slip_demo.py

# Modified from github.com/openai/CLIP
from collections import OrderedDict
import numpy as np
import timm
import torch
from torch import nn
from functools import partial
import math
from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
from timm.models.helpers import load_pretrained
from timm.models.layers import DropPath, to_2tuple, trunc_normal_
from timm.models.resnet import resnet26d, resnet50d
from timm.models.registry import register_model
import numpy as np
import sys
import torch.nn.functional as F


class LayerNorm(nn.LayerNorm):
    """Subclass torch's LayerNorm to handle fp16."""

    def forward(self, x: torch.Tensor):
        orig_type = x.dtype
        ret = super().forward(x.type(torch.float32))
        return ret.type(orig_type)


class QuickGELU(nn.Module):
    def forward(self, x: torch.Tensor):
        return x * torch.sigmoid(1.702 * x)


class ResidualAttentionBlock(nn.Module):
    def __init__(self, d_model: int, n_head: int, attn_mask: torch.Tensor = None):
        super().__init__()

        self.attn = nn.MultiheadAttention(d_model, n_head)
        self.ln_1 = LayerNorm(d_model)
        self.mlp = nn.Sequential(OrderedDict([
            ("c_fc", nn.Linear(d_model, d_model * 4)),
            ("gelu", QuickGELU()),
            ("c_proj", nn.Linear(d_model * 4, d_model))
        ]))
        self.ln_2 = LayerNorm(d_model)
        self.attn_mask = attn_mask

    def attention(self, x: torch.Tensor):
        self.attn_mask = self.attn_mask.to(dtype=x.dtype, device=x.device) if self.attn_mask is not None else None
        return self.attn(x, x, x, need_weights=False, attn_mask=self.attn_mask)[0]

    def forward(self, x: torch.Tensor):
        x = x + self.attention(self.ln_1(x))
        x = x + self.mlp(self.ln_2(x))
        return x


class Transformer(nn.Module):
    def __init__(self, width: int, layers: int, heads: int, attn_mask: torch.Tensor = None):
        super().__init__()
        self.width = width
        self.layers = layers
        self.resblocks = nn.Sequential(*[ResidualAttentionBlock(width, heads, attn_mask) for _ in range(layers)])

    def forward(self, x: torch.Tensor):
        return self.resblocks(x)


class CLIP(nn.Module):
    def __init__(self,
                 embed_dim: int,
                 # vision
                 vision_width: int,
                 vision_model: nn.Module,
                 # text
                 context_length: int,
                 vocab_size: int,
                 transformer_width: int,
                 transformer_heads: int,
                 transformer_layers: int,
                 **kwargs,
                 ):
        super().__init__()

        self.context_length = context_length
        self.vision_width = vision_width

        self.visual = vision_model

        self.transformer = Transformer(
            width=transformer_width,
            layers=transformer_layers,
            heads=transformer_heads,
            attn_mask=self.build_attention_mask(),
        )

        self.vocab_size = vocab_size
        self.token_embedding = nn.Embedding(vocab_size, transformer_width)
        self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width))
        self.ln_final = LayerNorm(transformer_width)

        self.image_projection = nn.Parameter(torch.empty(vision_width, embed_dim))
        self.text_projection = nn.Parameter(torch.empty(transformer_width, embed_dim))
        self.logit_scale = nn.Parameter(torch.ones([]) * np.log(1 / 0.07))

        self.initialize_parameters()

    def initialize_parameters(self):
        nn.init.normal_(self.token_embedding.weight, std=0.02)
        nn.init.normal_(self.positional_embedding, std=0.01)

        proj_std = (self.transformer.width ** -0.5) * ((2 * self.transformer.layers) ** -0.5)
        attn_std = self.transformer.width ** -0.5
        fc_std = (2 * self.transformer.width) ** -0.5
        for block in self.transformer.resblocks:
            nn.init.normal_(block.attn.in_proj_weight, std=attn_std)
            nn.init.normal_(block.attn.out_proj.weight, std=proj_std)
            nn.init.normal_(block.mlp.c_fc.weight, std=fc_std)
            nn.init.normal_(block.mlp.c_proj.weight, std=proj_std)

        nn.init.normal_(self.image_projection, std=self.vision_width ** -0.5)
        nn.init.normal_(self.text_projection, std=self.transformer.width ** -0.5)

    def build_attention_mask(self):
        # lazily create causal attention mask, with full attention between the vision tokens
        # pytorch uses additive attention mask; fill with -inf
        mask = torch.empty(self.context_length, self.context_length)
        mask.fill_(float("-inf"))
        mask.triu_(1)  # zero out the lower diagonal
        return mask

    def encode_image(self, image):
        x, attention = self.visual(image)
        x = x @ self.image_projection

        return x, attention

    def encode_text(self, text):
        x = self.token_embedding(text)  # [batch_size, n_ctx, d_model]
        x = x + self.positional_embedding
        x = x.permute(1, 0, 2)  # NLD -> LND
        x = self.transformer(x)
        x = x.permute(1, 0, 2)  # LND -> NLD
        x = self.ln_final(x)

        # x.shape = [batch_size, n_ctx, transformer.width]
        # take features from the eot embedding (eot_token is the highest number in each sequence)
        x = x[torch.arange(x.shape[0]), text.argmax(dim=-1)] @ self.text_projection

        return x

    def forward(self, image, text):
        image_embed = self.encode_image(image)
        text_embed = self.encode_text(text)

        return {'image_embed': image_embed,
                'text_embed': text_embed,
                'logit_scale': self.logit_scale.exp()}


class SIMCLR(nn.Module):
    def __init__(self,
                 # vision
                 vision_width: int,
                 vision_model: nn.Module,
                 # ssl
                 ssl_mlp_dim: int,
                 ssl_emb_dim: int,
                 **kwargs,
                 ):
        super().__init__()

        self.vision_width = vision_width
        self.visual = vision_model

        self.image_mlp = self._build_mlp(in_dim=vision_width, mlp_dim=ssl_mlp_dim, out_dim=ssl_emb_dim)

    def _build_mlp(self, in_dim, mlp_dim, out_dim):
        return nn.Sequential(OrderedDict([
            ("layer1", nn.Linear(in_dim, mlp_dim)),
            ("bn1", nn.SyncBatchNorm(mlp_dim)),
            ("relu1", nn.ReLU(inplace=True)),
            ("layer2", nn.Linear(mlp_dim, mlp_dim)),
            ("bn2", nn.SyncBatchNorm(mlp_dim)),
            ("relu2", nn.ReLU(inplace=True)),
            ("layer3", nn.Linear(mlp_dim, out_dim)),
        ]))

    def encode_image(self, image):
        x = self.visual(image)

        return x

    def forward(self, aug1, aug2):
        h1 = self.visual(aug1)
        h2 = self.visual(aug2)

        aug1_embed = self.image_mlp(h1)
        aug2_embed = self.image_mlp(h2)

        return {'aug1_embed': aug1_embed,
                'aug2_embed': aug2_embed}


class SLIP(CLIP):
    def __init__(self,
                 ssl_mlp_dim=4096,
                 ssl_emb_dim=256,
                 **kwargs,
                 ):
        super().__init__(**kwargs)

        self.image_mlp = self._build_mlp(in_dim=self.vision_width, mlp_dim=ssl_mlp_dim, out_dim=ssl_emb_dim)

    def _build_mlp(self, in_dim, mlp_dim, out_dim):
        return nn.Sequential(OrderedDict([
            ("layer1", nn.Linear(in_dim, mlp_dim)),
            ("bn1", nn.SyncBatchNorm(mlp_dim)),
            ("relu1", nn.ReLU(inplace=True)),
            ("layer2", nn.Linear(mlp_dim, mlp_dim)),
            ("bn2", nn.SyncBatchNorm(mlp_dim)),
            ("relu2", nn.ReLU(inplace=True)),
            ("layer3", nn.Linear(mlp_dim, out_dim)),
        ]))

    def forward(self, image, text, aug1, aug2):
        aug1_embed = self.image_mlp(self.visual(aug1))
        aug2_embed = self.image_mlp(self.visual(aug2))
        
        image_embed = self.encode_image(image)
        text_embed = self.encode_text(text)

        return {'image_embed': image_embed,
                'text_embed': text_embed,
                'logit_scale': self.logit_scale.exp(),
                'aug1_embed': aug1_embed,
                'aug2_embed': aug2_embed}


class Mlp(nn.Module):
    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.,):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.hidden_features = hidden_features
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop)

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        x = self.drop(x)
        return x


class Attention(nn.Module):
    def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0.,):
        super().__init__()
        self.num_heads = num_heads
        self.dim = dim
        head_dim = dim // num_heads
        # NOTE scale factor was wrong in my original version, can set manually to be compat with prev weights
        self.scale = qk_scale or head_dim ** -0.5

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)

    def forward(self, x):
        B, N, C = x.shape

        # qkv
        qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        q, k, v = qkv[0], qkv[1], qkv[2] # batch x head x seq x embed_chunk

        # attention matrix
        attn = (q @ k.transpose(-2, -1)) * self.scale # batch x head x seq x seq'
        attn = attn.softmax(dim=-1)
        attn = self.attn_drop(attn)
        x = (attn @ v) # batch x head x seq x embed_chunk
        x = x.transpose(1, 2) # batch x seq x head x embed_chunk
        x = x.reshape(B, N, C)

        # projection
        x = self.proj(x)
        x = self.proj_drop(x)

        return x, attn.mean(dim=1)


class Block(nn.Module):

    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0.,
                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm):
        super().__init__()
        self.norm1 = norm_layer(dim)
        self.attn = Attention(
            dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop)
        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
        self.norm2 = norm_layer(dim)
        mlp_hidden_dim = int(dim * mlp_ratio)
        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)

    def forward(self, x):
        B,N,C = x.shape

        # MHSA
        x_, attention = self.attn(self.norm1(x))
        x = x + self.drop_path(x_)
        # FFN
        x = x + self.drop_path(self.mlp(self.norm2(x)))

        return x, attention


class PatchEmbed(nn.Module):
    """ Image to Patch Embedding
    """
    def __init__(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768):
        super().__init__()
        img_size = to_2tuple(img_size)
        patch_size = to_2tuple(patch_size)
        num_patches = (img_size[1] // patch_size[1]) * (img_size[0] // patch_size[0])
        self.img_size = img_size
        self.patch_size = patch_size
        self.num_patches = num_patches

        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)

    def forward(self, x):
        B, C, H, W = x.shape
        # FIXME look at relaxing size constraints
        assert H == self.img_size[0] and W == self.img_size[1], \
            f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."
        x = self.proj(x).flatten(2).transpose(1, 2)
        return x


class HybridEmbed(nn.Module):
    """ CNN Feature Map Embedding
    Extract feature map from CNN, flatten, project to embedding dim.
    """
    def __init__(self, backbone, img_size=224, feature_size=None, in_chans=3, embed_dim=768):
        super().__init__()
        assert isinstance(backbone, nn.Module)
        img_size = to_2tuple(img_size)
        self.img_size = img_size
        self.backbone = backbone
        if feature_size is None:
            with torch.no_grad():
                # FIXME this is hacky, but most reliable way of determining the exact dim of the output feature
                # map for all networks, the feature metadata has reliable channel and stride info, but using
                # stride to calc feature dim requires info about padding of each stage that isn't captured.
                training = backbone.training
                if training:
                    backbone.eval()
                o = self.backbone(torch.zeros(1, in_chans, img_size[0], img_size[1]))[-1]
                feature_size = o.shape[-2:]
                feature_dim = o.shape[1]
                backbone.train(training)
        else:
            feature_size = to_2tuple(feature_size)
            feature_dim = self.backbone.feature_info.channels()[-1]
        self.num_patches = feature_size[0] * feature_size[1]
        self.proj = nn.Linear(feature_dim, embed_dim)

    def forward(self, x):
        x = self.backbone(x)[-1]
        x = x.flatten(2).transpose(1, 2)
        x = self.proj(x)
        return x


class VIT(nn.Module):
    """ Vision Transformer with support for patch or hybrid CNN input stage
    """
    def __init__(self, img_size=224, patch_size=16, in_chans=3, num_classes=1000, embed_dim=768, depth=12,
                 num_heads=12, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0.,
                 drop_path_rate=0., hybrid_backbone=None, norm_layer=nn.LayerNorm):
        super().__init__()
        self.num_classes = num_classes
        self.num_features = self.embed_dim = embed_dim  # num_features for consistency with other models

        if hybrid_backbone is not None:
            self.patch_embed = HybridEmbed(
                hybrid_backbone, img_size=img_size, in_chans=in_chans, embed_dim=embed_dim)
        else:
            self.patch_embed = PatchEmbed(
                img_size=img_size, patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim)
        num_patches = self.patch_embed.num_patches

        self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
        self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + 1, embed_dim))
        self.pos_drop = nn.Dropout(p=drop_rate)

        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)]  # stochastic depth decay rule
        self.blocks = nn.ModuleList([
            Block(
                dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
                drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer)
            for i in range(depth)])
        self.norm = norm_layer(embed_dim)

        # NOTE as per official impl, we could have a pre-logits representation dense layer + tanh here
        #self.repr = nn.Linear(embed_dim, representation_size)
        #self.repr_act = nn.Tanh()

        # Classifier head
        self.head = nn.Linear(embed_dim, num_classes) if num_classes > 0 else nn.Identity()

        trunc_normal_(self.pos_embed, std=.02)
        trunc_normal_(self.cls_token, std=.02)
        self.apply(self._init_weights)

        self.depth = depth

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)

    @torch.jit.ignore
    def no_weight_decay(self):
        return {'pos_embed', 'cls_token'}

    def get_classifier(self):
        return self.head

    def reset_classifier(self, num_classes, global_pool=''):
        self.num_classes = num_classes
        self.head = nn.Linear(self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()

    def forward_features(self, x):
        B = x.shape[0]
        x = self.patch_embed(x)

        cls_tokens = self.cls_token.expand(B, -1, -1)  # stole cls_tokens impl from Phil Wang, thanks
        x = torch.cat((cls_tokens, x), dim=1)
        x = x + self.pos_embed
        x = self.pos_drop(x)

        # for blk in self.blocks:
        #     x = blk(x)
        for idx, blk in enumerate(self.blocks):
            if idx < self.depth - 1:
                x, _ = blk(x)
            else:
                x, attention = blk(x)

        x = self.norm(x)
        return x, attention

    def forward(self, x):
        x, attention = self.forward_features(x)
        x = self.head(x)
        return x, attention


def vit_base_patch16_224(**kwargs):
    model = VIT(
        patch_size=16, embed_dim=768, depth=12, num_heads=12, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def vit_large_patch16_224(**kwargs):
    model = VIT(
        patch_size=16, embed_dim=1024, depth=24, num_heads=16, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def build_slip_model(state_dict: dict, key: str):
    if key == 'base':
        vision_model = vit_base_patch16_224(num_classes=0)
        model = SLIP(embed_dim=512, vision_width=768, vision_model=vision_model, context_length=77, vocab_size=49408,
                     transformer_width=512, transformer_heads=8, transformer_layers=12)
        msg = model.load_state_dict(state_dict)
        print('Load SLIP model:', msg)
    elif key == 'large':
        vision_model = vit_large_patch16_224(num_classes=0)
        model = SLIP(embed_dim=512, vision_width=1024, vision_model=vision_model, context_length=77, vocab_size=49408,
                     transformer_width=512, transformer_heads=8, transformer_layers=12)
        msg = model.load_state_dict(state_dict)
        print('Load SLIP model:', msg)
    return model.eval()




# def process_state_dict(state_dict):
#     for k in list(state_dict.keys()):
#         state_dict[k.replace('module.', '')] = state_dict.pop(k)
#     return state_dict
# ckpt = torch.load('./slip_vit_base_16.pth.tar')
# model = build_slip_model(process_state_dict(ckpt['state_dict']), 'base')
# input = torch.randn(16, 3, 224, 224)
# x, attention = model.encode_image(input)
# print(x.shape, attention.shape)