded run
myungkeun-cho/taobao_dlrm reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 2 no recorded run
nikithamalgifb/dlrm reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
calculate_metrics Ran
ran; our contract draft was wrong, not the code
myungkeun-cho/taobao_dlrm/dlrm_s_caffe2.py
code served (permissive licence) · get_code("332764c725473589")
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) 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 numpy as np
import sklearn.metrics

# Set random seed for reproducibility
np.random.seed(0)

# Create sample data for binary classification
# targets: list of arrays containing true binary labels
# scores: list of arrays containing predicted probabilities

# Create 2 batches of targets (binary labels: 0 or 1)
batch1_targets = np.array([0, 1, 1, 0])
batch2_targets = np.array([1, 0, 1, 1])
targets = [batch1_targets, batch2_targets]

# Create 2 batches of scores (probabilities between 0 and 1)
batch1_scores = np.array([0.1, 0.9, 0.8, 0.2])
batch2_scores = np.array([0.7, 0.3, 0.95, 0.85])
scores = [batch1_scores, batch2_scores]

ARGS = {
    'targets': targets,
    'scores': scores
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn — 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
collate_wrapper_criteo Ran
ran on a synthesized input
ravikucb/dnas/dlrm_data_pytorch.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/dlrm, is pointer only (licence: MIT) · get_code("283f3ba4a908ce84")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

collate_wrapper_random Ran
ran on a synthesized input
ravikucb/dnas/dlrm_data_pytorch.py
pointer only (licence: MIT); the stored copy get_code serves, from JShubhM15/recommender_systems, is pointer only (licence: MIT) · get_code("f2e2bab9a2fd0232")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

collect_named_outputs Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/compat/layers.py
code served (permissive licence) · get_code("89d240b361286d81")
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 EagerTensor, ndim 2, dtype family float, shape (2, 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 tensorflow as tf
from tensorflow.python.framework import ops
import torch

# Create a mock tensor-like object that TensorFlow can work with
# We'll use a real TensorFlow tensor
outputs_tensor = tf.constant([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)

# Set up collections - can be a string or list of strings
collections = ['end_points', 'summaries']

# Alias string to append
alias = 'inception_v3/logits'

ARGS = {
    'collections': collections,
    'alias': alias,
    'outputs': outputs_tensor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
combine_lines Ran
ran on a synthesized input
ravikucb/dnas/dnas_data_utils.py
pointer only (licence: MIT) · get_code("5c26eb8f78eded5a")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

convertUStringToDistinctIntsDict Ran
ran on a synthesized input
ravikucb/dnas/data_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/dlrm, is pointer only (licence: MIT) · get_code("0035cd64b0d2bc24")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

convertUStringToDistinctIntsUnique Ran
ran on a synthesized input
ravikucb/dnas/data_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/dlrm, is pointer only (licence: MIT) · get_code("892c2beb0c55d9bd")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

cosine_decay_with_warmup Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/core/learning_schedules.py
code served (permissive licence) · get_code("9c3002d3c253cd40")
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 EagerTensor, 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 numpy as np
import tensorflow as tf

# Define ARGS dict with all parameters for cosine_decay_with_warmup
ARGS = {
    "global_step": tf.constant(5, dtype=tf.int64),
    "learning_rate_base": 0.1,
    "total_steps": 100,
    "warmup_learning_rate": 0.0,
    "warmup_steps": 10,
    "hold_base_rate_steps": 0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
dash_separated_floats Ran
ran; our contract draft was wrong, not the code
nikithamalgifb/dlrm/dlrm_s_pytorch.py
code served (permissive licence) · get_code("ae34c5cfff000298")
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) and returned str. 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 argparse

def dash_separated_floats(value):
    vals = value.split('-')
    for val in vals:
        try:
            float(val)
        except ValueError:
            raise argparse.ArgumentTypeError('%s is not a valid dash separated list of floats' % value)
    return value

ARGS = {
    'value': '1.5-2.3-3.14-0.001'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
dash_separated_ints Ran
ran; our contract draft was wrong, not the code
nikithamalgifb/dlrm/dlrm_s_pytorch.py
code served (permissive licence) · get_code("42411aae7c7fe426")
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) and returned str. 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 argparse

def dash_separated_ints(value):
    vals = value.split('-')
    for val in vals:
        try:
            int(val)
        except ValueError:
            raise argparse.ArgumentTypeError('%s is not a valid dash separated list of ints' % value)
    return value

ARGS = {
    'value': '1-2-3-4-5'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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_value Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/layers/backbone.py
code served (permissive licence) · get_code("4b0efd503529d5da")
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 str. 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 six
from google.protobuf import struct_pb2

# Create test values for format_value function
text_value = six.text_type("hello")
float_value = 42.5
int_float_value = 10.0

# Create a ListValue with some test values
list_value = struct_pb2.ListValue()
list_value.values.add().string_value = "item1"
list_value.values.add().number_value = 3.14

# Create a Struct with some test values
struct_value = struct_pb2.Struct()
struct_value.fields["key1"].string_value = "value1"
struct_value.fields["key2"].number_value = 2.5

# Test with a simple integer
int_value = 42

# Test with None or other type
none_value = None

ARGS = {
    "value": text_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
google, six — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
get_file_arr Ran
ran on a synthesized input
alibaba/easyrec/git-lfs/git_lfs.py
code served (permissive licence) · get_code("5a7adc44c734dc67")
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 os
import tempfile
import shutil

# Create a temporary directory with some test files
temp_dir = tempfile.mkdtemp()
test_file_1 = os.path.join(temp_dir, "file1.txt")
test_file_2 = os.path.join(temp_dir, "file2.txt")
test_subdir = os.path.join(temp_dir, "subdir")

# Create test files
with open(test_file_1, 'w') as f:
    f.write("test content 1")
with open(test_file_2, 'w') as f:
    f.write("test content 2")
os.makedirs(test_subdir, exist_ok=True)

# Create a single file path for testing the else branch
single_file = os.path.join(temp_dir, "single_file.txt")
with open(single_file, 'w') as f:
    f.write("single file content")

# Define ARGS with both test cases available
ARGS = {
    "path": temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
get_local_sig Ran
ran on a synthesized input
alibaba/easyrec/git-lfs/git_lfs.py
code served (permissive licence) · get_code("bc60e924c5034e31")
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 str. 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 hashlib
import logging
import tempfile
import os

# Create temporary test files
temp_dir = tempfile.mkdtemp()

# Create a few small test files with known content
test_file_1 = os.path.join(temp_dir, 'test1.txt')
with open(test_file_1, 'wb') as f:
    f.write(b'test content 1')

test_file_2 = os.path.join(temp_dir, 'test2.txt')
with open(test_file_2, 'wb') as f:
    f.write(b'test content 2')

test_file_3 = os.path.join(temp_dir, 'test3.txt')
with open(test_file_3, 'wb') as f:
    f.write(b'test content 3')

# Create leaf_files parameter - a list of file paths
leaf_files = [test_file_1, test_file_2, test_file_3]

ARGS = {
    'leaf_files': leaf_files
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
get_variable_collections Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/compat/layers.py
code served (permissive licence) · get_code("482e02a23a3c5a45")
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

torch.manual_seed(0)

ARGS = {
    "variables_collections": {
        "model_vars": ["weight1", "bias1"],
        "optimizer_vars": ["momentum", "velocity"],
    },
    "name": "model_vars",
}
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 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
get_wide_embed_dim Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/layers/backbone.py
code served (permissive licence) · get_code("a7ae5d5be4e838b5")
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 int, ndim 0, dtype family int, 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
import torch.nn as nn
from google.protobuf import message

# Create a minimal protobuf-like block class that mimics the expected structure
class InputLayer(message.Message):
    def __init__(self, wide_output_dim=None):
        self._wide_output_dim = wide_output_dim
    
    def HasField(self, field_name):
        if field_name == 'wide_output_dim':
            return self._wide_output_dim is not None
        return False
    
    @property
    def wide_output_dim(self):
        return self._wide_output_dim

class Block(message.Message):
    def __init__(self, layer_type=None, input_layer=None):
        self._layer_type = layer_type
        self._input_layer = input_layer
    
    def WhichOneof(self, oneof_name):
        if oneof_name == 'layer':
            return self._layer_type
        return None
    
    @property
    def input_layer(self):
        return self._input_layer

# Create test blocks with input_layer containing wide_output_dim
block1 = Block(
    layer_type='input_layer',
    input_layer=InputLayer(wide_output_dim=128)
)

block2 = Block(
    layer_type='input_layer',
    input_layer=InputLayer(wide_output_dim=128)
)

block3 = Block(
    layer_type='other_layer',
    input_layer=None
)

blocks = [block1, block2, block3]

ARGS = {
    'blocks': blocks,
    'wide_embed_dim': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
google, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
manual_stepping Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/core/learning_schedules.py
code served (permissive licence) · get_code("691e547d271cb0d0")
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 EagerTensor, 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 tensorflow as tf

# Example 1: Basic case without warmup
global_step_1 = tf.constant(7, dtype=tf.int64)
boundaries_1 = [5, 10]
rates_1 = [0.1, 0.01, 0.001]
warmup_1 = False

# Example 2: With warmup enabled
global_step_2 = tf.constant(3, dtype=tf.int64)
boundaries_2 = [5, 10, 15]
rates_2 = [0.1, 0.05, 0.01, 0.001]
warmup_2 = True

# Use the first example for ARGS
ARGS = {
    'global_step': global_step_1,
    'boundaries': boundaries_1,
    'rates': rates_1,
    'warmup': warmup_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
merge_inputs Ran
ran on a synthesized input
alibaba/easyrec/easy_rec/python/layers/backbone.py
code served (permissive licence) · get_code("2e4eb0a163ff07c2")
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 EagerTensor, 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 tensorflow as tf
import logging

logging.basicConfig(level=logging.INFO)

# Create test tensors
tensor1 = tf.constant([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)
tensor2 = tf.constant([[9.0, 10.0, 11.0, 12.0], [13.0, 14.0, 15.0, 16.0]], dtype=tf.float32)

# Create list inputs
list1 = [1, 2, 3]
list2 = [4, 5, 6]

ARGS = {
    'inputs': [tensor1, tensor2],
    'axis': -1,
    'msg': 'test merge'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 2, dtype family float — no output shape is stated: observe_signature.py recorded SymbolicTensor ndim 2, which disagrees with the fixture run
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
path2name Ran
ran on a synthesized input
alibaba/easyrec/git-lfs/git_lfs.py
code served (permissive licence) · get_code("5f77fa0d16a4ac1a")
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 str. 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

def path2name(path):
    name = path.replace('//', '/')
    name = path.replace('/', '_')
    if name[-1] == '_':
        return name[:-1]
    elif name == '.':
        return 'curr_dir'
    else:
        return name

ARGS = {
    'path': '/home/user/documents/file.txt'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
time_wrap Ran
ran, honoured the contract we drafted
myungkeun-cho/taobao_dlrm/dlrm_s_pytorch.py
code served (permissive licence) · get_code("0b825b705dac66c0")
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) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. 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, 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 time
import torch

torch.manual_seed(0)

ARGS = {
    "use_gpu": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
concatCriteoAdData Not yet run
unverified (harvested, no recorded run)
ravikucb/dnas/data_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/dlrm, is pointer only (licence: MIT) · get_code("c1637c05b55a3b1f")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, ravikucb/dnas, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

dlrm_wrap Not yet run
unverified (harvested, no recorded run)
myungkeun-cho/taobao_dlrm/dlrm_s_pytorch.py
code served (permissive licence) · get_code("afee36a6506b0fa7")
How it ran

no invocation recorded.

exponential_decay_with_burnin Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/core/learning_schedules.py
code served (permissive licence) · get_code("8b33bbde992ac658")
How it ran

no invocation recorded.

fast_auc Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/core/metrics.py
code served (permissive licence) · get_code("d25a40937fdc87c5")
How it ran

no invocation recorded.

layer_norm Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/compat/layers.py
code served (permissive licence) · get_code("2afd8b1d98ef1308")
How it ran

no invocation recorded.

loss_fn_wrap Not yet run
unverified (harvested, no recorded run)
myungkeun-cho/taobao_dlrm/dlrm_s_pytorch.py
code served (permissive licence) · get_code("0e5d7c0e5fed2a07")
How it ran

no invocation recorded.

max_f1 Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/core/metrics.py
code served (permissive licence) · get_code("4160b3028830c70f")
How it ran

no invocation recorded.

metric_variable Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/core/easyrec_metrics/distribute_metrics_impl_pai.py
code served (permissive licence) · get_code("cc55ccf790502ef7")
How it ran

no invocation recorded.

metric_variable Not yet run
unverified (harvested, no recorded run)
alibaba/easyrec/easy_rec/python/core/easyrec_metrics/distribute_metrics_impl_tf.py
code served (permissive licence) · get_code("21d78d7b1274b6ff")
How it ran

no invocation recorded.

Also present in other papers' code

Of 28 samples lifted from this paper's linked repositories, 28 carry a function identity and 0 do not. Of the 28 distinct functions with an identity, 11 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 6 the same month, 9 later) and 17 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
calculate_metrics 3 other papers
1 the same month, 2 later
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation
1909.02107 (2019-09)
Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
dlrm_wrap 3 other papers
1 the same month, 2 later
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation
1909.02107 (2019-09)
Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
loss_fn_wrap 3 other papers
1 the same month, 2 later
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation
1909.02107 (2019-09)
Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
time_wrap 3 other papers
1 the same month, 2 later
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation
1909.02107 (2019-09)
Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
collate_wrapper_criteo 1 other paper
1 later
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
collate_wrapper_random 1 other paper
1 later
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
concatCriteoAdData 1 other paper
1 later
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
convertUStringToDistinctIntsDict 1 other paper
1 later
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
convertUStringToDistinctIntsUnique 1 other paper
1 later
1909.11810 (2019-09)
Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems
dash_separated_floats 1 other paper
1 the same month
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation
dash_separated_ints 1 other paper
1 the same month
1906.03109 (2019-06)
The Architectural Implications of Facebook's DNN-based Personalized Recommendation

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

With the advent of deep learning, neural network-based recommendation models have emerged as an important tool for tackling personalization and recommendation tasks. These networks differ significantly from other deep learning networks due to their need to handle categorical features and are not well studied or understood. In this paper, we develop a state-of-the-art deep learning recommendation model (DLRM) and provide its implementation in both PyTorch and Caffe2 frameworks. In addition, we design a specialized parallelization scheme utilizing model parallelism on the embedding tables to mitigate memory constraints while exploiting data parallelism to scale-out compute from the fully-connected layers. We compare DLRM against existing recommendation models and characterize its performance on the Big Basin AI platform, demonstrating its usefulness as a benchmark for future algorithmic experimentation and system co-design.

For agents

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

get_harvested_code_for_paper("1906.00091")
get_code_for_paper("1906.00091")
have("1906.00091")

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

curl https://syntology.ai/api/ran/1906.00091.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.00091.svg)](https://syntology.ai/paper/1906.00091)

Connect an agent — have() is free.