Cloned from madebyollin/taesdxl

README.md

@@ -0,0 +1,31 @@
+---
+license: mit
+---
+
+# 🍰 Tiny AutoEncoder for Stable Diffusion (XL)
+
+[TAESDXL](https://github.com/madebyollin/taesd) is very tiny autoencoder which uses the same "latent API" as [SDXL-VAE](https://huggingface.co/stabilityai/sdxl-vae).
+TAESDXL is useful for [real-time previewing](https://twitter.com/madebyollin/status/1679356448655163394) of the SDXL generation process.
+
+This repo contains `.safetensors` versions of the TAESDXL weights.
+
+For SD1.x / SD2.x, use [TAESD](https://huggingface.co/madebyollin/taesd/) instead (the SD and SDXL VAEs are [incompatible](https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/discussions/6#64b8a9c13707b7d603c6ac16)).
+
+## Using in 🧨 diffusers
+
+```python
+import torch
+from diffusers import DiffusionPipeline, AutoencoderTiny
+
+pipe = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+)
+pipe.vae = AutoencoderTiny.from_pretrained("madebyollin/taesdxl", torch_dtype=torch.float16)
+pipe = pipe.to("cuda")
+
+prompt = "slice of delicious New York-style cheesecake topped with berries, mint, chocolate crumble"
+image = pipe(prompt, num_inference_steps=50, generator=torch.Generator("cpu").manual_seed(0x7A35D)).images[0]
+image.save("cheesecake_sdxl.png")
+```
+
+![image/png](https://cdn-uploads.huggingface.co/production/uploads/630447d40547362a22a969a2/R4aJNPy5lqqbccdnVhSr8.png)

config.json

@@ -0,0 +1,39 @@
+{
+  "_class_name": "AutoencoderTiny",
+  "_diffusers_version": "0.30.0",
+  "act_fn": "relu",
+  "decoder_block_out_channels": [
+    64,
+    64,
+    64,
+    64
+  ],
+  "encoder_block_out_channels": [
+    64,
+    64,
+    64,
+    64
+  ],
+  "force_upcast": false,
+  "in_channels": 3,
+  "latent_channels": 4,
+  "latent_magnitude": 3,
+  "latent_shift": 0.5,
+  "num_decoder_blocks": [
+    3,
+    3,
+    3,
+    1
+  ],
+  "num_encoder_blocks": [
+    1,
+    3,
+    3,
+    3
+  ],
+  "out_channels": 3,
+  "scaling_factor": 1.0,
+  "shift_factor": 0.0,
+  "upsample_fn": "nearest",
+  "upsampling_scaling_factor": 2
+}

diffusion_pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1625815ec7e8c3b63a9fa045fe00be825ed7cdf1691a04f24130a69ac057a70e
+size 9823878

diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff4824aca94dd6111e0340fa749347fb74101060d9712cb5ef1ca8f1cf17502f
+size 9793292

taesdxl_decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6013131e7eb412ef20113f1acc2ea7d3e47e53196ca0530fa65d9b61d814b61
+size 4895612

taesdxl_encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f37c0b28f72ec4ca835dc7dbf05255bdf323cde8cf12a304674f106466c98ef
+size 4895600