bi BddlmZddlmZmZmZddlZddlmZddl m Z m Z ddl m Z ddlmZmZmZdd lmZd d lmZmZmZeGd d e ZGddee Zy)) dataclass)OptionalTupleUnionN) ConfigMixinregister_to_config) BaseOutput)GaussianFourierProjectionTimestepEmbedding Timesteps) ModelMixin)UNetMidBlock2Dget_down_block get_up_blockc0eZdZUdZej ed<y) UNet2DOutputz The output of [`UNet2DModel`]. Args: sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)`): The hidden states output from the last layer of the model. sampleN)__name__ __module__ __qualname____doc__torchTensor__annotations__Y/home/cdr/jupyterlab/.venv/lib/python3.12/site-packages/diffusers/models/unets/unet_2d.pyrrs LLrrc;eZdZdZdZdgZe d(deee e e e ffde de de de d ee d e d e d e e d fdee de e d fde e d fde de de de de de de dee de dee de de de dee dee d ee f8fd! Z d)d"ej d#eej e e fd$eej d%e d&eee ff d'ZxZS)* UNet2DModela A 2D UNet model that takes a noisy sample and a timestep and returns a sample shaped output. This model inherits from [`ModelMixin`]. Check the superclass documentation for it's generic methods implemented for all models (such as downloading or saving). Parameters: sample_size (`int` or `Tuple[int, int]`, *optional*, defaults to `None`): Height and width of input/output sample. Dimensions must be a multiple of `2 ** (len(block_out_channels) - 1)`. in_channels (`int`, *optional*, defaults to 3): Number of channels in the input sample. out_channels (`int`, *optional*, defaults to 3): Number of channels in the output. center_input_sample (`bool`, *optional*, defaults to `False`): Whether to center the input sample. time_embedding_type (`str`, *optional*, defaults to `"positional"`): Type of time embedding to use. freq_shift (`int`, *optional*, defaults to 0): Frequency shift for Fourier time embedding. flip_sin_to_cos (`bool`, *optional*, defaults to `True`): Whether to flip sin to cos for Fourier time embedding. down_block_types (`Tuple[str]`, *optional*, defaults to `("DownBlock2D", "AttnDownBlock2D", "AttnDownBlock2D", "AttnDownBlock2D")`): Tuple of downsample block types. mid_block_type (`str`, *optional*, defaults to `"UNetMidBlock2D"`): Block type for middle of UNet, it can be either `UNetMidBlock2D` or `None`. up_block_types (`Tuple[str]`, *optional*, defaults to `("AttnUpBlock2D", "AttnUpBlock2D", "AttnUpBlock2D", "UpBlock2D")`): Tuple of upsample block types. block_out_channels (`Tuple[int]`, *optional*, defaults to `(224, 448, 672, 896)`): Tuple of block output channels. layers_per_block (`int`, *optional*, defaults to `2`): The number of layers per block. mid_block_scale_factor (`float`, *optional*, defaults to `1`): The scale factor for the mid block. downsample_padding (`int`, *optional*, defaults to `1`): The padding for the downsample convolution. downsample_type (`str`, *optional*, defaults to `conv`): The downsample type for downsampling layers. Choose between "conv" and "resnet" upsample_type (`str`, *optional*, defaults to `conv`): The upsample type for upsampling layers. Choose between "conv" and "resnet" dropout (`float`, *optional*, defaults to 0.0): The dropout probability to use. act_fn (`str`, *optional*, defaults to `"silu"`): The activation function to use. attention_head_dim (`int`, *optional*, defaults to `8`): The attention head dimension. norm_num_groups (`int`, *optional*, defaults to `32`): The number of groups for normalization. attn_norm_num_groups (`int`, *optional*, defaults to `None`): If set to an integer, a group norm layer will be created in the mid block's [`Attention`] layer with the given number of groups. If left as `None`, the group norm layer will only be created if `resnet_time_scale_shift` is set to `default`, and if created will have `norm_num_groups` groups. norm_eps (`float`, *optional*, defaults to `1e-5`): The epsilon for normalization. resnet_time_scale_shift (`str`, *optional*, defaults to `"default"`): Time scale shift config for ResNet blocks (see [`~models.resnet.ResnetBlock2D`]). Choose from `default` or `scale_shift`. class_embed_type (`str`, *optional*, defaults to `None`): The type of class embedding to use which is ultimately summed with the time embeddings. Choose from `None`, `"timestep"`, or `"identity"`. num_class_embeds (`int`, *optional*, defaults to `None`): Input dimension of the learnable embedding matrix to be projected to `time_embed_dim` when performing class conditioning with `class_embed_type` equal to `None`. Tnorm sample_size in_channels out_channelscenter_input_sampletime_embedding_typetime_embedding_dim freq_shiftflip_sin_to_cosdown_block_types.mid_block_typeup_block_typesblock_out_channelslayers_per_blockmid_block_scale_factordownsample_paddingdownsample_type upsample_typedropoutact_fnattention_head_dimnorm_num_groupsattn_norm_num_groupsnorm_epsresnet_time_scale_shift add_attentionclass_embed_typenum_class_embedsnum_train_timestepsct*|||_|xs| ddz}t| t| k7rt d| d| dt| t| k7rt d| d| dt j || ddd |_|d k(rt| dd |_ d| dz}nH|dk(rt| d|||_ | d}n(|dk(r#t j|| d|_ | d}t||_ ||t j|||_n?|dk(rt|||_n(|dk(rt j|||_nd|_t j g|_d|_t j g|_| d}t)| D]V\} }!|}"| | }| t| dz k(}#t+|!| |"|||# |||||n|||||}$|j"j-|$X| d|_n%t/| d||||||||n| d||| |_t1t3| }%|%d}t)| D]w\} }&|}'|%| }|%t5| dzt| dz }"| t| dz k(}#t7|&| dz|"||'||# |||||n||||}(|j&j-|(y||nt5| ddzd})t j8| d|)||_t j<|_t j | d|dd |_ y)Nrz\Must provide the same number of `down_block_types` as `up_block_types`. `down_block_types`: z. `up_block_types`: .zbMust provide the same number of `block_out_channels` as `down_block_types`. `block_out_channels`: z. `down_block_types`: r)rr) kernel_sizepaddingfourier)embedding_sizescaler positionallearnedtimestepidentityr) num_layersr%r& temb_channelsadd_downsample resnet_eps resnet_act_fn resnet_groupsr7r2r;r3r5) r%rNr5rPrQoutput_scale_factorr;r7rR attn_groupsr<) rMr%r&prev_output_channelrN add_upsamplerPrQrRr7r;r4r5 ) num_channels num_groupseps)!super__init__r$len ValueErrornnConv2dconv_inr time_projr Embeddingr time_embeddingclass_embeddingIdentity ModuleList down_blocks mid_block up_blocks enumeraterappendrlistreversedminr GroupNorm conv_norm_outSiLUconv_actconv_out)+selfr$r%r&r'r(r)r*r+r,r-r.r/r0r1r2r3r4r5r6r7r8r9r:r;r<r=r>r?time_embed_dimtimestep_input_dimoutput_channelidown_block_type input_channelis_final_block down_blockreversed_block_out_channels up_block_typerVup_blocknum_groups_out __class__s+ r r]zUNet2DModel.__init___sA@ &+H/A!/Dq/H  C$7 7nooAAUVdUeefg  ! "c*:&; ;tvHuII_`p_qqrs  yy.@.CQR\bc  ) +6FXYZF[cefDN!"%7%:!:  L 0&'9!'> %^^FC0F "nn BN0#n6L6L2M1M1OPK%;CYCY?Z>Z%[ "~{3&4V[#{&[# ' SA B##F+v&v&  " k !F ;; * *i 7!))6<<?*[qcC UVUWHXDY>Y*[\Ii'F9 6**r)NrrFrINrT) DownBlock2DAttnDownBlock2Drrr) AttnUpBlock2Drr UpBlock2D)iiir rrconvrgsilurXNgh㈵>defaultTNNN)NT)rrrr _supports_gradient_checkpointing _skip_layerwise_casting_patternsr rrintrboolstrfloatr]rrrr __classcell__)rs@r r"r"(s1f(,$(.x$>B$)#/,0 $,t(8*j.B !()"#%#,-!.2'0"*.*.-1;XaeCsCx$89:XaXa Xa " Xa ! Xa%SMXaXaXa S/Xa! Xac3hXa"#s(OXaXa!&Xa  !Xa"#Xa$%Xa&'Xa()Xa*%SM+Xa,-Xa.'sm/Xa01Xa2"%3Xa45Xa6#3-7Xa8#3-9Xa:&c];XaXa|04 h+ h+ eS01h+u||, h+  h+ |U" # h+rr") dataclassesrtypingrrrrtorch.nnr`configuration_utilsrr utilsr embeddingsr r rmodeling_utilsrunet_2d_blocksrrrrr"rrr rsV")) BPP'HH  :   z+*kz+r