SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.15832 · NeurIPS · 2023

All Points Matter: Entropy-Regularized Distribution Alignment for Weakly-supervised 3D Segmentation

Dacheng Tao, Chaoyue Wang, Zhe Chen, Shanshan Zhao, Liyao Tang

arXiv · PDF · Open in the Atlas

Code that ran

16 of 33 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; 2 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 13 failures of our instrument; 17 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2305.15832.json · badge: /api/ran/2305.15832.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
LiyaoTang/ERDA canonical (classifier label)
classifier basis: classifier_2025
16 ran · 17 no recorded run
FunctionStatusWhere it lives
apply_reduction Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("65566e34fbee2d02")
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
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)

# Create a small target tensor with shape (batch=2, seq_len=3, features=4)
target = tf.constant(np.random.randn(2, 3, 4).astype(np.float32))

# Create a mask with shape (batch=2, seq_len=3) - one dimension less than target
mask = tf.constant(np.array([[1.0, 1.0, 0.0], [1.0, 0.0, 0.0]]).astype(np.float32))

# Test with different reduction types
reduction = 'mean'

# axis parameter specifies which dimension to reduce
axis = -2

ARGS = {
    'target': target,
    'reduction': reduction,
    'mask': mask,
    'axis': axis
}
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 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
check_mask Ran
ran, violated the contract we drafted
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("585514b22f60ec47")
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 EagerTensor, ndim 3, dtype family float, shape (2, 4, 1) 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
import numpy as np

# Create test tensors
x = tf.constant(np.random.randn(2, 4, 3), dtype=tf.float32)
mask = tf.constant(np.random.randn(2, 4), dtype=tf.float32)

ARGS = {
    "x": x,
    "mask": mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 3, dtype family float, shape (2, 4, 1) — shape (2, 4, 1) 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
dropout Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("66f4ca8e47d66009")
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 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
import numpy as np

# Set random seeds for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create small test tensors
features = tf.constant(np.random.randn(2, 4).astype(np.float32))
rate = 0.5
is_training = tf.constant(True)
noise_shape = None

ARGS = {
    "features": features,
    "rate": rate,
    "is_training": is_training,
    "noise_shape": noise_shape
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
fetch_supports_stage Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("9166f0042b70f6a0")
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 torch
import numpy as np

torch.manual_seed(0)

# Build the nested structure expected by fetch_supports_stage
# The function accesses: inputs['stage_list'][stage_n][stage_i]
# and inputs['neighbors'][stage_i]

# Each stage needs: ptype (e.g., 'p_out', 'p_sample') and ftype (e.g., 'f_out', 'latent', etc.)
stage_dict = {
    'p_out': torch.randn(2, 4),  # small tensor for points
    'f_out': torch.randn(2, 3),  # small tensor for features
    'p_sample': torch.randn(2, 4),
    'f_sample': torch.randn(2, 3),
}

# Create stage_list with all required stage types
inputs = {
    'stage_list': {
        'down': [stage_dict, stage_dict],  # list of stages
        'up': [stage_dict, stage_dict],
        'stages': [stage_dict, stage_dict],
    },
    'neighbors': [np.array([0, 1]), np.array([1, 0])],  # neighbor indices for each stage
}

# Test parameters
stage_n = 'down'
stage_i = 0
ftype = 'out'

ARGS = {
    'inputs': inputs,
    'stage_n': stage_n,
    'stage_i': stage_i,
    'ftype': ftype,
}
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 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_activation Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("6f022cb10fccee9d")
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 function. 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 tensorflow as tf

ARGS = {
    'activation': 'relu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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_class_weight Ran
ran, honoured the contract we drafted
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("36d027452f58eb43")
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 1, dtype family float, shape (13,) 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
import numpy as np

ARGS = {
    'dataset': 'S3DIS',
    'labels': None,
    'weight_type': 'class',
    'normalize': '',
    'version': ''
}
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 1, dtype family float, shape (13,) — shape (13,) 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_ftype Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("11f40ad83f37e95f")
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, 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 re

def get_ftype(ftype, raise_not_found=True):
    if ftype in ['out', 'fout', 'f_out']:
        ptype = 'p_out'
        ftype = 'f_out'
    elif any((re.fullmatch(f'{k}(\\d*mlp|mlp\\d*|linear|)', ftype) for k in ['latent', 'logits', 'probs'])):
        ptype = 'p_out'
        ftype = [k for k in ['latent', 'logits', 'probs'] if ftype.startswith(k)][0]
    elif ftype in ['sample', 'fsample', 'f_sample']:
        ptype = 'p_sample'
        ftype = 'f_sample' if ftype in ['sample', 'fsample'] else ftype
    elif raise_not_found:
        raise KeyError(f'not supported ftype = {ftype}')
    else:
        ftype = ptype = None
    return (ftype, ptype)

ARGS = {
    'ftype': 'out',
    'raise_not_found': True
}
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 tuple, ndim 1, dtype family float — a tuple 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_initializer Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("b9980bfcdc9fa508")
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 GlorotUniform. 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 tensorflow as tf

# Test various initializer string inputs
ARGS = {
    'k': 'xavier'  # Test with a valid string initializer
}

# Additional test cases to verify the function works with different inputs:
test_cases = [
    {'k': None},
    {'k': 'glorot'},
    {'k': 'glorotN'},
    {'k': 'normalf'},
    {'k': 'normalf0.05'},
    {'k': 'normal'},
    {'k': 'normal0.02'},
    {'k': 'fanin'},
    {'k': 'fanout'},
    {'k': 'msra'},
    {'k': 'he'},
    {'k': 'heout'},
    {'k': 'zeros'},
    {'k': 'ones'},
    {'k': tf.initializers.glorot_uniform()},  # Non-string initializer object
]

# Use the first test case as the main ARGS
ARGS = test_cases[0]
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GlorotUniform — a GlorotUniform output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GlorotUniform 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_kwargs Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("6eb3a7a026e881ba")
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
from dataclasses import dataclass

torch.manual_seed(0)

@dataclass
class BlockConfig:
    init: str = 'xavier'
    wd: str = '0.0001'
    bn: str = 'batch_norm'
    act: str = 'relu'

@dataclass
class Config:
    init: str = 'glorot'
    weight_decay: float = 0.0005
    bn_momentum: float = 0.9
    bn_eps: float = 1e-5
    bn: str = 'instance_norm'
    activation: str = 'tanh'

block_cfg = BlockConfig(
    init='xavier',
    wd='0.0001',
    bn='batch_norm',
    act='relu'
)

config = Config(
    init='glorot',
    weight_decay=0.0005,
    bn_momentum=0.9,
    bn_eps=1e-5,
    bn='instance_norm',
    activation='tanh'
)

ARGS = {
    'block_cfg': block_cfg,
    'config': config,
    'is_training': True,
    'act': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
get_kwargs_mlp Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("a36a27d7a5eeaf3e")
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)

class BlockConfig:
    def __init__(self, init='', wd='', bn='', act='', linearbn=''):
        self.init = init
        self.wd = wd
        self.bn = bn
        self.act = act
        self.linearbn = linearbn

class Config:
    def __init__(self, init='xavier', weight_decay=0.0001, bn_momentum=0.9, bn_eps=1e-5, bn='batch_norm', activation='relu', linearbn=''):
        self.init = init
        self.weight_decay = weight_decay
        self.bn_momentum = bn_momentum
        self.bn_eps = bn_eps
        self.bn = bn
        self.activation = activation
        self.linearbn = linearbn

block_cfg = BlockConfig(init='kaiming', wd='0.0002', bn='layer_norm', act='gelu', linearbn='true')
config = Config(init='xavier', weight_decay=0.0001, bn_momentum=0.95, bn_eps=1e-5, bn='batch_norm', activation='relu', linearbn='false')
is_training = True
act = True
_kwargs = {'hidden_size': 256}

ARGS = {
    'block_cfg': block_cfg,
    'config': config,
    'is_training': is_training,
    'act': act,
    '_kwargs': _kwargs
}
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
get_type Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("65a09d51e78e2ff8")
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 DType. 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 tensorflow as tf

def get_type(k):
    if k in [True, None, float, 'float', 'float32']:
        return tf.float32
    elif k in ['float16']:
        return tf.float16
    elif k in [int, 'int', 'int32']:
        return tf.int32
    elif k in [bool, 'bool']:
        return tf.bool
    elif hasattr(tf, k) and isinstance(k, str):
        return getattr(tf, k)
    else:
        return k

# Test with various input types
ARGS = {
    'k': 'float32'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DType — a DType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DType 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
masked_softmax Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("72be9593f760bb29")
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 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
import numpy as np

# Create small test tensors
batch_size = 2
seq_len = 4

# Create a vector to apply softmax to
vec = tf.constant([
    [1.0, 2.0, 3.0, 4.0],
    [0.5, 1.5, 2.5, 3.5]
], dtype=tf.float32)

# Create a valid mask (boolean mask indicating which positions are valid)
# True means valid, False means invalid (will be masked out)
valid_mask = tf.constant([
    [True, True, False, True],
    [True, False, True, True]
], dtype=tf.bool)

ARGS = {
    "vec": vec,
    "valid_mask": valid_mask,
    "axis": -1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
tf_cond Ran
ran, honoured the contract we drafted
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("0822e0ff9c4729a5")
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 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

# Define simple callable functions for true_fn and false_fn
def true_fn():
    return tf.constant(42.0)

def false_fn():
    return tf.constant(0.0)

# Create a scalar predicate (boolean tensor)
pred = tf.constant(True)

# Construct the arguments dictionary
ARGS = {
    "pred": pred,
    "true_fn": true_fn,
    "false_fn": false_fn,
    "name": "test_cond"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
tf_gather Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("862351355d71ffc6")
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 tensorflow as tf
import numpy as np

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

neighbor_idx = tf.constant([[0, 1], [1, 0]], dtype=tf.int32)

def shadow_fn_callable(shape, dtype):
    return tf.zeros(shape, dtype=dtype)

ARGS = {
    'supports': supports,
    'neighbor_idx': neighbor_idx,
    'shadow_fn': shadow_fn_callable,
    'get_mask': True
}
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 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
tf_scope Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("7153a8b6464f04ea")
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 function. 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 tensorflow as tf

def simple_function(x, y):
    """A simple function to be decorated"""
    return x + y

@tf.function
def decorated_func(x, y):
    """Function that will be wrapped by tf_scope decorator"""
    return x * y

def tf_scope(func):
    """ decorator: automatically wrap a var scope """

    def scopped_func(*args, name=None, reuse=None, **kwargs):
        if name is not None and (not reuse):
            with tf.variable_scope(name):
                return func(*args, **kwargs)
        elif name is not None and reuse:
            with tf.variable_scope(reuse, auxiliary_name_scope=False, reuse=True):
                with tf.name_scope(name):
                    return func(*args, **kwargs)
        elif reuse:
            with tf.variable_scope(reuse, reuse=True):
                return func(*args, **kwargs)
        else:
            return func(*args, **kwargs)
    return scopped_func

# Create a simple test function
def test_func(a, b):
    return a + b

# Apply the decorator
wrapped_func = tf_scope(test_func)

# Create test arguments
test_a = 5
test_b = 3

ARGS = {
    "func": test_func,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
to_valid_stage Ran
ran; our contract draft was wrong, not the code
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("532f90a2c87f9f32")
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 torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - stage_n: can be 'D', 'down', 'U', 'up', 'S', or 'stages'
# - short: boolean flag that determines output format
# The function validates stage_n and returns either short or long form

ARGS = {
    'stage_n': 'D',
    'short': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
_variable_on_cpu Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("ce6452caa8be618b")
How it ran

no invocation recorded.

_variable_with_weight_decay Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("d98ee661d06ca55e")
How it ran

no invocation recorded.

batch_norm Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("0c8f88cadeb2d45f")
How it ran

no invocation recorded.

calc_dist Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("34736bc6edd358fa")
How it ran

no invocation recorded.

calc_loss Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("87082c78c01cdcc8")
How it ran

no invocation recorded.

cls_contrast_head Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("2b0265835b2b9e32")
How it ran

no invocation recorded.

conv1d_1x1 Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("2101a28ba054c714")
How it ran

no invocation recorded.

dense_layer Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("6a695291eaa9bee6")
How it ran

no invocation recorded.

get_sample_idx Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("a3a2fe6e67287299")
How it ran

no invocation recorded.

get_scene_label Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("77f37437581eb0b7")
How it ran

no invocation recorded.

mlp_head Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("747ee8f1f284c303")
How it ran

no invocation recorded.

mlps Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("95ee8413bb72a5d2")
How it ran

no invocation recorded.

mlps_by_ops Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("b5717e76cb07f225")
How it ran

no invocation recorded.

normalize Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("114618f6f9dc6b02")
How it ran

no invocation recorded.

parse_stage Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("c1132af83963c82c")
How it ran

no invocation recorded.

pseudo_head Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("b62f118649752c42")
How it ran

no invocation recorded.

tf_get_variable Not yet run
unverified (harvested, no recorded run)
LiyaoTang/ERDA/models/head.py
code served (permissive licence) · get_code("5271d891a74ff73e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 33 samples lifted from this paper's linked repositories, 33 carry a function identity and 0 do not. Of the 33 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 32 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
_variable_on_cpu
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1612.00593 (2016-12)
PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation

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

Pseudo-labels are widely employed in weakly supervised 3D segmentation tasks where only sparse ground-truth labels are available for learning. Existing methods often rely on empirical label selection strategies, such as confidence thresholding, to generate beneficial pseudo-labels for model training. This approach may, however, hinder the comprehensive exploitation of unlabeled data points. We hypothesize that this selective usage arises from the noise in pseudo-labels generated on unlabeled data. The noise in pseudo-labels may result in significant discrepancies between pseudo-labels and model predictions, thus confusing and affecting the model training greatly. To address this issue, we propose a novel learning strategy to regularize the generated pseudo-labels and effectively narrow the gaps between pseudo-labels and model predictions. More specifically, our method introduces an Entropy Regularization loss and a Distribution Alignment loss for weakly supervised learning in 3D segmentation tasks, resulting in an ERDA learning strategy. Interestingly, by using KL distance to formulate the distribution alignment loss, it reduces to a deceptively simple cross-entropy-based loss which optimizes both the pseudo-label generation network and the 3D segmentation network simultaneously. Despite the simplicity, our method promisingly improves the performance. We validate the effectiveness through extensive experiments on various baselines and large-scale datasets. Results show that ERDA effectively enables the effective usage of all unlabeled data points for learning and achieves state-of-the-art performance under different settings. Remarkably, our method can outperform fully-supervised baselines using only 1% of true annotations. Code and model will be made publicly available at https://github.com/LiyaoTang/ERDA.

For agents

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

get_harvested_code_for_paper("2305.15832")
get_code_for_paper("2305.15832")
have("2305.15832")

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

curl https://syntology.ai/api/ran/2305.15832.json

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

[![Syntology run record](https://syntology.ai/api/ran/2305.15832.svg)](https://syntology.ai/paper/2305.15832)

Connect an agent — have() is free.