SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2201.12414 · NeurIPS · 2022

Posterior Matching for Arbitrary Conditioning

Ryan Strauss, Junier Oliva

arXiv · PDF · Open in the Atlas

Code that ran

14 of 26 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 11 have no contract verdict; 3 failures of our instrument; 12 no recorded run), measured 2026-11-21; 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/2201.12414.json · badge: /api/ran/2201.12414.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
deepmind/dm-haiku canonical (classifier label)
classifier basis not recorded on the link
11 ran · 3 no recorded run
lupalab/posterior-matching canonical (classifier label)
classifier basis: classifier_2025
1 ran · 9 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
FunctionStatusWhere it lives
calculate_bias_shape Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/bias.py
code served (permissive licence) · get_code("dcfe042d53ba91db")
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 int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from collections.abc import Sequence

torch.manual_seed(0)

# Test case 1: Basic usage with bias_dims specified
input_shape_1 = (2, 4, 3, 5)  # batch_size=2, then 3 feature dimensions
bias_dims_1 = (1, 2)

# Test case 2: Empty bias_dims
input_shape_2 = (2, 4, 3)
bias_dims_2 = ()

# Test case 3: bias_dims=None (should apply to all non-batch dims)
input_shape_3 = (2, 4, 3, 5)
bias_dims_3 = None

# Test case 4: Single dimension bias
input_shape_4 = (2, 8, 6)
bias_dims_4 = (2,)

# Use the first test case as the primary example
ARGS = {
    "input_shape": (2, 4, 3, 5),
    "bias_dims": (1, 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 tuple, ndim 1, dtype family int — 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
check_jax_usage Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/config.py
code served (permissive licence) · get_code("60d1392c43f39704")
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 bool, 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 dataclasses
import threading

@dataclasses.dataclass
class Config:
    check_jax_usage: bool
    module_auto_repr: bool
    restore_flatmap: bool
    rng_reserve_size: int

    @classmethod
    def default(cls) -> 'Config':
        return Config(check_jax_usage=False, module_auto_repr=True, restore_flatmap=False, rng_reserve_size=1)

main_thread_config = Config.default()

class ThreadLocalStorage(threading.local):

    def __init__(self):
        super().__init__()
        self.config = Config(**dataclasses.asdict(main_thread_config))

tls = ThreadLocalStorage()

def get_config() -> Config:
    return tls.config

def check_jax_usage(enabled: bool=True) -> bool:
    config = get_config()
    previous_value, config.check_jax_usage = (config.check_jax_usage, enabled)
    return previous_value

ARGS = {
    "enabled": 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 bool, 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
check_not_none Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/base.py
code served (permissive licence) · get_code("3dcc91ac4323976a")
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

def check_not_none(value, msg: str):
    if value is None:
        raise ValueError(msg)
    return value

# Case 1: value is not None - should return the value
ARGS = {
    "value": "test_value",
    "msg": "Value cannot be None"
}
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 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
compute_adjusted_padding Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/conv.py
code served (permissive licence) · get_code("6a0640e81b20a064")
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 int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - input_size: int - the input size for ConvTranspose
# - output_size: int - desired output size
# - kernel_size: int - kernel size for convolution
# - stride: int - stride for convolution
# - padding: str - either 'VALID' or 'SAME'
# - dilation: int - dilation factor (default=1)
#
# The function validates that input_size matches expected_input_size
# for the given parameters. We need to construct valid combinations.

# For 'VALID' padding:
# expected_input_size = (output_size - kernel_size + stride) // stride
# So: input_size = (output_size - kernel_size + stride) // stride

# Let's construct a valid case for 'VALID':
input_size_valid = 5
kernel_size_valid = 3
stride_valid = 2
dilation_valid = 1
# expected_input_size = (output_size - kernel_size + stride) // stride
# We need: input_size = (output_size - kernel_size + stride) // stride
# 5 = (output_size - 3 + 2) // 2
# 10 = output_size - 1
output_size_valid = 11

# Verify: expected = (11 - 3 + 2) // 2 = 10 // 2 = 5 ✓

ARGS = {
    'input_size': 5,
    'output_size': 11,
    'kernel_size': 3,
    'stride': 2,
    'padding': 'VALID',
    'dilation': 1
}
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, ndim 1, dtype family int — 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_width_settings Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from openai/vdvae
pointer only · get_code("9419abf027f0115d")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for 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.

module_auto_repr Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/config.py
code served (permissive licence) · get_code("533c7323fed6b625")
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 bool, 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 dataclasses
import threading

@dataclasses.dataclass
class Config:
    check_jax_usage: bool
    module_auto_repr: bool
    restore_flatmap: bool
    rng_reserve_size: int

    @classmethod
    def default(cls) -> 'Config':
        return Config(check_jax_usage=False, module_auto_repr=True, restore_flatmap=False, rng_reserve_size=1)

main_thread_config = Config.default()

class ThreadLocalStorage(threading.local):

    def __init__(self):
        super().__init__()
        self.config = Config(**dataclasses.asdict(main_thread_config))

tls = ThreadLocalStorage()

def get_config() -> Config:
    return tls.config

def module_auto_repr(enabled: bool) -> bool:
    config = get_config()
    previous_value, config.module_auto_repr = (config.module_auto_repr, enabled)
    return previous_value

ARGS = {
    "enabled": 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 bool, 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
multinomial Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/basic.py
code served (permissive licence) · get_code("c2aac7a3368f16d4")
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 ArrayImpl, ndim 2, dtype family int, shape (2, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import jax
import jax.numpy as jnp

# Set seed for reproducibility
jax.config.update("jax_default_prng_impl", "threefry2x32")

# Create a JAX PRNG key
rng = jax.random.PRNGKey(0)

# Create logits tensor with shape (batch_size=2, num_categories=4)
# This represents unnormalized log-probabilities for 2 samples with 4 categories each
logits = jnp.array([
    [1.0, 2.0, 0.5, 1.5],
    [0.2, 1.8, 0.9, 0.1]
], dtype=jnp.float32)

# Number of samples to draw
num_samples = 3

ARGS = {
    "rng": rng,
    "logits": logits,
    "num_samples": num_samples
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 2, dtype family int, shape (2, 3) — shape (2, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ndim_at_least Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/basic.py
code served (permissive licence) · get_code("71a177ba7a67e972")
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 bool, 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 jax
import jax.numpy as jnp
import numpy as np

# Test case 1: JAX array with sufficient dimensions
jax_array_3d = jnp.ones((2, 3, 4))

# Test case 2: NumPy array with fewer dimensions
numpy_array_1d = np.array([1, 2, 3, 4])

# Test case 3: Python list that will be converted to JAX array
python_list_2d = [[1, 2], [3, 4], [5, 6]]

# Test case 4: NumPy array with exact number of dimensions
numpy_array_2d = np.ones((3, 4))

ARGS = {
    "x": jax_array_3d,
    "num_dims": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, 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
one_hot Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/basic.py
code served (permissive licence) · get_code("f6f51ced321b0151")
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 ArrayImpl, ndim 3, dtype family float, shape (2, 3, 5) 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 jax
import jax.numpy as jnp

# Set seed for reproducibility
jax.config.update('jax_platform_name', 'cpu')

# Create a small tensor of indices for one-hot encoding
# Shape: (2, 3) - batch of 2, with 3 indices each
x = jnp.array([[0, 1, 2],
               [1, 0, 3]], dtype=jnp.int32)

# Number of classes
num_classes = 5

# Data type for output
dtype = jnp.float32

ARGS = {
    'x': x,
    'num_classes': num_classes,
    'dtype': dtype
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 3, dtype family float, shape (2, 3, 5) — shape (2, 3, 5) 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
pad_channels Ran
ran; our contract draft was wrong, not the code
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("10fcb117ffee5145")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

parse_layer_string Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from openai/vdvae
pointer only · get_code("95d3df9fe1921426")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for 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.

rng_reserve_size Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/config.py
code served (permissive licence) · get_code("de4b41a490b3f147")
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 dataclasses
import threading

@dataclasses.dataclass
class Config:
    check_jax_usage: bool
    module_auto_repr: bool
    restore_flatmap: bool
    rng_reserve_size: int

    @classmethod
    def default(cls) -> 'Config':
        return Config(check_jax_usage=False, module_auto_repr=True, restore_flatmap=False, rng_reserve_size=1)

main_thread_config = Config.default()

class ThreadLocalStorage(threading.local):

    def __init__(self):
        super().__init__()
        self.config = Config(**dataclasses.asdict(main_thread_config))

tls = ThreadLocalStorage()

def get_config() -> Config:
    return tls.config

def rng_reserve_size(size: int) -> int:
    """Change amount of RNG keys reserved when calling ``next_rng_key``.

  Args:
    size: amount of keys to reserve when splitting off a key
      through ``next_rng_key()``, defaults to 1. Reserving larger blocks of keys
      can improve compilation and run-time of your model. Changing the
      reservation size will change RNG keys returned by ``next_rng_key``, and
      will change the generated random numbers.
  Returns:
    The previous value of the rng_reserve_size setting.
  """
    if size <= 0:
        raise ValueError(f'RNG reserve size needs to be more than 0, got {size}.')
    config = get_config()
    before, config.rng_reserve_size = (config.rng_reserve_size, size)
    return before

ARGS = {
    "size": 4
}
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 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
to_dimension_numbers Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/conv.py
code served (permissive licence) · get_code("d53951146fe6a788")
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 ConvDimensionNumbers. 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.

from jax import lax

# Test case 1: 2D convolution, channels last, not transpose
ARGS = {
    "num_spatial_dims": 2,
    "channels_last": True,
    "transpose": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvDimensionNumbers — a ConvDimensionNumbers output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ConvDimensionNumbers output the recorded ndim 2 and dtype family int 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_haiku_dict Ran
ran on a synthesized input
deepmind/dm-haiku/haiku/_src/data_structures.py
code served (permissive licence) · get_code("93a365e0dfc01f8c")
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 jax
import jax.numpy as jnp
from collections import OrderedDict

# Create a simple two-level nested mapping structure
# The function expects Mapping[K, V] where the structure should be nestable
inner_dict_1 = {"weight": jnp.array([1.0, 2.0, 3.0, 4.0]), "bias": jnp.array([0.1, 0.2])}
inner_dict_2 = {"scale": jnp.array([5.0, 6.0, 7.0]), "offset": jnp.array([0.5])}

structure = {
    "layer_1": inner_dict_1,
    "layer_2": inner_dict_2,
}

ARGS = {
    "structure": structure,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — 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
Block Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("4086bb07fed39601")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

Encoder Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("701b897fce8a34e9")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

LogisticMixture Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("a42f907e5c7b7a4d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

PosteriorMatchingDecoder Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("b0f8e415415e8379")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

PosteriorMatchingDecoderBlock Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("18273b05e6809fa2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

PosteriorMatchingVDVAE Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("6e0c276984d84108")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

_LogisticMixtureDist Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("00c1496a0adf1aa4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

get_1x1 Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("93ec9c5f4286cbf7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

get_3x3 Not yet run
unverified (harvested, no recorded run)
lupalab/posterior-matching/posterior_matching/models/vdvae.py
pointer only (licence: NONE) · get_code("aa70735ae381c2da")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, lupalab/posterior-matching when it was checked. 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.

safe_get_module_name Not yet run
unverified (harvested, no recorded run)
deepmind/dm-haiku/haiku/_src/base.py
code served (permissive licence) · get_code("9fb0c6433beb4e20")
How it ran

no invocation recorded.

to_immutable_dict Not yet run
unverified (harvested, no recorded run)
deepmind/dm-haiku/haiku/_src/data_structures.py
code served (permissive licence) · get_code("96c9cc45cd31b59c")
How it ran

no invocation recorded.

to_mutable_dict Not yet run
unverified (harvested, no recorded run)
deepmind/dm-haiku/haiku/_src/data_structures.py
code served (permissive licence) · get_code("dce7b6e14066c25e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 26 samples lifted from this paper's linked repositories, 26 carry a function identity and 0 do not. Of the 26 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 24 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_width_settings
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2011.10650 (2020-11)
Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images
parse_layer_string
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2011.10650 (2020-11)
Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images

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 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Arbitrary conditioning is an important problem in unsupervised learning, where we seek to model the conditional densities p(x u | x o ) that underly some data, for all possible non-intersecting subsets o, u ⊂ {1, . . . , d}. However, the vast majority of density estimation only focuses on modeling the joint distribution p(x), in which important conditional dependencies between features are opaque. We propose a simple and general framework, coined Posterior Matching, that enables Variational Autoencoders (VAEs) to perform arbitrary conditioning, without modification to the VAE itself. Posterior Matching applies to the numerous existing VAE-based approaches to joint density estimation, thereby circumventing the specialized models required by previous approaches to arbitrary conditioning. We find that Posterior Matching is comparable or superior to current state-of-the-art methods for a variety of tasks with an assortment of VAEs (e.g. discrete, hierarchical, VaDE).

For agents

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

get_harvested_code_for_paper("2201.12414")
get_code_for_paper("2201.12414")
have("2201.12414")

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

curl https://syntology.ai/api/ran/2201.12414.json

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

[![Syntology run record](https://syntology.ai/api/ran/2201.12414.svg)](https://syntology.ai/paper/2201.12414)

Connect an agent — have() is free.