SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.04200 · CVPR · 2022

MaskGIT: Masked Generative Image Transformer

Han Zhang, Lu Jiang, Huiwen Chang, Ce Liu, T Freeman

arXiv · PDF · Open in the Atlas

Code that ran

14 of 21 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (9 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 4 failures of our instrument; 7 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2202.04200.json · badge: /api/ran/2202.04200.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
valeoai/maskgit-pytorch community
Papers with Code archive
6 ran · 2 no recorded run
LAION-AI/phenaki community
Papers with Code archive
3 ran · 1 no recorded run
lucidrains/phenaki-pytorch community
Papers with Code archive
2 ran · 1 no recorded run
alibaba/graph-gpt community
Papers with Code archive
2 ran · 1 no recorded run
dome272/MaskGIT-pytorch community
Papers with Code archive
1 ran · 2 no recorded run
FunctionStatusWhere it lives
AdaNorm Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("db6abd243470c35a")
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 an instance of its own class, AdaNorm: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

# Define ARGS dictionary with all parameters for AdaNorm.__init__
ARGS = {
    'x_dim': 4,
    'y_dim': 3,
}
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 AdaNorm — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a AdaNorm 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
Attention Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("7fd4e0ce7bd6b27f")
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 an instance of its own class, Attention: the run constructed the object and did not call it, so no result was computed. 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
import math

torch.manual_seed(0)

# Define the RMSNorm and QKNorm classes as they are dependencies
class RMSNorm(nn.Module):
    def __init__(self, dim: int, eps: float=1e-05, linear=True, bias=True):
        super().__init__()
        self.eps = eps
        self.linear = linear
        self.add_bias = bias
        if self.linear:
            self.weight = nn.Parameter(torch.ones(dim))
        if self.add_bias:
            self.bias = nn.Parameter(torch.zeros(dim))

    def _norm(self, x):
        return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)

    def forward(self, x):
        output = self._norm(x.float()).type_as(x)
        if self.linear:
            output = self.weight * output
        if self.add_bias:
            output = output + self.bias
        return output

class QKNorm(torch.nn.Module):
    def __init__(self, dim: int):
        super().__init__()
        self.query_norm = RMSNorm(dim, linear=False, bias=False)
        self.key_norm = RMSNorm(dim, linear=False, bias=False)

    def forward(self, q, k, v):
        q = self.query_norm(q)
        k = self.key_norm(k)
        return (q.to(v), k.to(v))

# Create sample input
batch_size = 2
seq_len = 3
embed_dim = 4
num_heads = 2

# Create input tensor
x = torch.randn(batch_size, seq_len, embed_dim)

# Optional mask (can be None)
mask = None

ARGS = {
    "embed_dim": embed_dim,
    "num_heads": num_heads,
    "dropout": 0.1,
    "use_flash": True,
    "bias": False,
}
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 Attention — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Attention 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
Attention Ran
ran on a synthesized input · constructed an object; no result was computed
LAION-AI/phenaki/maskgit.py
pointer only (licence: NONE) · get_code("7e79dc89343596b8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LAION-AI/phenaki 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.

Block Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("c1c45512a825fc3b")
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 an instance of its own class, Block: the run constructed the object and did not call it, so no result was computed. 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)

# Build smallest stand-in models and arguments

# Block parameters
dim = 4
heads = 2
mlp_dim = 8
dropout = 0.1

# Create block instance
block = nn.Module()

# Create sample input tensors
# x: (batch, sequence_length, dim)
x = torch.randn(2, 3, dim)

# cond: conditioning input that goes through mlp to produce 6 chunks
# The mlp output is (batch, dim * 6), so cond should be (batch, some_dim)
# The block's mlp is: nn.Sequential(nn.SiLU(), nn.Linear(dim, dim * 6))
# So cond input should have shape (batch, dim)
cond = torch.randn(2, dim)

# mask: optional attention mask, can be None or a tensor
# Based on forward: mask is reshaped to (b, 1, 1, h_w) in Attention
# So mask should be broadcastable, typically (batch, 1, 1, seq_len) or similar
mask = None  # or could be torch.zeros(2, 1, 1, 3)

ARGS = {
    "dim": dim,
    "heads": heads,
    "mlp_dim": mlp_dim,
    "dropout": dropout,
}
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 Block — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Block 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
ContinuousPositionBias Ran
ran on a synthesized input · constructed an object; no result was computed
lucidrains/phenaki-pytorch/phenaki_pytorch/phenaki_pytorch.py
code served (permissive licence) · get_code("220f01cece0c0d8b")
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 an instance of its own class, ContinuousPositionBias: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    'dim': 8,
    'heads': 4,
    'num_dims': 2,
    'layers': 2,
    'log_dist': True,
    'cache_rel_pos': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ContinuousPositionBias — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ContinuousPositionBias 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
Encoder Ran
ran on a synthesized input
dome272/MaskGIT-pytorch/bidirectional_transformer.py
code served (permissive licence) · get_code("6ee8c0748ff579f0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

INIT_ARGS = {
    'dim': 8,
    'hidden_dim': 16
}

ARGS = {
    'x': torch.randn(2, 5, 8)  # (batch_size=2, seq_len=5, dim=8)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
FeedForward Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("9bdc5ed60c59ed33")
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 an instance of its own class, FeedForward: the run constructed the object and did not call it, so no result was computed. 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)

# FeedForward is an nn.Module that takes:
# - dim: input/output dimension
# - h_dim: hidden dimension
# - multiple_of: alignment for hidden dimension (default 256)
# - bias: whether to use bias in Linear layers (default False)
# - dropout: dropout rate (default 0.0)

# The forward method expects input x of shape (batch, dim)
# It applies linear layers w1 and w3, multiplies them after SiLU,
# optionally applies dropout, then applies w2

ARGS = {
    "dim": 64,
    "h_dim": 128,
    "multiple_of": 256,
    "bias": False,
    "dropout": 0.1,
}
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 FeedForward — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a FeedForward 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
FeedForward Ran
ran; our contract draft was wrong, not the code
LAION-AI/phenaki/maskgit.py
pointer only (licence: NONE) · get_code("37eef98f3b06bc78")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LAION-AI/phenaki 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.

QKNorm Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("3c34ec2ff05d28fd")
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 an instance of its own class, QKNorm: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

class RMSNorm(nn.Module):

    def __init__(self, dim: int, eps: float=1e-05, linear=True, bias=True):
        super().__init__()
        self.eps = eps
        self.linear = linear
        self.add_bias = bias
        if self.linear:
            self.weight = nn.Parameter(torch.ones(dim))
        if self.add_bias:
            self.bias = nn.Parameter(torch.zeros(dim))

    def _norm(self, x):
        return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)

    def forward(self, x):
        output = self._norm(x.float()).type_as(x)
        if self.linear:
            output = self.weight * output
        if self.add_bias:
            output = output + self.bias
        return output

class QKNorm(torch.nn.Module):

    def __init__(self, dim: int):
        super().__init__()
        self.query_norm = RMSNorm(dim, linear=False, bias=False)
        self.key_norm = RMSNorm(dim, linear=False, bias=False)

    def forward(self, q, k, v):
        q = self.query_norm(q)
        k = self.key_norm(k)
        return (q.to(v), k.to(v))

# Construct arguments for QKNorm
dim = 4

ARGS = {
    "dim": dim,
}
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 QKNorm — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a QKNorm 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
RMSNorm Ran
ran on a synthesized input · constructed an object; no result was computed
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("86a159a05ae09375")
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 an instance of its own class, RMSNorm: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "dim": 4,
    "eps": 1e-05,
    "linear": True,
    "bias": True,
}
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 RMSNorm — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RMSNorm 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
Transformer Ran
ran on a synthesized input · constructed an object; no result was computed
LAION-AI/phenaki/maskgit.py
pointer only (licence: NONE) · get_code("c86e4af300cafb7a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LAION-AI/phenaki 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.

_batch_unmask_without_for_loop Ran
ran; our contract draft was wrong, not the code
alibaba/graph-gpt/src/utils/generation_utils.py
code served (permissive licence) · get_code("93bc9d37c053dcd3")
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 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 torch.nn as nn
from dataclasses import dataclass, field
from typing import Optional, Dict, Any
from transformers import __version__

torch.manual_seed(0)

@dataclass
class GenerationConfig:
    """
    Configuration for masked diffusion generation, designed for use with Hydra.

    This class holds all parameters for the generation process. It is instantiated
    by Hydra from a YAML configuration file.
    """
    alg: str = 'origin'
    alg_temp: Optional[float] = None
    steps: int = 512
    eps: float = 0.001
    parallel_gen: bool = False
    temperature: float = 0.0
    top_p: Optional[float] = None
    top_k: Optional[int] = None
    max_length: int = 20
    max_new_tokens: Optional[int] = None
    num_return_sequences: int = 1
    return_dict_in_generate: bool = False
    output_history: bool = False
    mask_token_id: Optional[int] = None
    pad_token_id: Optional[int] = None
    bos_token_id: Optional[int] = None
    eos_token_id: Optional[int] = None
    generation_kwargs: Dict[str, Any] = field(default_factory=dict)
    _from_model_config: bool = False
    _commit_hash: Optional[str] = None
    transformers_version: str = field(default=__version__)

    def __post_init__(self):
        """
        This method is called by dataclasses after the object is created.
        It's the perfect place for validation logic.
        """
        self.validate()

    def validate(self):
        """
        Validates the configuration parameters. This is the same method
        as in the original class.
        """
        import warnings
        if self.alg not in ['origin', 'maskgit_plus', 'topk_margin', 'entropy']:
            warnings.warn(f"Unknown algorithm '{self.alg}' specified. Valid options are: 'origin', 'maskgit_plus', 'topk_margin', 'entropy'")
        if self.temperature < 0:
            raise ValueError('`temperature` must be non-negative.')
        if self.steps <= 0:
            raise ValueError('`steps` must be positive.')

# Create a small GenerationConfig instance
cfg = GenerationConfig(
    alg='origin',
    alg_temp=None,
    steps=512,
    eps=0.001,
    parallel_gen=False,
    temperature=0.0,
    top_p=None,
    top_k=None,
    mask_token_id=0,
    pad_token_id=1,
    bos_token_id=2,
    eos_token_id=3,
)

# Create tensors for the function
batch_size = 2
seq_length = 4
vocab_size = 5

# x: token ids with some mask tokens (mask_token_id=0)
x = torch.tensor([
    [1, 0, 2, 3],
    [1, 2, 0, 0]
], dtype=torch.long)

# logits: (batch_size, seq_length, vocab_size)
logits = torch.randn(batch_size, seq_length, vocab_size)

# timesteps: schedule of timesteps for diffusion
timesteps = torch.tensor([1000, 500, 250, 100, 50, 10, 1], dtype=torch.long)

# i: current step index
i = 0

ARGS = {
    'x': x,
    'logits': logits,
    'timesteps': timesteps,
    'i': i,
    'cfg': cfg,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, transformers — 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
leaky_relu Ran
ran; our contract draft was wrong, not the code
lucidrains/phenaki-pytorch/phenaki_pytorch/phenaki_pytorch.py
code served (permissive licence) · get_code("f11bef5e87ce3904")
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 LeakyReLU. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "p": 0.1
}
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 LeakyReLU — a LeakyReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LeakyReLU 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
sample_tokens Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
alibaba/graph-gpt/src/utils/generation_utils.py
code served (permissive licence) · get_code("3028b4a113e4e8f8")
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 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# Create a small logits tensor with shape (batch_size=2, vocab_size=4)
logits = torch.randn(2, 4)

ARGS = {
    "logits": logits,
    "temperature": 1.0,
    "top_p": 0.9,
    "top_k": 3,
    "margin_confidence": False,
    "neg_entropy": False,
}
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 tuple, ndim 2, 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
BidirectionalTransformer Not yet run
unverified (harvested, no recorded run)
dome272/MaskGIT-pytorch/bidirectional_transformer.py
code served (permissive licence) · get_code("b3029d867f4bb08d")
How it ran

no invocation recorded.

GenerationConfig Not yet run
unverified (harvested, no recorded run)
alibaba/graph-gpt/src/utils/generation_utils.py
code served (permissive licence) · get_code("def2e2196cd17208")
How it ran

no invocation recorded.

MaskGit Not yet run
unverified (harvested, no recorded run)
lucidrains/phenaki-pytorch/phenaki_pytorch/phenaki_pytorch.py
code served (permissive licence) · get_code("d683bf3267136071")
How it ran

no invocation recorded.

MaskGit Not yet run
unverified (harvested, no recorded run)
LAION-AI/phenaki/maskgit.py
pointer only (licence: NONE) · get_code("78ffd63b86b6fe34")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LAION-AI/phenaki 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.

Transformer Not yet run
unverified (harvested, no recorded run)
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("d14b816d459cd10d")
How it ran

no invocation recorded.

TransformerEncoder Not yet run
unverified (harvested, no recorded run)
valeoai/maskgit-pytorch/Network/transformer.py
code served (permissive licence) · get_code("ae6080e4032dcb9a")
How it ran

no invocation recorded.

weights_init Not yet run
unverified (harvested, no recorded run)
dome272/MaskGIT-pytorch/bidirectional_transformer.py
code served (permissive licence) · get_code("323141756998f1c6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 21 samples lifted from this paper's linked repositories, 21 carry a function identity and 0 do not. Of the 21 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 20 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
leaky_relu
brick of this paper's assembly: helper
2 other papers
2 later
2410.03143 (2024-10)
ECHOPulse: ECG controlled echocardio-grams video generation
2410.11758 (2024-10)
Latent Action Pretraining from Videos

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

Class-conditional Image Generation (b) Image Manipulation (c) Image Extrapolation Flamingo Tram Input Input Figure 1. Example generation by MaskGIT on image synthesis and manipulation tasks. We show that MaskGIT is a flexible model that can generate high-quality samples on (a) class-conditional synthesis, (b) class-conditional image manipulation, e.g. replacing selected objects in the bounding box with ones from the given classes, and (c) image extrapolation. Examples shown here have resolutions 512ˆ512, 512ˆ512, and 512ˆ2560 in the three columns, respectively. Zoom in to see the details.

For agents

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

get_harvested_code_for_paper("2202.04200")
get_code_for_paper("2202.04200")
have("2202.04200")

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

curl https://syntology.ai/api/ran/2202.04200.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.04200.svg)](https://syntology.ai/paper/2202.04200)

Connect an agent — have() is free.