foscat.healpix_vit_skip

Attributes

in_nside

Classes

MLP	Base class for all neural network modules.
HealpixViTSkip	Base class for all neural network modules.

Module Contents

class foscat.healpix_vit_skip.MLP(d: int, hidden_mult: int = 4, drop: float = 0.0)

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F


class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

net

forward(x)

class foscat.healpix_vit_skip.HealpixViTSkip(*, in_nside: int, n_chan_in: int, level_dims: List[int], depth_token: int, num_heads_token: int, cell_ids: numpy.ndarray, task: Literal['regression', 'segmentation', 'global'] = 'regression', out_channels: int = 1, mlp_ratio_token: float = 4.0, KERNELSZ: int = 3, gauge_type: Literal['cosmo', 'phi'] = 'cosmo', G: int = 1, prefer_foscat_gpu: bool = True, dropout: float = 0.1, dtype: Literal['float32', 'float64'] = 'float32', pos_embed_per_level: bool = True)

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F


class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

in_nside

n_chan_in

level_dims

token_down

C_fine

embed_dim

depth_token

num_heads_token

mlp_ratio_token

task = 'regression'

out_channels = 1

KERNELSZ = 3

gauge_type = 'cosmo'

G = 1

prefer_foscat_gpu = True

dropout

dtype = 'float32'

pos_embed_per_level = True

cell_ids_fine

f

hconv_levels: List[foscat.SphericalStencil.SphericalStencil] = []

level_cell_ids: List[numpy.ndarray]

token_nside

token_cell_ids

hconv_token

hconv_head

nsides_levels

ntokens_levels

pe_w1

pe_w2

pe_bn1

pe_bn2

enc_w1

enc_w2

enc_bn1

enc_bn2

temporal_encoders

n_tokens

pos_token

encoder_token

dec_q

dec_k

dec_v

dec_attn

dec_mlp

level_pos

dec_refine_w1

dec_refine_w2

dec_refine_bn1

dec_refine_bn2

runtime_device

forward(x: torch.Tensor, runtime_ids: numpy.ndarray | None = None) → torch.Tensor

foscat.healpix_vit_skip.in_nside = 4