SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2607.07720 · 2026

Omni-Sleep: A Sleep Foundation Model via Hierarchical Contrastive Learning of CNS-ANS Dynamics

Song Wang, Chen Wei, Quanying Liu, Mo Wang, Kexin Lou, Zhoujie Hou

arXiv · PDF

Code that ran

13 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 12 have no contract verdict; 1 failure of our instrument; 2 no recorded run), measured 2026-11-22; 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/2607.07720.json · badge: /api/ran/2607.07720.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
AutoBrain-sleep/OmniSleep community
Papers with Code archive
13 ran · 2 no recorded run
FunctionStatusWhere it lives
DSConv Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("186c4f54da0b3d2f")
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, shape (2, 8, 16) 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 torch.nn as nn

class DSConv(nn.Module):

    def __init__(self, in_ch, out_ch, kernel_size, stride=1, padding=0, dilation=1):
        super().__init__()
        self.depthwise = nn.Sequential(nn.Conv1d(in_ch, in_ch, kernel_size, stride, padding, dilation=dilation, groups=in_ch, bias=False), nn.BatchNorm1d(in_ch), nn.ReLU6(inplace=True))
        self.pointwise = nn.Sequential(nn.Conv1d(in_ch, out_ch, kernel_size=1, bias=False), nn.BatchNorm1d(out_ch), nn.ReLU6(inplace=True))

    def forward(self, x):
        return self.pointwise(self.depthwise(x))


INIT_ARGS = {
    'in_ch': 4,
    'out_ch': 8,
    'kernel_size': 3,
    'stride': 1,
    'padding': 1,
    'dilation': 1,
}

ARGS = {
    'x': torch.randn(2, 4, 16),
}

if __name__ == '__main__':
    model = DSConv(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
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, shape (2, 8, 16) — shape (2, 8, 16) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ECG_Encoder_ResNet Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("de641cdd23a33180")
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, shape (2, 64) 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 torch.nn as nn

class ResBlock(nn.Module):

    def __init__(self, in_channels, out_channels, stride=1):
        super(ResBlock, self).__init__()
        self.conv1 = nn.Conv1d(in_channels, out_channels, kernel_size=5, stride=stride, padding=2, bias=False)
        self.bn1 = nn.BatchNorm1d(out_channels)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv1d(out_channels, out_channels, kernel_size=5, stride=1, padding=2, bias=False)
        self.bn2 = nn.BatchNorm1d(out_channels)
        self.downsample = None
        if stride != 1 or in_channels != out_channels:
            self.downsample = nn.Sequential(nn.Conv1d(in_channels, out_channels, kernel_size=1, stride=stride, bias=False), nn.BatchNorm1d(out_channels))

    def forward(self, x):
        identity = x
        if self.downsample is not None:
            identity = self.downsample(x)
        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)
        out = self.conv2(out)
        out = self.bn2(out)
        out += identity
        out = self.relu(out)
        return out

class ECG_Encoder_ResNet(nn.Module):

    def __init__(self, output_dim=256):
        super(ECG_Encoder_ResNet, self).__init__()
        self.conv1 = nn.Conv1d(1, 32, kernel_size=15, stride=2, padding=7, bias=False)
        self.bn1 = nn.BatchNorm1d(32)
        self.relu = nn.ReLU(inplace=True)
        self.maxpool = nn.MaxPool1d(kernel_size=3, stride=2, padding=1)
        self.layer1 = self._make_layer(32, 64, blocks=2, stride=2)
        self.layer2 = self._make_layer(64, 128, blocks=2, stride=2)
        self.layer3 = self._make_layer(128, 256, blocks=2, stride=2)
        self.layer4 = self._make_layer(256, 256, blocks=2, stride=2)
        self.avgpool = nn.AdaptiveAvgPool1d(1)
        self.fc = nn.Linear(256, output_dim)

    def _make_layer(self, in_channels, out_channels, blocks, stride):
        layers = []
        layers.append(ResBlock(in_channels, out_channels, stride))
        for _ in range(1, blocks):
            layers.append(ResBlock(out_channels, out_channels, stride=1))
        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)
        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)
        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.fc(x)
        return x

INIT_ARGS = {
    'output_dim': 64
}

ARGS = {
    'x': torch.randn(2, 1, 512)
}

if __name__ == '__main__':
    model = ECG_Encoder_ResNet(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {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 2, dtype family float, shape (2, 64) — shape (2, 64) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
EnvelopeBranch Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("300163b9306e2fa8")
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, shape (2, 4, 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 torch
import torch.nn as nn

class EnvelopeBranch(nn.Module):

    def __init__(self, stride=2):
        super().__init__()
        self.pool = nn.AvgPool1d(kernel_size=5, stride=stride, padding=2)

    def forward(self, x):
        return self.pool(torch.abs(x))


INIT_ARGS = {
    'stride': 2
}

ARGS = {
    'x': torch.randn(2, 4, 16)
}

if __name__ == '__main__':
    module = EnvelopeBranch(**INIT_ARGS)
    output = module(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
    print(f"Output: {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 3, dtype family float, shape (2, 4, 8) — shape (2, 4, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
HybridEffNetEncoder Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("b1dfbc030b5b1b30")
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
import numpy as np
from collections import OrderedDict

INIT_ARGS = {
    'in_channel': 1,
    'depth': [1, 1],
    'channels': [4, 6, 8],
    'expansion': 2
}

# Create input tensor with shape (batch=2, in_channel=1, feature_dim=16)
input_tensor = torch.randn(2, 1, 16)

ARGS = {
    'x': input_tensor
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 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)
MBConv Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("e94a071c89ce6097")
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, shape (2, 8, 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 torch
import torch.nn as nn
from collections import OrderedDict

class Bottleneck(nn.Module):

    def __init__(self, in_channel, out_channel, expansion, activation, stride=1, padding=1):
        super().__init__()
        self.stride = stride
        hidden_dim = in_channel * expansion
        self.conv1 = nn.Conv1d(in_channel, hidden_dim, kernel_size=1, bias=False)
        self.b0 = nn.BatchNorm1d(hidden_dim)
        self.conv2 = nn.Conv1d(hidden_dim, hidden_dim, kernel_size=3, groups=hidden_dim, padding=padding, stride=stride, bias=False)
        self.b1 = nn.BatchNorm1d(hidden_dim)
        self.conv3 = nn.Conv1d(hidden_dim, out_channel, kernel_size=1, stride=1, bias=False)
        self.b2 = nn.BatchNorm1d(out_channel)
        self.d = nn.Dropout(p=0.1)
        self.act = activation(inplace=True)

    def forward(self, x):
        use_res_connect = self.stride == 1 and x.shape[1] == self.conv3.out_channels
        y = self.act(self.b0(self.conv1(x)))
        y = self.act(self.b1(self.conv2(y)))
        y = self.b2(self.conv3(y))
        y = self.d(y)
        return x + y if use_res_connect else y

class MBConv(nn.Module):

    def __init__(self, in_channel, out_channels, expansion, layers, activation=nn.ReLU6, stride=2):
        super().__init__()
        self.stack = OrderedDict()
        self.stack['s0'] = Bottleneck(in_channel, out_channels, expansion, activation, stride=stride)
        for i in range(1, layers):
            self.stack['s' + str(i)] = Bottleneck(out_channels, out_channels, expansion, activation, stride=1)
        self.stack = nn.Sequential(self.stack)

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


INIT_ARGS = {
    'in_channel': 4,
    'out_channels': 8,
    'expansion': 2,
    'layers': 2,
    'activation': nn.ReLU6,
    'stride': 2,
}

ARGS = {
    'x': torch.randn(2, 4, 16),
}

if __name__ == '__main__':
    model = MBConv(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {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 3, dtype family float, shape (2, 8, 8) — shape (2, 8, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MultiScaleEnvelopeStem Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("55ac875449460c73")
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, shape (2, 16, 32) 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 torch.nn as nn

class DSConv(nn.Module):

    def __init__(self, in_ch, out_ch, kernel_size, stride=1, padding=0, dilation=1):
        super().__init__()
        self.depthwise = nn.Sequential(nn.Conv1d(in_ch, in_ch, kernel_size, stride, padding, dilation=dilation, groups=in_ch, bias=False), nn.BatchNorm1d(in_ch), nn.ReLU6(inplace=True))
        self.pointwise = nn.Sequential(nn.Conv1d(in_ch, out_ch, kernel_size=1, bias=False), nn.BatchNorm1d(out_ch), nn.ReLU6(inplace=True))

    def forward(self, x):
        return self.pointwise(self.depthwise(x))

class EnvelopeBranch(nn.Module):

    def __init__(self, stride=2):
        super().__init__()
        self.pool = nn.AvgPool1d(kernel_size=5, stride=stride, padding=2)

    def forward(self, x):
        return self.pool(torch.abs(x))

class MultiScaleEnvelopeStem(nn.Module):

    def __init__(self, in_channels, out_channels):
        super().__init__()
        assert out_channels % 4 == 0
        branch_c = out_channels // 4
        self.branch_high = DSConv(in_channels, branch_c, kernel_size=3, stride=2, padding=1)
        self.branch_mid = DSConv(in_channels, branch_c, kernel_size=7, stride=2, padding=3)
        self.branch_low = DSConv(in_channels, branch_c, kernel_size=7, stride=2, padding=12, dilation=4)
        self.branch_env_pre = EnvelopeBranch(stride=2)
        self.branch_env_post = nn.Sequential(nn.Conv1d(in_channels, branch_c, 1, bias=False), nn.BatchNorm1d(branch_c), nn.ReLU6(inplace=True))
        self.fusion = nn.Sequential(nn.Conv1d(branch_c * 4, out_channels, kernel_size=1, bias=False), nn.BatchNorm1d(out_channels), nn.ReLU6(inplace=True))

    def forward(self, x):
        h, m, l, e = (self.branch_high(x), self.branch_mid(x), self.branch_low(x), self.branch_env_post(self.branch_env_pre(x)))
        return self.fusion(torch.cat([h, m, l, e], dim=1))

INIT_ARGS = {
    'in_channels': 3,
    'out_channels': 16,
}

ARGS = {
    'x': torch.randn(2, 3, 64),
}

if __name__ == '__main__':
    model = MultiScaleEnvelopeStem(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 16, 16)")
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, shape (2, 16, 32) — shape (2, 16, 32) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ResBlock Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("adb9d7776b9364c5")
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, shape (2, 4, 16) 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 torch.nn as nn

class ResBlock(nn.Module):

    def __init__(self, in_channels, out_channels, stride=1):
        super(ResBlock, self).__init__()
        self.conv1 = nn.Conv1d(in_channels, out_channels, kernel_size=5, stride=stride, padding=2, bias=False)
        self.bn1 = nn.BatchNorm1d(out_channels)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv1d(out_channels, out_channels, kernel_size=5, stride=1, padding=2, bias=False)
        self.bn2 = nn.BatchNorm1d(out_channels)
        self.downsample = None
        if stride != 1 or in_channels != out_channels:
            self.downsample = nn.Sequential(nn.Conv1d(in_channels, out_channels, kernel_size=1, stride=stride, bias=False), nn.BatchNorm1d(out_channels))

    def forward(self, x):
        identity = x
        if self.downsample is not None:
            identity = self.downsample(x)
        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)
        out = self.conv2(out)
        out = self.bn2(out)
        out += identity
        out = self.relu(out)
        return out


INIT_ARGS = {
    'in_channels': 3,
    'out_channels': 4,
    'stride': 1
}

ARGS = {
    'x': torch.randn(2, 3, 16)
}

if __name__ == '__main__':
    model = ResBlock(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 4, 16)")
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, shape (2, 4, 16) — shape (2, 4, 16) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
RespBasicBlock Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("4c83191c74f35d99")
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 for RespBasicBlock.__init__(in_channels, out_channels, stride=1)
INIT_ARGS = {
    'in_channels': 8,
    'out_channels': 12,
    'stride': 2
}

# ARGS for forward(x)
# Input shape: (batch_size, in_channels, sequence_length)
ARGS = {
    'x': torch.randn(2, 8, 16)
}
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)
RespBranchEncoder Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("bb1357a8098ad9d1")
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, shape (2, 256) 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 torch.nn as nn

# Construct the class
class RespBasicBlock(nn.Module):

    def __init__(self, in_channels, out_channels, stride=1):
        super(RespBasicBlock, self).__init__()
        self.resp_conv1 = nn.Conv1d(in_channels, out_channels, kernel_size=3, stride=stride, padding=1, bias=False)
        self.resp_bn1 = nn.BatchNorm1d(out_channels)
        self.resp_gelu = nn.GELU()
        self.resp_conv2 = nn.Conv1d(out_channels, out_channels, kernel_size=3, stride=1, padding=1, bias=False)
        self.resp_bn2 = nn.BatchNorm1d(out_channels)
        self.resp_downsample = None
        if stride != 1 or in_channels != out_channels:
            self.resp_downsample = nn.Sequential(nn.Conv1d(in_channels, out_channels, kernel_size=1, stride=stride, bias=False), nn.BatchNorm1d(out_channels))

    def forward(self, x):
        identity = x
        out = self.resp_conv1(x)
        out = self.resp_bn1(out)
        out = self.resp_gelu(out)
        out = self.resp_conv2(out)
        out = self.resp_bn2(out)
        if self.resp_downsample is not None:
            identity = self.resp_downsample(x)
        out += identity
        out = self.resp_gelu(out)
        return out

class RespBranchEncoder(nn.Module):

    def __init__(self):
        super(RespBranchEncoder, self).__init__()
        self.resp_stem = nn.Sequential(nn.Conv1d(1, 32, kernel_size=7, stride=2, padding=3, bias=False), nn.BatchNorm1d(32), nn.GELU(), nn.MaxPool1d(kernel_size=3, stride=2, padding=1))
        self.resp_layer1 = RespBasicBlock(32, 32, stride=1)
        self.resp_layer2 = RespBasicBlock(32, 64, stride=2)
        self.resp_layer3 = RespBasicBlock(64, 128, stride=2)
        self.resp_layer4 = RespBasicBlock(128, 256, stride=2)
        self.resp_avgpool = nn.AdaptiveAvgPool1d(1)

    def forward(self, x):
        x = self.resp_stem(x)
        x = self.resp_layer1(x)
        x = self.resp_layer2(x)
        x = self.resp_layer3(x)
        x = self.resp_layer4(x)
        x = self.resp_avgpool(x)
        return x.flatten(1)

INIT_ARGS = {}

ARGS = {
    'x': torch.randn(2, 1, 256)
}

if __name__ == '__main__':
    model = RespBranchEncoder(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {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 2, dtype family float, shape (2, 256) — shape (2, 256) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
RespTriStreamModel Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("a2fb964d893f99e7")
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, shape (2, 128) 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 torch.nn as nn

class RespBasicBlock(nn.Module):

    def __init__(self, in_channels, out_channels, stride=1):
        super(RespBasicBlock, self).__init__()
        self.resp_conv1 = nn.Conv1d(in_channels, out_channels, kernel_size=3, stride=stride, padding=1, bias=False)
        self.resp_bn1 = nn.BatchNorm1d(out_channels)
        self.resp_gelu = nn.GELU()
        self.resp_conv2 = nn.Conv1d(out_channels, out_channels, kernel_size=3, stride=1, padding=1, bias=False)
        self.resp_bn2 = nn.BatchNorm1d(out_channels)
        self.resp_downsample = None
        if stride != 1 or in_channels != out_channels:
            self.resp_downsample = nn.Sequential(nn.Conv1d(in_channels, out_channels, kernel_size=1, stride=stride, bias=False), nn.BatchNorm1d(out_channels))

    def forward(self, x):
        identity = x
        out = self.resp_conv1(x)
        out = self.resp_bn1(out)
        out = self.resp_gelu(out)
        out = self.resp_conv2(out)
        out = self.resp_bn2(out)
        if self.resp_downsample is not None:
            identity = self.resp_downsample(x)
        out += identity
        out = self.resp_gelu(out)
        return out

class RespBranchEncoder(nn.Module):

    def __init__(self):
        super(RespBranchEncoder, self).__init__()
        self.resp_stem = nn.Sequential(nn.Conv1d(1, 32, kernel_size=7, stride=2, padding=3, bias=False), nn.BatchNorm1d(32), nn.GELU(), nn.MaxPool1d(kernel_size=3, stride=2, padding=1))
        self.resp_layer1 = RespBasicBlock(32, 32, stride=1)
        self.resp_layer2 = RespBasicBlock(32, 64, stride=2)
        self.resp_layer3 = RespBasicBlock(64, 128, stride=2)
        self.resp_layer4 = RespBasicBlock(128, 256, stride=2)
        self.resp_avgpool = nn.AdaptiveAvgPool1d(1)

    def forward(self, x):
        x = self.resp_stem(x)
        x = self.resp_layer1(x)
        x = self.resp_layer2(x)
        x = self.resp_layer3(x)
        x = self.resp_layer4(x)
        x = self.resp_avgpool(x)
        return x.flatten(1)

class RespTriStreamModel(nn.Module):

    def __init__(self, output_dim=256):
        super(RespTriStreamModel, self).__init__()
        self.resp_flow_encoder = RespBranchEncoder()
        self.resp_thor_encoder = RespBranchEncoder()
        self.resp_abdo_encoder = RespBranchEncoder()
        self.resp_projector = nn.Sequential(nn.Linear(768, 512, bias=False), nn.BatchNorm1d(512), nn.GELU(), nn.Dropout(0.2), nn.Linear(512, output_dim, bias=False))

    def forward(self, x):
        x_flow = x[:, 0:1, :]
        x_thor = x[:, 1:2, :]
        x_abdo = x[:, 2:3, :]
        feat_flow = self.resp_flow_encoder(x_flow)
        feat_thor = self.resp_thor_encoder(x_thor)
        feat_abdo = self.resp_abdo_encoder(x_abdo)
        resp_combined = torch.cat([feat_flow, feat_thor, feat_abdo], dim=1)
        resp_out = self.resp_projector(resp_combined)
        return resp_out

INIT_ARGS = {
    'output_dim': 128
}

ARGS = {
    'x': torch.randn(2, 3, 256)
}
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 2, dtype family float, shape (2, 128) — shape (2, 128) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
RoPETransformerLayer Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("196f0de89b434a95")
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, shape (2, 3, 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 torch
import torch.nn as nn
import torch.nn.functional as F

class RotaryPositionalEmbeddings(nn.Module):

    def __init__(self, d_model, max_seq_len=10000):
        super().__init__()
        inv_freq = 1.0 / 10000 ** (torch.arange(0, d_model, 2).float() / d_model)
        self.register_buffer('inv_freq', inv_freq)

    def forward(self, x, seq_len=None):
        if seq_len is None:
            seq_len = x.shape[1]
        t = torch.arange(seq_len, device=x.device, dtype=self.inv_freq.dtype)
        freqs = torch.einsum('i,j->ij', t, self.inv_freq)
        emb = torch.cat((freqs, freqs), dim=-1)
        return emb[None, :, :]

def apply_rotary_pos_emb(x, cos, sin):

    def rotate_half(x):
        x1, x2 = x.chunk(2, dim=-1)
        return torch.cat((-x2, x1), dim=-1)
    return x * cos + rotate_half(x) * sin

class RoPETransformerLayer(nn.Module):

    def __init__(self, d_model, nhead, dim_feedforward=1024, dropout=0.1):
        super().__init__()
        self.nhead = nhead
        self.head_dim = d_model // nhead
        self.q_proj = nn.Linear(d_model, d_model)
        self.k_proj = nn.Linear(d_model, d_model)
        self.v_proj = nn.Linear(d_model, d_model)
        self.out_proj = nn.Linear(d_model, d_model)
        self.linear1 = nn.Linear(d_model, dim_feedforward)
        self.dropout = nn.Dropout(dropout)
        self.linear2 = nn.Linear(dim_feedforward, d_model)
        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.dropout1 = nn.Dropout(dropout)
        self.dropout2 = nn.Dropout(dropout)
        self.rope = RotaryPositionalEmbeddings(self.head_dim)

    def forward(self, src):
        B, N, C = src.shape
        x = self.norm1(src)
        q = self.q_proj(x).view(B, N, self.nhead, self.head_dim).transpose(1, 2)
        k = self.k_proj(x).view(B, N, self.nhead, self.head_dim).transpose(1, 2)
        v = self.v_proj(x).view(B, N, self.nhead, self.head_dim).transpose(1, 2)
        cos_sin = self.rope(v, seq_len=N)
        cos, sin = (cos_sin.cos(), cos_sin.sin())
        cos, sin = (cos.unsqueeze(1), sin.unsqueeze(1))
        q = apply_rotary_pos_emb(q, cos, sin)
        k = apply_rotary_pos_emb(k, cos, sin)
        attn_output = F.scaled_dot_product_attention(q, k, v)
        attn_output = attn_output.transpose(1, 2).contiguous().view(B, N, C)
        src = src + self.dropout1(self.out_proj(attn_output))
        x = self.norm2(src)
        x = self.linear2(self.dropout(F.gelu(self.linear1(x))))
        src = src + self.dropout2(x)
        return src

INIT_ARGS = {
    'd_model': 8,
    'nhead': 2,
    'dim_feedforward': 16,
    'dropout': 0.1
}

ARGS = {
    'src': torch.randn(2, 3, 8)
}

if __name__ == '__main__':
    model = RoPETransformerLayer(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: {ARGS['src'].shape}")
    assert output.shape == ARGS['src'].shape, "Output shape mismatch"
    print("Success!")
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, shape (2, 3, 8) — shape (2, 3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
RotaryPositionalEmbeddings Ran
ran on a synthesized input
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("3aa1c6232f1025f9")
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, shape (1, 5, 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 torch
import torch.nn as nn

class RotaryPositionalEmbeddings(nn.Module):

    def __init__(self, d_model, max_seq_len=10000):
        super().__init__()
        inv_freq = 1.0 / 10000 ** (torch.arange(0, d_model, 2).float() / d_model)
        self.register_buffer('inv_freq', inv_freq)

    def forward(self, x, seq_len=None):
        if seq_len is None:
            seq_len = x.shape[1]
        t = torch.arange(seq_len, device=x.device, dtype=self.inv_freq.dtype)
        freqs = torch.einsum('i,j->ij', t, self.inv_freq)
        emb = torch.cat((freqs, freqs), dim=-1)
        return emb[None, :, :]


INIT_ARGS = {
    'd_model': 8,
    'max_seq_len': 10000,
}

ARGS = {
    'x': torch.randn(2, 5, 8),
    'seq_len': None,
}

if __name__ == '__main__':
    model = RotaryPositionalEmbeddings(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output:\n{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 3, dtype family float, shape (1, 5, 8) — shape (1, 5, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
apply_rotary_pos_emb Ran
ran; our contract draft was wrong, not the code
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("0f92704bc4a6e1f6")
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 Tensor, ndim 3, dtype family float, shape (2, 3, 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 torch

torch.manual_seed(0)

# Create small tensors for the function
# x: input tensor that will be rotated, shape (batch=2, seq_len=3, dim=4)
x = torch.randn(2, 3, 4)

# cos: cosine positional embedding, same shape as x
cos = torch.randn(2, 3, 4)

# sin: sine positional embedding, same shape as x
sin = torch.randn(2, 3, 4)

ARGS = {
    "x": x,
    "cos": cos,
    "sin": sin,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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, shape (2, 3, 4) — shape (2, 3, 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
Bottleneck Not yet run
unverified (harvested, no recorded run)
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("68f38c0292702117")
How it ran

no invocation recorded.

FiveModalSleepModel Not yet run
unverified (harvested, no recorded run)
AutoBrain-sleep/OmniSleep/src/omnisleep/models.py
code served (permissive licence) · get_code("d2a198a26c4643fe")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 15 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.

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

Sleep physiology arises from the coordinated dynamics of the central nervous system (CNS) and autonomic nervous system (ANS), as reflected by multimodal polysomnography signals including EEG, EOG, EMG, ECG, and respiration. However, existing sleep foundation models often fuse heterogeneous biosignals in a topology-agnostic manner, overlooking their physiological organization. We introduce Omni-Sleep, a sleep foundation model that uses the CNS/ANS partition as a physiological prior for topology-constrained representation learning. Omni-Sleep learns structured representations through three objectives: intra-system consistency, which captures shared subsystem-level factors within neural and cardio-respiratory signals; inter-system synchronization, which aligns subsystem trajectories to model brain-body dynamics; and latent-space masked temporal modeling, which captures longhorizon sleep dynamics. Pre-trained on over 100,000 hours of multi-center multimodal PSG data, Omni-Sleep is evaluated on sleep staging and multi-disease classification. Across datasets and modalityablation settings, Omni-Sleep outperforms strong foundation-model baselines, showing improved label efficiency, cross-dataset generalization, and robustness to missing modalities. These results highlight the value of physiological hierarchy for generalizable sleep representation learning. Code is available at https://github.com/AutoBrain-sleep/OmniSleep.

For agents

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

get_harvested_code_for_paper("2607.07720")
get_code_for_paper("2607.07720")
have("2607.07720")

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

curl https://syntology.ai/api/ran/2607.07720.json

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

[![Syntology run record](https://syntology.ai/api/ran/2607.07720.svg)](https://syntology.ai/paper/2607.07720)

Connect an agent — have() is free.