modeling_vitmix.py

from transformers import PreTrainedModel
from .configuration_vitmix import ViTMixConfig


# Model architecture gracefully stolen from lucidrains https://github.com/lucidrains/vit-pytorch/blob/main/vit_pytorch/simple_vit.py
import torch
from torch import nn

from einops import rearrange
from einops.layers.torch import Rearrange

from st_moe_pytorch import SparseMoEBlock, MoE
# I thin this is 'including a copy of this notice'... tell me if I'm wrong
####################################
#MIT License

#Copyright (c) 2020 Phil Wang

#Permission is hereby granted, free of charge, to any person obtaining a copy
#of this software and associated documentation files (the "Software"), to deal
#in the Software without restriction, including without limitation the rights
#to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
#copies of the Software, and to permit persons to whom the Software is
#furnished to do so, subject to the following conditions:

#The above copyright notice and this permission notice shall be included in all
#copies or substantial portions of the Software.

#THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
#IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
#FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
#AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
#LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
#OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
#SOFTWARE.
##################################]
# helpers

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

def posemb_sincos_2d(h, w, dim, temperature: int = 10000, dtype = torch.float32):
    y, x = torch.meshgrid(torch.arange(h), torch.arange(w), indexing="ij")
    assert (dim % 4) == 0, "feature dimension must be multiple of 4 for sincos emb"
    omega = torch.arange(dim // 4) / (dim // 4 - 1)
    omega = 1.0 / (temperature ** omega)

    y = y.flatten()[:, None] * omega[None, :]
    x = x.flatten()[:, None] * omega[None, :]
    pe = torch.cat((x.sin(), x.cos(), y.sin(), y.cos()), dim=1)
    return pe.type(dtype)

# classes

class FeedForward(nn.Module):
    def __init__(self, dim, hidden_dim):
        super().__init__()
        self.net = nn.Sequential(
            nn.LayerNorm(dim),
            nn.Linear(dim, hidden_dim),
            nn.GELU(),
            nn.Linear(hidden_dim, dim),
        )
    def forward(self, x):
        return self.net(x)

class Attention(nn.Module):
    def __init__(self, dim, heads = 8, dim_head = 64):
        super().__init__()
        inner_dim = dim_head *  heads
        self.heads = heads
        self.scale = dim_head ** -0.5
        self.norm = nn.LayerNorm(dim)

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

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

    def forward(self, x):
        x = self.norm(x)

        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

        attn = self.attend(dots)

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

class Transformer(nn.Module):
    ### Here is my MOE modification
    def __init__(self, dim, depth, heads, dim_head, mlp_dim, num_experts):
        super().__init__()
        self.norm = nn.LayerNorm(dim)
        self.layers = nn.ModuleList([])
        for _ in range(depth):
            if _ % 2 == 0: #make every other FNN an expert layer.
                self.layers.append(nn.ModuleList([
                    Attention(dim, heads = heads, dim_head = dim_head),
                    FeedForward(dim, mlp_dim)
                ]))
            else:
                self.layers.append(nn.ModuleList([
                    Attention(dim, heads = heads, dim_head = dim_head),
                    SparseMoEBlock(
                        MoE(dim = dim,
                            num_experts = num_experts,
                            gating_top_n = 2,
                            threshold_train = 0.2,
                            threshold_eval = 0.2,
                            capacity_factor_train = 1.25,
                            capacity_factor_eval = 2.,
                            balance_loss_coef = 1e-2,
                            router_z_loss_coef = 1e-3,
                        ),
                        add_ff_before = True,
                        add_ff_after = True
                    )
                ]))
    def forward(self, x):
        for attne, ff in self.layers:
            x = attne(x) + x
            try:
                x = ff(x) + x
            except:
                x = ff(x)[0]+x # I won't bother returning aux_loss... probably a bad idea
        return self.norm(x)

class SimpleViTMIX(nn.Module):
    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, channels = 3, dim_head = 64, num_experts = 12):
        super().__init__()
        image_height, image_width = pair(image_size)
        patch_height, patch_width = pair(patch_size)

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

        patch_dim = channels * patch_height * patch_width

        self.to_patch_embedding = nn.Sequential(
            Rearrange("b c (h p1) (w p2) -> b (h w) (p1 p2 c)", p1 = patch_height, p2 = patch_width),
            nn.LayerNorm(patch_dim),
            nn.Linear(patch_dim, dim),
            nn.LayerNorm(dim),
        )

        self.pos_embedding = posemb_sincos_2d(
            h = image_height // patch_height,
            w = image_width // patch_width,
            dim = dim,
        )

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

        self.pool = "mean"
        self.to_latent = nn.Identity()

        self.linear_head = nn.Linear(dim, num_classes)

    def forward(self, img):
        device = img.device

        x = self.to_patch_embedding(img)
        x += self.pos_embedding.to(device, dtype=x.dtype)

        x = self.transformer(x)
        x = x.mean(dim = 1)

        x = self.to_latent(x)
        return self.linear_head(x)



class ViTMixModel(PreTrainedModel):
    config_class = ViTMixConfig
    def __init__(self, config):
        super().__init__(config)
        self.model = SimpleViTMIX(
                image_size = config.image_size,
                patch_size = config.patch_size,
                num_classes = config.num_classes,
                dim = config.dim,
                depth = config.depth,
                heads = config.heads,
                mlp_dim = config.mlp_dim,
                num_experts = config.num_experts
            )
    def forward(self,tensor, labels = None):
        logits = self.model(tensor)
        if labels is not None:
            loss = torch.nn.cross_entropy(logits, labels)
            return {"loss": loss, "logits": logits}
        return {"logits": logits}