SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2206.08155 · NeurIPS · 2022

Zero-Shot Video Question Answering via Frozen Bidirectional Language Models

Cordelia Schmid, Ivan Laptev, Antoine Miech, Antoine Yang, Josef Sivic, Pretrained Bilm, Running Frozenbilm

arXiv · PDF · Open in the Atlas

Code that ran

14 of 34 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; 1 honoured the contract we drafted, 1 violated it and 12 have no contract verdict; 0 failures of our instrument; 20 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/2206.08155.json · badge: /api/ran/2206.08155.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
antoyang/FrozenBiLM canonical (classifier label)
classifier basis: classifier_2025
0 ran · 7 no recorded run
sts-vlcc/sts-vlcc community
Papers with Code archive
14 ran · 8 no recorded run
klauscc/dam community
Papers with Code archive
0 ran · 4 no recorded run
copy not recorded community
not in the paper's declared repositories
0 ran · 1 no recorded run
FunctionStatusWhere it lives
Adapter Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("020a7d784fec3e8b")
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, Adapter: 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 = {
    "ds_factor": 2,
    "hidden_dim": 8,
    "ln_after": False,
    "ln_before": 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 Adapter — 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 Adapter 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
ConvLayer Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("be2fdff3546c4ee9")
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, ConvLayer: 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
from transformers.activations import ACT2FN

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.hidden_size = 4
        self.conv_kernel_size = 3
        self.conv_groups = 1
        self.conv_act = 'tanh'
        self.hidden_dropout_prob = 0.1
        self.layer_norm_eps = 1e-12

config = Config()

batch_size = 2
seq_length = 3
hidden_size = 4

hidden_states = torch.randn(batch_size, seq_length, hidden_size)
residual_states = torch.randn(batch_size, seq_length, hidden_size)
input_mask = torch.ones(batch_size, seq_length)

ARGS = {
    'config': config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvLayer — 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 ConvLayer 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
DebertaV2Embeddings Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("fb069d6e466f4cf8")
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, DebertaV2Embeddings: 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)

class MockConfig:
    def __init__(self):
        self.vocab_size = 100
        self.hidden_size = 8
        self.embedding_size = 8
        self.max_position_embeddings = 512
        self.type_vocab_size = 2
        self.layer_norm_eps = 1e-12
        self.hidden_dropout_prob = 0.1
        self.pad_token_id = 0
        self.position_biased_input = True

config = MockConfig()
features_dim = 6

ARGS = {
    'config': config,
    'features_dim': features_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 DebertaV2Embeddings — 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 DebertaV2Embeddings 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
DebertaV2Intermediate Ran
ran on a synthesized input
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("713015c84afd123d")
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 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
from torch import nn
from transformers.activations import ACT2FN

class Config:
    def __init__(self):
        self.hidden_size = 4
        self.intermediate_size = 3
        self.hidden_act = "gelu"

INIT_ARGS = {
    "config": Config()
}

ARGS = {
    "hidden_states": torch.randn(2, 4)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float — no output shape is stated: observe_signature.py 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)
DebertaV2LMPredictionHead Ran
ran on a synthesized input
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("4f4202be5c178fde")
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
from torch import nn
from types import SimpleNamespace

# Create a minimal config object
config = SimpleNamespace(
    hidden_size=8,
    vocab_size=12,
    hidden_act='gelu',
    layer_norm_eps=1e-7
)

INIT_ARGS = {
    'config': config
}

# For the forward method, we need:
# - hidden_states: tensor of shape (batch, seq_len, hidden_size)
# - embedding_weight: tensor of shape (vocab_size, hidden_size)
# - bias: optional tensor of shape (vocab_size,)

batch_size = 2
seq_len = 5
hidden_size = config.hidden_size
vocab_size = config.vocab_size

hidden_states = torch.randn(batch_size, seq_len, hidden_size)
embedding_weight = torch.randn(vocab_size, hidden_size)

ARGS = {
    'hidden_states': hidden_states,
    'embedding_weight': embedding_weight,
    'bias': None
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch, transformers — 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)
DebertaV2OnlyMLMHead Ran
ran on a synthesized input
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("491522c8a24566a6")
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
from torch import nn

class SimpleConfig:
    def __init__(self):
        self.hidden_size = 8
        self.vocab_size = 12
        self.hidden_act = "gelu"
        self.layer_norm_eps = 1e-7

config = SimpleConfig()

INIT_ARGS = {
    "config": config
}

batch_size = 2
seq_length = 5
hidden_size = config.hidden_size
vocab_size = config.vocab_size

sequence_output = torch.randn(batch_size, seq_length, hidden_size)
embedding_weight = torch.randn(vocab_size, hidden_size)

ARGS = {
    "sequence_output": sequence_output,
    "embedding_weight": embedding_weight,
    "bias": None
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch, transformers — 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)
DebertaV2Output Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("d3b79844fb5a1268")
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, DebertaV2Output: 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
from torch.nn import LayerNorm

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.intermediate_size = 8
        self.hidden_size = 4
        self.layer_norm_eps = 1e-12
        self.hidden_dropout_prob = 0.1

config = Config()
ds_factor = 2
dropout = 0.1

ARGS = {
    "config": config,
    "ds_factor": ds_factor,
    "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 DebertaV2Output — 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 DebertaV2Output 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
DebertaV2PreTrainedModel Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("48742f01e32d514b")
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, DebertaV2PreTrainedModel: 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
from transformers import DebertaV2Config

torch.manual_seed(0)

# Create a minimal config
config = DebertaV2Config(
    vocab_size=1000,
    hidden_size=4,
    num_hidden_layers=2,
    num_attention_heads=2,
    intermediate_size=8,
    hidden_act="gelu",
    hidden_dropout_prob=0.1,
    attention_probs_dropout_prob=0.1,
    max_position_embeddings=512,
    type_vocab_size=2,
    initializer_range=0.02,
    layer_norm_eps=1e-12,
    num_labels=3,
)

ARGS = {
    'config': config,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DebertaV2PreTrainedModel — 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 DebertaV2PreTrainedModel 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
DebertaV2SelfOutput Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("a14cfd37805f8d1a")
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, DebertaV2SelfOutput: 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
from torch.nn import LayerNorm

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.hidden_size = 8
        self.layer_norm_eps = 1e-12
        self.hidden_dropout_prob = 0.1

config = Config()

ARGS = {
    "config": config,
    "ds_factor": 2,
    "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 DebertaV2SelfOutput — 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 DebertaV2SelfOutput 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
StableDropout Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("104d374e92036a6f")
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, StableDropout: 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)

# StableDropout.__init__ takes only drop_prob as a parameter
# The forward method uses self.training and self.drop_prob
# The other methods are called internally

ARGS = {
    'drop_prob': 0.5
}
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 StableDropout — 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 StableDropout 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
XDropout Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("1f587485f1c0247d")
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, XDropout: 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

torch.manual_seed(0)

class DropoutContext(object):
    def __init__(self):
        self.dropout = 0
        self.mask = None
        self.scale = 1
        self.reuse_mask = True

# Create a small input tensor (batch_size=2, dim=4)
input_tensor = torch.randn(2, 4)

# Create a DropoutContext with dropout probability
local_ctx = DropoutContext()
local_ctx.dropout = 0.5
local_ctx.mask = None
local_ctx.scale = 1.0
local_ctx.reuse_mask = True

ARGS = {
    'input': input_tensor,
    'local_ctx': local_ctx
}
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 XDropout — 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 XDropout 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
XSoftmax Ran
ran on a synthesized input · constructed an object; no result was computed
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("d4b3d76edcfa3131")
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, XSoftmax: 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)

# Create input tensor with shape [2, 4, 3]
input_tensor = torch.randn(2, 4, 3)

# Create mask tensor with same shape - 1s indicate valid positions, 0s indicate positions to ignore
mask = torch.ones(2, 4, 3, dtype=torch.int)
# Set some positions to 0 to create masked regions
mask[0, 1, :] = 0

# Dimension to apply softmax on
dim = -1

ARGS = {
    "input": input_tensor,
    "mask": mask,
    "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 XSoftmax — 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 XSoftmax 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
build_relative_position Ran
ran, honoured the contract we drafted
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("0464dc8467b456dc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family int, shape (1, 5, 7) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch

torch.manual_seed(0)

ARGS = {
    "query_size": 5,
    "key_size": 7,
    "bucket_size": 32,
    "max_position": 128,
}
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 Tensor, ndim 3, dtype family int, shape (1, 5, 7) — shape (1, 5, 7) 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
make_log_bucket_position Ran
ran, violated the contract we drafted
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("567c8b010f8e20d2")
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 ndarray, ndim 1, dtype family int, shape (8,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct arguments for make_log_bucket_position
# The function takes:
# - relative_pos: array of relative positions
# - bucket_size: integer size of buckets
# - max_position: integer maximum position value

# Create a small array of relative positions
relative_pos = np.array([-10, -5, -2, 0, 2, 5, 10, 15], dtype=np.int32)

# Set bucket_size to a reasonable value
bucket_size = 8

# Set max_position larger than the max absolute value in relative_pos
max_position = 32

ARGS = {
    "relative_pos": relative_pos,
    "bucket_size": bucket_size,
    "max_position": max_position,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (8,) — shape (8,) 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
BaseModelOutput Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("06f89381bbcaa62a")
How it ran

no invocation recorded.

BaseModelOutput Not yet run
unverified (harvested, no recorded run)
klauscc/dam/model/deberta.py
code served (permissive licence) · get_code("5c632e836a01b0f1")
How it ran

no invocation recorded.

DebertaV2Attention Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("19240ea1e9c179bc")
How it ran

no invocation recorded.

DebertaV2Attention Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("ed339d8069718237")
How it ran

no invocation recorded.

DebertaV2Embeddings Not yet run
unverified (harvested, no recorded run)
klauscc/dam/model/deberta.py
code served (permissive licence) · get_code("d402090dd56a6f46")
How it ran

no invocation recorded.

DebertaV2Encoder Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("0905316df8bf0d3e")
How it ran

no invocation recorded.

DebertaV2Encoder Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("deded63a5e7d8699")
How it ran

no invocation recorded.

DebertaV2Encoder Not yet run
unverified (harvested, no recorded run)
klauscc/dam/model/deberta.py
code served (permissive licence) · get_code("cc3b22c7c43000bf")
How it ran

no invocation recorded.

DebertaV2ForMaskedLM Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("24dd2616a44d83c7")
How it ran

no invocation recorded.

DebertaV2Layer Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("b76e671720991509")
How it ran

no invocation recorded.

DebertaV2Layer Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("32cb0a816174decd")
How it ran

no invocation recorded.

DebertaV2Model Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("6a90151a5129dc8e")
How it ran

no invocation recorded.

DebertaV2Model Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("f576b70e9c2bc9c4")
How it ran

no invocation recorded.

DebertaV2Model Not yet run
unverified (harvested, no recorded run)
klauscc/dam/model/deberta.py
code served (permissive licence) · get_code("63c2fffe385cf5e9")
How it ran

no invocation recorded.

DisentangledSelfAttention Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("f5ca472802a4cfb0")
How it ran

no invocation recorded.

DisentangledSelfAttention Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("b2ac2b6c1c0b64dc")
How it ran

no invocation recorded.

DropoutContext Not yet run
unverified (harvested, no recorded run)
sts-vlcc/sts-vlcc/model/deberta.py
code served (permissive licence) · get_code("ae0fc5a9e16f1c5d")
How it ran

no invocation recorded.

build_relative_position Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("b1852809457a4b1f")
How it ran

no invocation recorded.

get_mask Not yet run
unverified (harvested, no recorded run)
this paper's copy was not recorded; identical code first harvested from whu-zqh/panda
pointer only · get_code("05732f1a5126672d")
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. No invocation recorded.

make_log_bucket_position Not yet run
unverified (harvested, no recorded run)
antoyang/FrozenBiLM/model/deberta.py
code served (permissive licence) · get_code("29371b9c270911e1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 34 samples lifted from this paper's linked repositories, 34 carry a function identity and 0 do not. Of the 27 distinct functions with an identity, 4 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 4 later) and 23 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_mask
brick of this paper's assembly: helper
10 other papers
2 dated earlier, 8 later
1906.03158 (2019-06)
Matching the Blanks: Distributional Similarity for Relation Learning
2110.00976 (2021-10)
LexGLUE: A Benchmark Dataset for Legal Language Understanding in English
2207.07087 (2022-07)
Parameter-Efficient Prompt Tuning Makes Generalized and Calibrated Neural Text Retrievers
and 7 more not named here
build_relative_position
brick of this paper's assembly: helper
7 other papers
7 later
2207.07087 (2022-07)
Parameter-Efficient Prompt Tuning Makes Generalized and Calibrated Neural Text Retrievers
2210.12360 (2022-10)
Prompt-Tuning Can Be Much Better Than Fine-Tuning on Cross-lingual Understanding With Multilingual Language Models
2211.05414 (2022-11)
ADEPT: A DEbiasing PrompT Framework
and 4 more not named here
make_log_bucket_position
brick of this paper's assembly: helper
7 other papers
1 dated earlier, 6 later · 1 only as an assembly brick
2006.03654 (2020-06)
DeBERTa: Decoding-enhanced BERT with Disentangled Attention
2207.07087 (2022-07)
Parameter-Efficient Prompt Tuning Makes Generalized and Calibrated Neural Text Retrievers
2210.12360 (2022-10)
Prompt-Tuning Can Be Much Better Than Fine-Tuning on Cross-lingual Understanding With Multilingual Language Models
and 4 more not named here
make_log_bucket_position
brick of this paper's assembly: helper
1 other paper
1 later
2309.01327 (2023-09)
Can I Trust Your Answer? Visually Grounded Video Question Answering

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

Video question answering (VideoQA) is a complex task that requires diverse multimodal data for training. Manual annotation of question and answers for videos, however, is tedious and prohibits scalability. To tackle this problem, recent methods consider zero-shot settings with no manual annotation of visual question-answer. In particular, a promising approach adapts frozen autoregressive language models pretrained on Web-scale text-only data to multi-modal inputs. In contrast, we here build on frozen bidirectional language models (BiLM) and show that such an approach provides a stronger and cheaper alternative for zero-shot VideoQA. In particular, (i) we combine visual inputs with the frozen BiLM using light trainable modules, (ii) we train such modules using Web-scraped multi-modal data, and finally (iii) we perform zero-shot VideoQA inference through masked language modeling, where the masked text is the answer to a given question. Our proposed approach, FrozenBiLM, outperforms the state of the art in zero-shot VideoQA by a significant margin on a variety of datasets, including LSMDC-FiB, iVQA, MSRVTT-QA, MSVD-QA, ActivityNet-QA, TGIF-FrameQA, How2QA and TVQA. It also demonstrates competitive performance in the few-shot and fully-supervised setting. Our code and models are publicly available at [1]. Little cute toy poodle dog running fast on the beach.

For agents

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

get_harvested_code_for_paper("2206.08155")
get_code_for_paper("2206.08155")
have("2206.08155")

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

curl https://syntology.ai/api/ran/2206.08155.json

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

[![Syntology run record](https://syntology.ai/api/ran/2206.08155.svg)](https://syntology.ai/paper/2206.08155)

Connect an agent — have() is free.