clean/video/dfd_fcg/src/clip/model_syno.py

import torch
import random
import pickle
import numpy as np
import torch.nn.functional as F
from torch import nn


from enum import IntEnum, auto, IntFlag
from collections import OrderedDict
from typing import Tuple, Union, List


class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, inplanes, planes, stride=1):
        super().__init__()

        # all conv layers have stride 1. an avgpool is performed after the second convolution when stride > 1
        self.conv1 = nn.Conv2d(inplanes, planes, 1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.relu1 = nn.ReLU(inplace=True)

        self.conv2 = nn.Conv2d(planes, planes, 3, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.relu2 = nn.ReLU(inplace=True)

        self.avgpool = nn.AvgPool2d(stride) if stride > 1 else nn.Identity()

        self.conv3 = nn.Conv2d(planes, planes * self.expansion, 1, bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
        self.relu3 = nn.ReLU(inplace=True)

        self.downsample = None
        self.stride = stride

        if stride > 1 or inplanes != planes * Bottleneck.expansion:
            # downsampling layer is prepended with an avgpool, and the subsequent convolution has stride 1
            self.downsample = nn.Sequential(OrderedDict([
                ("-1", nn.AvgPool2d(stride)),
                ("0", nn.Conv2d(inplanes, planes *
                 self.expansion, 1, stride=1, bias=False)),
                ("1", nn.BatchNorm2d(planes * self.expansion))
            ]))

    def forward(self, x: torch.Tensor):
        identity = x

        out = self.relu1(self.bn1(self.conv1(x)))
        out = self.relu2(self.bn2(self.conv2(out)))
        out = self.avgpool(out)
        out = self.bn3(self.conv3(out))

        if self.downsample is not None:
            identity = self.downsample(x)

        out += identity
        out = self.relu3(out)
        return out


class AttentionPool2d(nn.Module):
    def __init__(self, spacial_dim: int, embed_dim: int, num_heads: int, output_dim: int = None):
        super().__init__()
        self.positional_embedding = nn.Parameter(torch.randn(
            spacial_dim ** 2 + 1, embed_dim) / embed_dim ** 0.5)
        self.k_proj = nn.Linear(embed_dim, embed_dim)
        self.q_proj = nn.Linear(embed_dim, embed_dim)
        self.v_proj = nn.Linear(embed_dim, embed_dim)
        self.c_proj = nn.Linear(embed_dim, output_dim or embed_dim)
        self.num_heads = num_heads

    def forward(self, x):
        x = x.flatten(start_dim=2).permute(2, 0, 1)  # NCHW -> (HW)NC
        x = torch.cat([x.mean(dim=0, keepdim=True), x], dim=0)  # (HW+1)NC
        x = x + self.positional_embedding[:, None, :].to(x.dtype)  # (HW+1)NC
        x, _ = F.multi_head_attention_forward(
            query=x[:1], key=x, value=x,
            embed_dim_to_check=x.shape[-1],
            num_heads=self.num_heads,
            q_proj_weight=self.q_proj.weight,
            k_proj_weight=self.k_proj.weight,
            v_proj_weight=self.v_proj.weight,
            in_proj_weight=None,
            in_proj_bias=torch.cat(
                [self.q_proj.bias, self.k_proj.bias, self.v_proj.bias]),
            bias_k=None,
            bias_v=None,
            add_zero_attn=False,
            dropout_p=0,
            out_proj_weight=self.c_proj.weight,
            out_proj_bias=self.c_proj.bias,
            use_separate_proj_weight=True,
            training=self.training,
            need_weights=False
        )
        return x.squeeze(0)


class ModifiedResNet(nn.Module):
    """
    A ResNet class that is similar to torchvision's but contains the following changes:
    - There are now 3 "stem" convolutions as opposed to 1, with an average pool instead of a max pool.
    - Performs anti-aliasing strided convolutions, where an avgpool is prepended to convolutions with stride > 1
    - The final pooling layer is a QKV attention instead of an average pool
    """

    def __init__(self, layers, output_dim, heads, input_resolution=224, width=64):
        super().__init__()
        self.output_dim = output_dim
        self.input_resolution = input_resolution

        # the 3-layer stem
        self.conv1 = nn.Conv2d(3, width // 2, kernel_size=3,
                               stride=2, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(width // 2)
        self.relu1 = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv2d(width // 2, width // 2,
                               kernel_size=3, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(width // 2)
        self.relu2 = nn.ReLU(inplace=True)
        self.conv3 = nn.Conv2d(
            width // 2, width, kernel_size=3, padding=1, bias=False)
        self.bn3 = nn.BatchNorm2d(width)
        self.relu3 = nn.ReLU(inplace=True)
        self.avgpool = nn.AvgPool2d(2)

        # residual layers
        self._inplanes = width  # this is a *mutable* variable used during construction
        self.layer1 = self._make_layer(width, layers[0])
        self.layer2 = self._make_layer(width * 2, layers[1], stride=2)
        self.layer3 = self._make_layer(width * 4, layers[2], stride=2)
        self.layer4 = self._make_layer(width * 8, layers[3], stride=2)

        embed_dim = width * 32  # the ResNet feature dimension
        self.attnpool = AttentionPool2d(
            input_resolution // 32, embed_dim, heads, output_dim)

    def _make_layer(self, planes, blocks, stride=1):
        layers = [Bottleneck(self._inplanes, planes, stride)]

        self._inplanes = planes * Bottleneck.expansion
        for _ in range(1, blocks):
            layers.append(Bottleneck(self._inplanes, planes))

        return nn.Sequential(*layers)

    def forward(self, x):
        def stem(x):
            x = self.relu1(self.bn1(self.conv1(x)))
            x = self.relu2(self.bn2(self.conv2(x)))
            x = self.relu3(self.bn3(self.conv3(x)))
            x = self.avgpool(x)
            return x

        x = x.type(self.conv1.weight.dtype)
        x = stem(x)
        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)
        x = self.attnpool(x)

        return x


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)


############# ORIGINAL #############
# preserve for text modules
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 MultiheadAttentionAttrExtract(nn.Module):
    '''
    Simple reimplementation of nn.MultiheadAttention with key, value return
    '''

    def __init__(
        self,
        embed_dim,
        n_head,
        attn_record=False
    ):
        super().__init__()

        self.in_proj_weight = nn.Parameter(torch.empty((3 * embed_dim, embed_dim)))
        self.in_proj_bias = nn.Parameter(torch.empty(3 * embed_dim))
        self.out_proj = nn.Linear(embed_dim, embed_dim)

        self.n_head = n_head

        # recordings
        self.attn_record = attn_record
        self.aff = None

    def forward(
        self,
        x: torch.Tensor
    ):
        # x.shape = (batch, frames, grid**2 + 1, width)
        batch, frames = x.shape[:2]

        # Original ViT Self-Attention
        q, k, v = F.linear(
            x,
            self.in_proj_weight,
            self.in_proj_bias
        ).chunk(3, dim=-1)

        view_as = (*q.shape[:3], self.n_head, -1)

        q = q.view(*view_as)
        k = k.view(*view_as)
        v = v.view(*view_as)

        aff = torch.einsum('ntqhc,ntkhc->ntqkh', q / (q.size(-1) ** 0.5), k)

        aff = aff.softmax(dim=-2)
        mix = torch.einsum('ntqlh,ntlhc->ntqhc', aff, v)

        out = self.out_proj(mix.flatten(-2))

        # record attentions
        if self.attn_record:
            self.aff = aff

        return dict(
            q=q,
            k=k,
            v=v,
            out=out
        )


class VResidualAttentionBlock(nn.Module):
    def __init__(
        self,
        d_model: int,
        n_head: int,
        mlp_ratio: int,
        block_index: int,
        attn_record: bool = False,
        store_attrs: List[str] = [],
    ):
        super().__init__()
        # modified
        self.attn = MultiheadAttentionAttrExtract(
            d_model,
            n_head,
            attn_record=attn_record
        )

        self.block_index = block_index
        self.store_attrs = store_attrs

        self.ln_1 = LayerNorm(d_model)
        self.mlp = nn.Sequential(OrderedDict([
            ("c_fc", nn.Linear(d_model, int(d_model * mlp_ratio))),
            ("gelu", QuickGELU()),
            ("c_proj", nn.Linear(int(d_model * mlp_ratio), d_model))
        ]))
        self.ln_2 = LayerNorm(d_model)

        # preserve attrs
        self.attr = {}

    def pop_attr(self):
        ret = self.get_attr()
        self.attr.clear()
        return ret

    def get_attr(self):
        return {k: self.attr[k] for k in self.attr}

    def set_attr(self, **attr):
        self.attr = {
            k: attr[k]
            for k in attr
            if k in self.store_attrs
        }

    def attention(self, x: torch.Tensor):
        return self.attn(x)

    def forward(self, x: torch.Tensor):
        self.pop_attr()
        data = self.attention(self.ln_1(x))

        x = x + data["out"]
        x = x + self.mlp(self.ln_2(x))

        data["emb"] = x
        self.set_attr(**data)

        return data


class VTransformer(nn.Module):
    def __init__(
        self,
        width: int,
        layers: int,
        heads: int,
        mlp_ratio: int,
        num_frames: int,
        attn_record: bool = False,
        store_attrs: List[str] = []
    ):
        super().__init__()
        self.width = width
        self.heads = heads
        self.layers = layers

        self.resblocks = nn.Sequential(*[
            VResidualAttentionBlock(
                d_model=width,
                n_head=heads,
                block_index=i,
                mlp_ratio=mlp_ratio,
                attn_record=attn_record,
                store_attrs=store_attrs
            )
            for i in range(layers)
        ])

    def forward(self, x: torch.Tensor):
        for blk in self.resblocks:
            x = blk(x)["emb"]
        return x


class VisionTransformer(nn.Module):
    def __init__(
        self,
        input_resolution: int,
        patch_size: int,
        width: int,
        layers: int,
        heads: int,
        output_dim: int,
        mlp_ratio: int,
        num_frames: int,
        attn_record: bool = False,
        store_attrs: List[str] = [],
    ):
        super().__init__()
        self.input_resolution = input_resolution
        self.patch_size = patch_size
        self.patch_num = (input_resolution // patch_size) ** 2
        self.output_dim = output_dim

        self.conv1 = nn.Conv2d(
            in_channels=3,
            out_channels=width,
            kernel_size=patch_size,
            stride=patch_size,
            bias=False
        )

        scale = width ** -0.5
        self.class_embedding = nn.Parameter(scale * torch.randn(width))
        self.positional_embedding = nn.Parameter(scale * torch.randn((input_resolution // patch_size) ** 2 + 1, width))
        self.ln_pre = LayerNorm(width)

        self.transformer = VTransformer(
            # structure
            width,
            layers,
            heads,
            mlp_ratio,
            num_frames=num_frames,
            # generic
            attn_record=attn_record,
            store_attrs=store_attrs
        )

        self.ln_post = LayerNorm(width)
        self.proj = nn.Parameter(scale * torch.randn(width, output_dim))

    def _prepare(self, x: torch.Tensor):
        batch, frames = x.shape[:2]
        # x.shape = [batch,  frames, 3, px, px]
        x = self.conv1(x.flatten(0, 1)).unflatten(0, (batch, frames))
        # x.shape = [batch, frames, width, grid, grid]
        x = x.flatten(-2).transpose(-1, -2)
        # x.shape = [batch, frames, grid ** 2, width]
        x = torch.cat(
            [
                self.class_embedding.to(x.dtype) +
                torch.zeros(
                    x.shape[0],
                    x.shape[1],
                    1,
                    x.shape[-1],
                    dtype=x.dtype,
                    device=x.device
                ),
                x
            ],
            dim=-2
        )  # shape = [batch, frames, grid ** 2 + 1, width]
        x = x + self.positional_embedding.to(x.dtype)
        x = self.ln_pre(x)
        return x

    def _transformer(self, x: torch.Tensor):
        x = self.transformer(x)
        return x

    def _finalize(self, x: torch.Tensor):
        x = self.ln_post(x[..., 0, :])

        if self.proj is not None:
            x = x @ self.proj
        return x

    def forward(self, x: torch.Tensor):
        x = self._prepare(x)
        x = self._transformer(x)
        x = self._finalize(x)
        return x


class CLIP(nn.Module):
    def __init__(
        self,
        embed_dim: int,
        # vision
        image_resolution: int,
        vision_layers: Union[Tuple[int, int, int, int], int],
        vision_width: int,
        vision_patch_size: int,
        vision_mlp_ratio: int,
        # text
        context_length: int,
        vocab_size: int,
        transformer_width: int,
        transformer_heads: int,
        transformer_layers: int,
        store_attrs: List[str] = [],
        # video
        num_frames=1,
        **model_kargs
    ):
        super().__init__()

        self.context_length = context_length

        if isinstance(vision_layers, (tuple, list)):
            vision_heads = vision_width * 32 // 64
            self.visual = ModifiedResNet(
                layers=vision_layers,
                output_dim=embed_dim,
                heads=vision_heads,
                input_resolution=image_resolution,
                width=vision_width
            )
        else:
            vision_heads = vision_width // 64
            self.visual = VisionTransformer(
                num_frames=num_frames,
                input_resolution=image_resolution,
                patch_size=vision_patch_size,
                width=vision_width,
                layers=vision_layers,
                heads=vision_heads,
                output_dim=embed_dim,
                mlp_ratio=vision_mlp_ratio,
                store_attrs=store_attrs,
                ** model_kargs
            )

        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.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)

        if isinstance(self.visual, ModifiedResNet):
            if self.visual.attnpool is not None:
                std = self.visual.attnpool.c_proj.in_features ** -0.5
                nn.init.normal_(self.visual.attnpool.q_proj.weight, std=std)
                nn.init.normal_(self.visual.attnpool.k_proj.weight, std=std)
                nn.init.normal_(self.visual.attnpool.v_proj.weight, std=std)
                nn.init.normal_(self.visual.attnpool.c_proj.weight, std=std)

            for resnet_block in [self.visual.layer1, self.visual.layer2, self.visual.layer3, self.visual.layer4]:
                for name, param in resnet_block.named_parameters():
                    if name.endswith("bn3.weight"):
                        nn.init.zeros_(param)

        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)

        if self.text_projection is not None:
            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

    @property
    def dtype(self):
        return self.visual.conv1.weight.dtype

    def encode_frames(self, image):
        return self.visual(image.type(self.dtype))

    def encode_text(self, text):
        x = self.token_embedding(text).type(
            self.dtype)  # [batch_size, n_ctx, d_model]

        x = x + self.positional_embedding.type(self.dtype)
        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).type(self.dtype)

        # 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_features = self.encode_frames(image)
        text_features = self.encode_text(text)

        # normalized features
        image_features = image_features / image_features.norm(dim=-1, keepdim=True)
        text_features = text_features / text_features.norm(dim=-1, keepdim=True)

        # cosine similarity as logits
        logit_scale = self.logit_scale.exp()
        logits_per_image = logit_scale * image_features @ text_features.transpose(-1, -2)
        logits_per_text = logits_per_image.transpose(-1, -2)

        # shape = [global_batch_size, global_batch_size]
        return logits_per_image, logits_per_text


def convert_weights(model: nn.Module):
    """Convert applicable model parameters to fp16"""

    def _convert_weights_to_fp16(l):
        if isinstance(l, (nn.Conv1d, nn.Conv2d, nn.Linear)):
            l.weight.data = l.weight.data.half()
            if l.bias is not None:
                l.bias.data = l.bias.data.half()

        if isinstance(l, nn.MultiheadAttention):
            for attr in [*[f"{s}_proj_weight" for s in ["in", "q", "k", "v"]], "in_proj_bias", "bias_k", "bias_v"]:
                tensor = getattr(l, attr)
                if tensor is not None:
                    tensor.data = tensor.data.half()

        for name in ["text_projection", "proj"]:
            if hasattr(l, name):
                attr = getattr(l, name)
                if attr is not None:
                    attr.data = attr.data.half()

    model.apply(_convert_weights_to_fp16)


def build_model(state_dict: dict, **model_kargs):
    vit = "visual.proj" in state_dict

    if vit:
        vision_width = state_dict["visual.conv1.weight"].shape[0]
        vision_layers = len(
            [
                k for k in state_dict.keys()
                if k.startswith("visual.") and k.endswith(".attn.in_proj_weight")
            ]
        )
        vision_patch_size = state_dict["visual.conv1.weight"].shape[-1]
        vision_mlp_ratio = (
            state_dict["visual.transformer.resblocks.0.mlp.c_fc.weight"].shape[0] /
            state_dict["visual.transformer.resblocks.0.mlp.c_fc.weight"].shape[1]
        )
        grid_size = round(
            (state_dict["visual.positional_embedding"].shape[0] - 1) ** 0.5)
        image_resolution = vision_patch_size * grid_size
    else:
        counts: list = [
            len(
                set(k.split(".")[2]
                    for k in state_dict
                    if k.startswith(f"visual.layer{b}"))
            )
            for b in [1, 2, 3, 4]
        ]
        vision_layers = tuple(counts)
        vision_width = state_dict["visual.layer1.0.conv1.weight"].shape[0]
        output_width = round(
            (state_dict["visual.attnpool.positional_embedding"].shape[0] - 1) ** 0.5)
        vision_patch_size = None
        vision_mlp_ratio = None
        assert output_width ** 2 + \
            1 == state_dict["visual.attnpool.positional_embedding"].shape[0]
        image_resolution = output_width * 32

    embed_dim = state_dict["text_projection"].shape[1]
    context_length = state_dict["positional_embedding"].shape[0]
    vocab_size = state_dict["token_embedding.weight"].shape[0]
    transformer_width = state_dict["ln_final.weight"].shape[0]
    transformer_heads = transformer_width // 64
    transformer_layers = len(set(k.split(".")[2] for k in state_dict if k.startswith("transformer.resblocks")))

    model = CLIP(
        embed_dim,
        image_resolution, vision_layers, vision_width, vision_patch_size, vision_mlp_ratio,
        context_length, vocab_size, transformer_width, transformer_heads, transformer_layers,
        **model_kargs
    )

    for key in ["input_resolution", "context_length", "vocab_size"]:
        if key in state_dict:
            del state_dict[key]

    convert_weights(model)
    model.load_state_dict(state_dict, strict=False)
    return model.eval()