SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2411.01409 · NeurIPS · 2024

Classifier-guided Gradient Modulation for Enhanced Multimodal Learning

Zhou Zhao, Tao Jin, Zirun Guo, Jingyuan Chen

arXiv · PDF · Open in the Atlas

Code that ran

17 of 25 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 13 failures of our instrument; 8 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/2411.01409.json · badge: /api/ran/2411.01409.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
zrguo/CGGM canonical (classifier label)
classifier basis: census
12 ran · 1 no recorded run
zrguo/cggm community
Papers with Code archive
5 ran · 7 no recorded run
FunctionStatusWhere it lives
Linear Ran
ran; our contract draft was wrong, not the code · constructed an object; no result was computed
zrguo/CGGM/modules/transformer.py
pointer only (licence: MIT); the stored copy get_code serves, from YvesWang/Machine_Translation_NLP, is pointer only (licence: NONE) · get_code("10865bbb99140edd")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from YvesWang/Machine_Translation_NLP, and GitHub detected no licence in that repository 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.

ResNet101 Ran
ran; our contract draft was wrong, not the code
zrguo/CGGM/models/segmodel.py
code served (permissive licence) · get_code("6aa213f79f005ec4")
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 ResNet. 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)

ARGS = {
    "output_stride": 16,
    "BatchNorm": nn.BatchNorm2d,
    "input_dim": 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 ResNet — a ResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ResNet 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
augment_gaussian_noise Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/CGGM/datasets/BratsDataset.py
code served (permissive licence) · get_code("1fe4d67c3bc4510b")
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 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import random

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

# Create a small sample data array
data_sample = np.random.randn(2, 4).astype(np.float32)

# Define the arguments dictionary
ARGS = {
    "data_sample": data_sample,
    "noise_variance": (0, 0.1)
}
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 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
buffered_future_mask Ran
ran, violated the contract we drafted
zrguo/CGGM/modules/transformer.py
pointer only (licence: MIT); the stored copy get_code serves, from kenford953/graphcage, is pointer only (licence: NONE) · get_code("184a5eddedaf6a42")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from kenford953/graphcage, and GitHub detected no licence in that repository 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.

build_aspp Ran
ran; our contract draft was wrong, not the code
zrguo/CGGM/models/segmodel.py
code served (permissive licence) · get_code("137ee892a73b5311")
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 ASPP. 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)

ARGS = {
    "output_stride": 16,
    "BatchNorm": nn.BatchNorm2d,
}
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 ASPP — a ASPP output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ASPP 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_decoder Ran
ran; our contract draft was wrong, not the code
zrguo/CGGM/models/segmodel.py
code served (permissive licence) · get_code("22ecd809e4d1de39")
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 Decoder. 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 a minimal BatchNorm class that works as expected
BatchNorm = nn.BatchNorm2d

# Define parameters for build_decoder
num_classes = 10
input_heads = 1

ARGS = {
    "num_classes": num_classes,
    "BatchNorm": BatchNorm,
    "input_heads": input_heads,
}
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 Decoder — a Decoder output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Decoder 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
cal_cos Ran
ran; our contract draft was wrong, not the code
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("ed6f4e9e545f8f07")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# The function cal_cos takes:
# - cls_grad: a list (iterable) of tensors
# - fusion_grad: a tensor

# Each element in cls_grad is cloned and viewed as 1D
# fusion_grad is cloned and viewed as 1D
# F.cosine_similarity is called with dim=0 on two 1D tensors

# Create small tensors with distinct shapes
fusion_grad = torch.randn(4)  # 1D tensor of size 4

# cls_grad is a list of tensors, each should be viewable to 1D
# Using different shapes to make them distinct
cls_grad = [
    torch.randn(2, 2),  # Can be viewed as shape (4,)
    torch.randn(4),      # Already 1D with size 4
    torch.randn(1, 4),   # Can be viewed as shape (4,)
]

ARGS = {
    "cls_grad": cls_grad,
    "fusion_grad": fusion_grad,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cutout Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/CGGM/datasets/BratsDataset.py
code served (permissive licence) · get_code("653ae4081b06cab6")
How it ran

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

import numpy as np

# Construct valid arguments for the cutout function
np.random.seed(0)

# Create a small 3D image array (modalities, height, width)
# Using distinct sizes: 3 modalities, 4 height, 5 width
image = np.random.randint(0, 255, size=(3, 4, 5), dtype=np.uint8)

# Create a small 2D label array (height, width)
label = np.random.randint(0, 10, size=(4, 5), dtype=np.uint8)

# Define patch size that fits within the image dimensions
patch_size = (2, 2)

ARGS = {
    "image": image,
    "label": label,
    "patch_size": patch_size
}
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 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
eval_food Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("d67ae69618e8f777")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function expects:
# - results: a tensor that can be viewed as (-1, 101) and has .cpu().detach().numpy()
# - truths: a tensor that can be viewed as (-1) and has .cpu().detach().numpy()

# Create results tensor: shape (2, 101) - 2 samples, 101 classes (food101 dataset)
results = torch.randn(2, 101)

# Create truths tensor: shape (2,) - 2 ground truth labels
truths = torch.tensor([0, 50], dtype=torch.long)

ARGS = {
    "results": results,
    "truths": truths
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
eval_senti Ran
ran; our contract draft was wrong, not the code
zrguo/CGGM/src/eval_metrics.py
code served (permissive licence) · get_code("cae47ad9ef47ac82")
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 float, 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 torch
import numpy as np

torch.manual_seed(0)

# Create small tensors for results and truths
# The function calls .view(-1), .cpu(), .detach(), .numpy() on these
# So they need to be torch tensors
results = torch.randn(2, 4)  # Small tensor that will be flattened
truths = torch.randn(2, 4)   # Small tensor that will be flattened

ARGS = {
    "results": results,
    "truths": truths,
    "exclude_zero": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
fill_with_neg_inf Ran
ran, honoured the contract we drafted
zrguo/CGGM/modules/transformer.py
pointer only (licence: MIT); the stored copy get_code serves, from kenford953/graphcage, is pointer only (licence: NONE) · get_code("8f266b9d616de9b9")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from kenford953/graphcage, and GitHub detected no licence in that repository 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.

make_positions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/CGGM/modules/position_embedding.py
pointer only (licence: MIT); the stored copy get_code serves, from kenford953/graphcage, is pointer only (licence: NONE) · get_code("21958d540bd87a8c")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from kenford953/graphcage, and GitHub detected no licence in that repository 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.

multiclass_acc Ran
ran, honoured the contract we drafted
zrguo/CGGM/src/eval_metrics.py
code served (permissive licence) · get_code("65494cdd1508772e")
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 float64, ndim 0, 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 numpy as np

# Construct arguments for multiclass_acc
# The function takes predictions and ground truth labels
# It rounds both, compares them, and returns accuracy

# Create small batch of predictions (continuous values that will be rounded)
preds = np.array([0.2, 0.8, 0.3, 0.9, 0.1, 0.7])

# Create small batch of ground truth labels (0s and 1s, or continuous values to be rounded)
truths = np.array([0.1, 0.9, 0.2, 1.0, 0.0, 0.8])

ARGS = {
    "preds": preds,
    "truths": truths,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
train_eval_food Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("b39cbc0e9aa13c48")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from sklearn.metrics import accuracy_score, f1_score
import torch.nn as nn

torch.manual_seed(0)

class DummyModel(nn.Module):
    def __init__(self):
        super(DummyModel, self).__init__()
        self.fc = nn.Linear(4, 101)

    def forward(self, x):
        return self.fc(x)

model = DummyModel()

# Create dummy input and target tensors
results = torch.randn(2, 101)  # Simulating model output
truths = torch.randint(0, 101, (2,))  # Simulating true labels

ARGS = {
    'results': results,
    'truths': truths
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
train_eval_iemo Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("ad9faeb1beaacf40")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create test tensors
# results: should be viewable as (-1, 4) and have cpu(), detach(), numpy() methods
# This means it's a torch tensor with shape (batch_size, 4)
results = torch.randn(2, 4)

# truths: should be viewable as (-1) and have cpu(), detach(), numpy() methods
# This means it's a torch tensor with shape (batch_size,)
truths = torch.tensor([0, 2])

ARGS = {
    "results": results,
    "truths": truths
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
train_eval_senti Ran
ran; our contract draft was wrong, not the code
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("7b0ce64e017cea1a")
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 float, 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 torch
import numpy as np

torch.manual_seed(0)

# Create small tensor arguments
results = torch.tensor([[0.5, -0.3, 0.8, -0.1], [0.2, 0.9, -0.4, 0.6]], dtype=torch.float32)
truths = torch.tensor([[1, 0, 1, -1], [0, 1, -1, 1]], dtype=torch.float32)

ARGS = {
    "results": results,
    "truths": truths,
    "exclude_zero": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
weighted_accuracy Ran
ran, honoured the contract we drafted
zrguo/CGGM/src/eval_metrics.py
code served (permissive licence) · get_code("6d4da8365bb0f5d1")
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 float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Analyze the function:
# - test_preds_emo: used with > 0 comparison and np.sum, so needs to be array-like
# - test_truth_emo: used with > 0 comparison and np.sum, so needs to be array-like
# Both should be 1D arrays of numeric values to compute binary classification metrics

# Create small test data
np.random.seed(0)

# Create prediction scores (continuous values)
test_preds_emo = np.array([0.5, -0.3, 0.2, 0.8, -0.5, 0.1])

# Create ground truth scores (continuous values that will be thresholded at 0)
test_truth_emo = np.array([0.7, -0.2, 0.3, 0.9, -0.4, -0.1])

ARGS = {
    "test_preds_emo": test_preds_emo,
    "test_truth_emo": test_truth_emo,
}
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 float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Classifier Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("f70f65ad58a53350")
How it ran

no invocation recorded.

ClassifierGuided Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("f927307b92481f41")
How it ran

no invocation recorded.

FoodModel Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("1cd046b456ad9f31")
How it ran

no invocation recorded.

TransformerEncoder Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("34ea1f3d71361140")
How it ran

no invocation recorded.

TransformerEncoderLayer Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("3dfcc20147c77637")
How it ran

no invocation recorded.

initiate Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("e2fe3744d6f3b381")
How it ran

no invocation recorded.

train_model Not yet run
unverified (harvested, no recorded run)
zrguo/cggm/src/foodtrain.py
code served (permissive licence) · get_code("341db6e7ca441245")
How it ran

no invocation recorded.

train_model Not yet run
unverified (harvested, no recorded run)
zrguo/CGGM/src/segtrain.py
code served (permissive licence) · get_code("27a4ff497fab5848")
How it ran

no invocation recorded.

Also present in other papers' code

Of 25 samples lifted from this paper's linked repositories, 25 carry a function identity and 0 do not. Of the 25 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 6 later) and 19 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
Linear
brick of this paper's assembly: helper
58 other papers
55 dated earlier, 2 later, 1 same year, order not determinable · 5 only as an assembly brick
1409.0473 (2014-09)
Neural Machine Translation by Jointly Learning to Align and Translate
1706.03762 (2017-06)
Attention Is All You Need
1904.01681 (2019-04)
Augmented Neural ODEs
and 55 more not named here
fill_with_neg_inf 16 other papers
11 dated earlier, 4 later, 1 same year, order not determinable · 3 only as an assembly brick
1906.00295 (2019-06)
Multimodal Transformer for Unaligned Multimodal Language Sequences
2007.00808 (2020-07)
Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval
2109.04096 (2021-09)
A Three-Stage Learning Framework for Low-Resource Knowledge-Grounded Dialogue Generation
and 13 more not named here
make_positions
brick of this paper's assembly: helper
9 other papers
7 dated earlier, 2 later · 3 only as an assembly brick
1906.00295 (2019-06)
Multimodal Transformer for Unaligned Multimodal Language Sequences
2111.08456 (2021-11)
Trustworthy Multimodal Regression with Mixture of Normal-inverse Gamma Distributions
2210.17444 (2022-10)
Multimodal Information Bottleneck: Learning Minimal Sufficient Unimodal and Multimodal Representations
and 6 more not named here
buffered_future_mask 7 other papers
5 dated earlier, 2 later · 2 only as an assembly brick
1906.00295 (2019-06)
Multimodal Transformer for Unaligned Multimodal Language Sequences
2111.08456 (2021-11)
Trustworthy Multimodal Regression with Mixture of Normal-inverse Gamma Distributions
Pan_Wnet_Audio-Guided_Video_Object_Segmentation_via_Wavelet-Based_Cross-Modal_Denoising_Networks_CVPR_2022_paper (2022 (year only))
Wnet: Audio-Guided Video Object Segmentation via Wavelet-Based Cross-Modal Denoising Networks
and 4 more not named here
multiclass_acc 5 other papers
3 dated earlier, 2 later
1906.00295 (2019-06)
Multimodal Transformer for Unaligned Multimodal Language Sequences
2109.00181 (2021-09)
CTAL: Pre-training Cross-modal Transformer for Audio-and-Language Representations
2407.02751 (2024-07)
Emotion and Intent Joint Understanding in Multimodal Conversation: A Benchmarking Dataset
and 2 more not named here
weighted_accuracy 2 other papers
1 dated earlier, 1 later
1906.00295 (2019-06)
Multimodal Transformer for Unaligned Multimodal Language Sequences
2412.07121 (2024-12)
Bridging the Gap for Test-Time Multimodal Sentiment Analysis

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

Multimodal learning has developed very fast in recent years. However, during the multimodal training process, the model tends to rely on only one modality based on which it could learn faster, thus leading to inadequate use of other modalities. Existing methods to balance the training process always have some limitations on the loss functions, optimizers and the number of modalities and only consider modulating the magnitude of the gradients while ignoring the directions of the gradients. To solve these problems, in this paper, we present a novel method to balance multimodal learning with Classifier-Guided Gradient Modulation (CGGM), considering both the magnitude and directions of the gradients. We conduct extensive experiments on four multimodal datasets: UPMC-Food 101, CMU-MOSI, IEMOCAP and BraTS 2021, covering classification, regression and segmentation tasks. The results show that CGGM outperforms all the baselines and other stateof-the-art methods consistently, demonstrating its effectiveness and versatility. Our code is available at https://github.com/zrguo/CGGM.

For agents

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

get_harvested_code_for_paper("2411.01409")
get_code_for_paper("2411.01409")
have("2411.01409")

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

curl https://syntology.ai/api/ran/2411.01409.json

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

[![Syntology run record](https://syntology.ai/api/ran/2411.01409.svg)](https://syntology.ai/paper/2411.01409)

Connect an agent — have() is free.