SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2209.00189 · ICML · 2022

Federated Learning with Label Distribution Skew via Logits Calibration

Bo Li, Jie Zhang, Shuang Wu, Shouhong Ding, Chao Wu, Zhiqi Li, Jianghe Xu

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 1 failure of our instrument; 1 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/2209.00189.json · badge: /api/ran/2209.00189.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
bytedance/feddecorr community
Papers with Code archive
4 ran · 0 no recorded run
thejungwon/gc-fed community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
FedAvg Ran
ran on a synthesized input · constructed an object; no result was computed
bytedance/feddecorr/approach/fedlogitcal.py
code served (permissive licence) · get_code("af8f3034a8ca7b6f")
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, FedAvg: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 unittest.mock import Mock
import logging

torch.manual_seed(0)

# Create a minimal model that satisfies the usage in FedAvg
class SimpleModel(nn.Module):
    def __init__(self):
        super(SimpleModel, self).__init__()
        self.fc1 = nn.Linear(4, 8)
        self.fc2 = nn.Linear(8, 3)
    
    def forward(self, x):
        x = torch.relu(self.fc1(x))
        x = self.fc2(x)
        return x

# Create minimal dataloaders (list of (x, y) tuples)
def create_dataloader(num_samples=8, input_dim=4, num_classes=3):
    data = []
    for i in range(num_samples):
        x = torch.randn(2, input_dim)
        y = torch.randint(0, num_classes, (2,))
        data.append((x, y))
    return data

# Create logger
logger = logging.getLogger('test')
logger.setLevel(logging.INFO)
handler = logging.StreamHandler()
logger.addHandler(handler)

# Build arguments dict
ARGS = {
    'args': Mock(
        n_comm_round=2,
        print_interval=1,
        save_interval=2,
        ckptdir='/tmp',
        approach='fed_avg',
        log_file_name='test'
    ),
    'appr_args': Mock(feddecorr=False, feddecorr_coef=0.1),
    'logger': logger,
    'party_list_rounds': [[0, 1], [0, 1]],
    'party2nets': {
        0: SimpleModel(),
        1: SimpleModel()
    },
    'global_net': SimpleModel(),
    'party2loaders': {
        0: create_dataloader(num_samples=8),
        1: create_dataloader(num_samples=8)
    },
    'global_train_dl': create_dataloader(num_samples=8),
    'test_dl': create_dataloader(num_samples=8)
}
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 FedAvg — 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 FedAvg 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
FedDecorrLoss Ran
ran on a synthesized input
bytedance/feddecorr/approach/fedlogitcal.py
code served (permissive licence) · get_code("f64d951d32b9b6d8")
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 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 torch
import torch.nn as nn

class FedDecorrLoss(nn.Module):

    def __init__(self):
        super(FedDecorrLoss, self).__init__()
        self.eps = 1e-08

    def _off_diagonal(self, mat):
        n, m = mat.shape
        assert n == m
        return mat.flatten()[:-1].view(n - 1, n + 1)[:, 1:].flatten()

    def forward(self, x):
        N, C = x.shape
        if N == 1:
            return 0.0
        x = x - x.mean(dim=0, keepdim=True)
        x = x / torch.sqrt(self.eps + x.var(dim=0, keepdim=True))
        corr_mat = torch.matmul(x.t(), x)
        loss = self._off_diagonal(corr_mat).pow(2).mean()
        loss = loss / N
        return loss


INIT_ARGS = {}

ARGS = {
    'x': torch.randn(2, 4)
}

if __name__ == '__main__':
    model = FedDecorrLoss(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output: {output}")
    print(f"Output type: {type(output)}")
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 0, 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)
FedLogitCal Ran
ran on a synthesized input · constructed an object; no result was computed
bytedance/feddecorr/approach/fedlogitcal.py
code served (permissive licence) · get_code("1c10c22b09b2f3ce")
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, FedLogitCal: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 numpy as np
from unittest.mock import Mock
import logging

torch.manual_seed(0)

# Simple model that supports return_features
class SimpleNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.fc1 = nn.Linear(4, 8)
        self.fc2 = nn.Linear(8, 3)
    
    def forward(self, x, return_features=False):
        features = self.fc1(x)
        out = self.fc2(features)
        if return_features:
            return out, features
        return out

# Create simple dataloaders as lists of (x, y) tuples
def create_dataloader(n_samples=8, input_dim=4, num_classes=3):
    data = []
    for i in range(n_samples):
        x = torch.randn(input_dim)
        y = torch.randint(0, num_classes, (1,)).item()
        data.append((x, y))
    return data

# Logger mock
logger = Mock(spec=logging.Logger)
logger.info = Mock()

# Args namespace
class ArgsNamespace:
    def __init__(self):
        self.n_comm_round = 2
        self.epochs = 1
        self.lr = 0.01
        self.rho = 0.9
        self.weight_decay = 1e-5
        self.print_interval = 1
        self.save_interval = 2
        self.ckptdir = "/tmp"
        self.approach = "test"
        self.log_file_name = "test_log"

args = ArgsNamespace()

# Approch args
class ApprArgsNamespace:
    def __init__(self):
        self.feddecorr = False
        self.feddecorr_coef = 0.1

appr_args = ApprArgsNamespace()

# Create party networks
party2nets = {
    0: SimpleNet(),
    1: SimpleNet(),
}

# Global network
global_net = SimpleNet()

# Create party dataloaders
party2loaders = {
    0: create_dataloader(n_samples=8),
    1: create_dataloader(n_samples=6),
}

# Global train and test dataloaders
global_train_dl = create_dataloader(n_samples=10)
test_dl = create_dataloader(n_samples=8)

# Party list for each round
party_list_rounds = [
    [0, 1],
    [0, 1],
]

ARGS = {
    'args': args,
    'appr_args': appr_args,
    'logger': logger,
    'party_list_rounds': party_list_rounds,
    'party2nets': party2nets,
    'global_net': global_net,
    'party2loaders': party2loaders,
    'global_train_dl': global_train_dl,
    'test_dl': test_dl,
}
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 FedLogitCal — 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 FedLogitCal output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_accuracy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bytedance/feddecorr/approach/fedlogitcal.py
code served (permissive licence) · get_code("e7e5bb878fd22400")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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

torch.manual_seed(0)

# Build a minimal model that takes input and outputs logits
class SimpleClassifier(nn.Module):
    def __init__(self, input_dim=4, num_classes=3):
        super().__init__()
        self.linear = nn.Linear(input_dim, num_classes)
    
    def forward(self, x):
        return self.linear(x)

# Create model instance
model = SimpleClassifier(input_dim=4, num_classes=3)
model.train()  # Start in training mode so the function can switch it

# Create a simple dataloader (list of batches with (x, target) tuples)
# Batch size 2, input dimension 4, 3 classes
dataloader = [
    (torch.randn(2, 4), torch.tensor([0, 1], dtype=torch.int64)),
    (torch.randn(2, 4), torch.tensor([2, 0], dtype=torch.int64)),
]

ARGS = {
    "model": model,
    "dataloader": dataloader,
}
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 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
FedLCClient Not yet run
unverified (harvested, no recorded run)
thejungwon/gc-fed/algorithms/fedlc.py
pointer only (licence: NONE) · get_code("6c35df54afa9cd25")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, thejungwon/gc-fed 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.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 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 4 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
FedDecorrLoss
brick of this paper's assembly: loss
1 other paper
1 later · 1 only as an assembly brick
2210.00226 (2022-10)
Towards Understanding and Mitigating Dimensional Collapse in Heterogeneous Federated Learning

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

Traditional federated optimization methods perform poorly with heterogeneous data (i.e. , accuracy reduction), especially for highly skewed data. In this paper, we investigate the label distribution skew in FL, where the distribution of labels varies across clients. First, we investigate the label distribution skew from a statistical view. We demonstrate both theoretically and empirically that previous methods based on softmax cross-entropy are not suitable, which can result in local models heavily overfitting to minority classes and missing classes. Additionally, we theoretically introduce a deviation bound to measure the deviation of the gradient after local update. At last, we propose FedLC (Federated learning via Logits Calibration), which calibrates the logits before softmax cross-entropy according to the probability of occurrence of each class. FedLC applies a fine-grained calibrated cross-entropy loss into local update by adding a pairwise label margin. Extensive experiments on federated datasets and real-world datasets demonstrate that FedLC leads to a more accurate global model and much improved performance. Furthermore, integrating other FL methods into our approach can further enhance the performance of the global model.

For agents

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

get_harvested_code_for_paper("2209.00189")
get_code_for_paper("2209.00189")
have("2209.00189")

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

curl https://syntology.ai/api/ran/2209.00189.json

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

[![Syntology run record](https://syntology.ai/api/ran/2209.00189.svg)](https://syntology.ai/paper/2209.00189)

Connect an agent — have() is free.