Upload PixArtSigmaPipeline

README.md

@@ -1 +1,4 @@
+---
+library_name: diffusers
+---
 This is still not trained, this is an attempt at sticking a 16 channels VAE to a Pixart-Sigma pipeline.

transformer/config.json

@@ -1,14 +1,16 @@
 {
   "_class_name": "PixArtTransformer2DModel",
   "_diffusers_version": "0.30.2",
+  "_name_or_path": "frutiemax/TwistedReality-pixart-512ms",
   "activation_fn": "gelu-approximate",
   "attention_bias": true,
   "attention_head_dim": 72,
   "attention_type": "default",
   "caption_channels": 4096,
   "cross_attention_dim": 1152,
+  "double_self_attention": false,
   "dropout": 0.0,
-  "in_channels": 16,
+  "in_channels": 4,
   "interpolation_scale": 1,
   "norm_elementwise_affine": false,
   "norm_eps": 1e-06,
@@ -17,9 +19,12 @@
   "num_attention_heads": 16,
   "num_embeds_ada_norm": 1000,
   "num_layers": 28,
-  "out_channels": 32,
+  "num_vector_embeds": null,
+  "only_cross_attention": false,
+  "out_channels": 8,
   "patch_size": 2,
   "sample_size": 64,
   "upcast_attention": false,
-  "use_additional_conditions": false
+  "use_additional_conditions": null,
+  "use_linear_projection": false
 }

transformer/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d4a16d9e4b6c2b1c89a204e1435d3bddfe05d93fc7bca9dd6d29a408331bc6b9
-size 2444156432
+oid sha256:a92c9ace52b60fe1da0c5997859c8ecc6ad1da55a68d5519110c89b66b59ee61
+size 2443492488

vae/config.json

@@ -1,12 +1,12 @@
 {
   "_class_name": "AutoencoderKL",
   "_diffusers_version": "0.30.2",
-  "_name_or_path": "AuraDiffusion/16ch-vae",
+  "_name_or_path": "ostris/vae-kl-f8-d16",
   "act_fn": "silu",
   "block_out_channels": [
+    128,
     128,
     256,
-    512,
     512
   ],
   "down_block_types": [
@@ -15,18 +15,18 @@
     "DownEncoderBlock2D",
     "DownEncoderBlock2D"
   ],
-  "force_upcast": false,
+  "force_upcast": true,
   "in_channels": 3,
   "latent_channels": 16,
   "latents_mean": null,
   "latents_std": null,
   "layers_per_block": 2,
-  "mid_block_add_attention": false,
+  "mid_block_add_attention": true,
   "norm_num_groups": 32,
   "out_channels": 3,
-  "sample_size": 1024,
-  "scaling_factor": 0.18215,
-  "shift_factor": 0,
+  "sample_size": 512,
+  "scaling_factor": 0.2614,
+  "shift_factor": null,
   "up_block_types": [
     "UpDecoderBlock2D",
     "UpDecoderBlock2D",

vae/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ead97b12214d10d347a57425f2387242710c8e101e09466bcf91bc0f28d65e9c
-size 326896452
+oid sha256:b6468f242218bc72981284b6fda570cc7c07ce043e84979cc1bb7d99d39a64a2
+size 229094420