Source code for geowatch.tasks.fusion.architectures.wu_mae

"""
A liberated of WU's MAE pretrained backbone

import sys
sys.path.append('/data/joncrall/dvc-repos/smart_expt_dvc/models/wu/MAE-2023-02-09')
import pred_features

import liberator
lib = liberator.Liberator()
lib.add_dynamic(pred_features.ViT)
lib.expand(['pred_features'])
print(lib.current_sourcecode())
"""
from einops.layers.torch import Rearrange
import torch.nn as nn
import torch
from einops import rearrange
from einops import repeat




class PreNorm(nn.Module):
    def __init__(self, dim, fn):
        super().__init__()
        self.norm = nn.LayerNorm(dim)
        self.fn = fn



    def forward(self, x, **kwargs):
        return self.fn(self.norm(x), **kwargs)






class FeedForward(nn.Module):
    def __init__(self, dim, hidden_dim, dropout=0.):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(dim, hidden_dim),
            nn.GELU(),
            nn.Dropout(dropout),
            nn.Linear(hidden_dim, dim),
            nn.Dropout(dropout)
        )



    def forward(self, x, mask):
        return self.net(x)






class Attention(nn.Module):
    def __init__(self, dim, heads=8, dim_head=64, dropout=0.):
        super().__init__()
        inner_dim = dim_head * heads
        project_out = not (heads == 1 and dim_head == dim)

        self.heads = heads
        self.scale = dim_head ** -0.5

        self.attend = nn.Softmax(dim=-1)
        self.dropout = nn.Dropout(dropout)

        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias=False)

        self.to_out = nn.Sequential(
            nn.Linear(inner_dim, dim),
            nn.Dropout(dropout)
        ) if project_out else nn.Identity()



    def forward(self, x, mask):
        qkv = self.to_qkv(x).chunk(3, dim=-1)
        q, k, v = map(
            lambda t: rearrange(
                t, 'b n (h d) -> b h n d', h=self.heads), qkv)

        #dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale
        dots = torch.einsum('ijkd,ijld->ijkl', q, k) * self.scale
        attn = torch.einsum('ijkl,ijkl->ijkl',
                            dots,
                            repeat(mask,
                                   'b n -> b h n d',
                                   h=self.heads,
                                   d=dots.shape[-1]))
        attn = self.attend(dots)
        attn = self.dropout(attn)

        out = torch.matmul(attn, v)
        out = rearrange(out, 'b h n d -> b n (h d)')
        return self.to_out(out)






def pair(t):
    return t if isinstance(t, tuple) else (t, t)





class Transformer(nn.Module):
    """
    Example:
        >>> dim = 4
        >>> depth = 3
        >>> heads = 2
        >>> dim_head = 2
        >>> mlp_dim = 2
        >>> self = Transformer(dim, depth, heads, dim_head, mlp_dim)
        >>> x = torch.rand(2, 3, dim)
        >>> mask = torch.rand(2, 3) > 0
        >>> out = self.forward(x, mask)
        >>> print(f'out.shape={out.shape}')
    """

    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout=0.):
        super().__init__()
        self.layers = nn.ModuleList([])
        for _ in range(depth):
            self.layers.append(nn.ModuleList([
                PreNorm(
                    dim,
                    Attention(
                        dim,
                        heads=heads,
                        dim_head=dim_head,
                        dropout=dropout)),
                PreNorm(dim, FeedForward(dim, mlp_dim, dropout=dropout))
            ]))



    def forward(self, x, mask):
        for attn, ff in self.layers:
            x = attn(x, mask=mask) + x
            x = ff(x, mask=mask) + x
        return x






class ViT(nn.Module):
    def __init__(self, *, image_size, image_patch_size, frames, frame_patch_size, dim,
                 depth, heads, mlp_dim, channels=6, dim_head=64, dropout=0., emb_dropout=0.):
        super().__init__()
        image_height, image_width = pair(image_size)
        patch_height, patch_width = pair(image_patch_size)

        assert image_height % patch_height == 0 and image_width % patch_width == 0, 'Image dimensions must be divisible by the patch size.'
        assert frames % frame_patch_size == 0, 'Frames must be divisible by frame patch size'

        num_patches = (image_height // patch_height) * \
            (image_width // patch_width) * (frames // frame_patch_size)
        patch_dim = channels * patch_height * patch_width * frame_patch_size

        #assert pool in {'cls', 'mean'}, 'pool type must be either cls (cls token) or mean (mean pooling)'

        self.to_patch_embedding = nn.Sequential(
            Rearrange(
                'b (f pf) c (h p1) (w p2) -> b (f h w) (p1 p2 pf c)',
                p1=patch_height,
                p2=patch_width,
                pf=frame_patch_size),
            nn.Linear(patch_dim, dim),
        )

        self.pos_embedding = nn.Parameter(torch.randn(1, num_patches + 1, dim))
        #self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
        self.dropout = nn.Dropout(emb_dropout)

        self.transformer = Transformer(
            dim, depth, heads, dim_head, mlp_dim, dropout)

        #self.pool = pool
        #self.to_latent = nn.Identity()

        # self.mlp_head = nn.Sequential(
        # nn.LayerNorm(dim),
        #nn.Linear(dim, num_classes)
        # )



    def forward(self, video):
        x = self.to_patch_embedding(video)
        b, n, _ = x.shape

        #cls_tokens = repeat(self.cls_token, '1 1 d -> b 1 d', b = b)
        #x = torch.cat((cls_tokens, x), dim=1)
        x += self.pos_embedding[:, :(n + 1)]
        x = self.dropout(x)

        x = self.transformer(x)

        #x = x.mean(dim = 1) if self.pool == 'mean' else x[:, 0]
        #x = self.to_latent(x)
        return x






def wu_backbone():
    """
    from torch_liberator import Pretrained
    ckpt_fpath = '/home/joncrall/remote/toothbrush/data/dvc-repos/smart_expt_dvc/models/wu/MAE-2023-02-09/goldenMae-epoch=07-val_loss=0.23.ckpt'
    initializer = Pretrained(ckpt_fpath, association='embedding')
    vit = wu_backbone()
    result = initializer.forward(vit)
    """
    vit = ViT(
        image_size=128,
        image_patch_size=4,
        frames=4,
        frame_patch_size=2,
        dim=16,
        depth=12,
        heads=12,
        mlp_dim=1024,
        dropout=0.1
    )
    return vit