vae_decoder_native

Native VAE Decoder

Description

Native R torch implementation of the SDXL VAE decoder. Replaces TorchScript decoder for better GPU compatibility.

Usage

vae_decoder_native(latent_channels = 4, out_channels = 3,
                   block_channels = c(512, 512, 256, 128), norm_groups = 32)

Arguments

• latent_channels: Number of latent channels (4 for SD/SDXL, 16 for FLUX/SD3)
• out_channels: Number of output channels (default 3 for RGB)
• block_channels: Decoder block channels (reversed encoder block_out_channels; default matches SD/SDXL and FLUX)
• norm_groups: Group norm groups (default 32; must divide every entry of block_channels)

Value

An nn_module representing the VAE decoder

Examples

decoder <- vae_decoder_native()
load_decoder_weights(decoder, "path/to/decoder.pt")
latents <- torch::torch_randn(c(1, 4, 64, 64))
image <- decoder(latents)

• ← vocab_size
• unigram_tokenizer →