vicgalle
diff --git a/‎README.md
+115-155 b/‎README.md
+115-155
diff --git a/‎aesthetic_embeddings/aivazovsky.pt
2.23 KB b/‎aesthetic_embeddings/aivazovsky.pt
2.23 KB
diff --git a/‎aesthetic_embeddings/cloudcore.pt
2.23 KB b/‎aesthetic_embeddings/cloudcore.pt
2.23 KB
diff --git a/‎aesthetic_embeddings/gloomcore.pt
2.23 KB b/‎aesthetic_embeddings/gloomcore.pt
2.23 KB
diff --git a/‎aesthetic_embeddings/glowwave.pt
2.23 KB b/‎aesthetic_embeddings/glowwave.pt
2.23 KB
diff --git a/‎aesthetic_embeddings/laion_7plus.pt
3.73 KB b/‎aesthetic_embeddings/laion_7plus.pt
3.73 KB
diff --git a/‎aesthetic_embeddings/sac_8plus.pt
3.73 KB b/‎aesthetic_embeddings/sac_8plus.pt
3.73 KB
diff --git a/‎assets/grid-0033.png
1.64 MB b/‎assets/grid-0033.png
1.64 MB
diff --git a/‎assets/grid-0035.png
1.59 MB b/‎assets/grid-0035.png
1.59 MB
diff --git a/‎assets/grid-0043.png
1.9 MB b/‎assets/grid-0043.png
1.9 MB
diff --git a/‎assets/grid-0047.png
1.78 MB b/‎assets/grid-0047.png
1.78 MB
diff --git a/‎assets/grid-0072.png
3.14 MB b/‎assets/grid-0072.png
3.14 MB
diff --git a/‎assets/grid-0073.png
3.29 MB b/‎assets/grid-0073.png
3.29 MB
diff --git a/‎assets/grid-0075.png
3.1 MB b/‎assets/grid-0075.png
3.1 MB
diff --git a/‎assets/grid-0089.png
2.91 MB b/‎assets/grid-0089.png
2.91 MB
diff --git a/‎assets/grid-0091.png
2.93 MB b/‎assets/grid-0091.png
2.93 MB
diff --git a/‎assets/grid-0096.png
2.93 MB b/‎assets/grid-0096.png
2.93 MB
diff --git a/‎assets/grid-0097.png
3.07 MB b/‎assets/grid-0097.png
3.07 MB
diff --git a/‎assets/grid-0131.png
4.43 MB b/‎assets/grid-0131.png
4.43 MB
diff --git a/‎assets/grid-0133.png
4.75 MB b/‎assets/grid-0133.png
4.75 MB
diff --git a/‎assets/grid-0135.png
4.91 MB b/‎assets/grid-0135.png
4.91 MB
diff --git a/‎configs/stable-diffusion/v1-inference-aesthetic.yaml
+74 b/‎configs/stable-diffusion/v1-inference-aesthetic.yaml
+74
@@ -0,0 +1,74 @@
+model:
+  base_learning_rate: 1.0e-04
+  target: ldm.models.diffusion.ddpm.LatentDiffusion
+  params:
+    linear_start: 0.00085
+    linear_end: 0.0120
+    num_timesteps_cond: 1
+    log_every_t: 200
+    timesteps: 1000
+    first_stage_key: "jpg"
+    cond_stage_key: "txt"
+    image_size: 64
+    channels: 4
+    cond_stage_trainable: false # Note: different from the one we trained before
+    conditioning_key: crossattn
+    monitor: val/loss_simple_ema
+    scale_factor: 0.18215
+    use_ema: False
+
+    scheduler_config: # 10000 warmup steps
+      target: ldm.lr_scheduler.LambdaLinearScheduler
+      params:
+        warm_up_steps: [10000]
+        cycle_lengths: [10000000000000] # incredibly large number to prevent corner cases
+        f_start: [1.e-6]
+        f_max: [1.]
+        f_min: [1.]
+
+    unet_config:
+      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
+      params:
+        image_size: 32 # unused
+        in_channels: 4
+        out_channels: 4
+        model_channels: 320
+        attention_resolutions: [4, 2, 1]
+        num_res_blocks: 2
+        channel_mult: [1, 2, 4, 4]
+        num_heads: 8
+        use_spatial_transformer: True
+        transformer_depth: 1
+        context_dim: 768
+        use_checkpoint: True
+        legacy: False
+
+    first_stage_config:
+      target: ldm.models.autoencoder.AutoencoderKL
+      params:
+        embed_dim: 4
+        monitor: val/rec_loss
+        ddconfig:
+          double_z: true
+          z_channels: 4
+          resolution: 256
+          in_channels: 3
+          out_ch: 3
+          ch: 128
+          ch_mult:
+            - 1
+            - 2
+            - 4
+            - 4
+          num_res_blocks: 2
+          attn_resolutions: []
+          dropout: 0.0
+        lossconfig:
+          target: torch.nn.Identity
+
+    cond_stage_config:
+      target: ldm.modules.encoders.modules.PersonalizedCLIPEmbedder
+      params:
+        aesthetic_embedding_path: "aesthetic_embeddings/sac_8plus.pt"
+        T: 3
+        lr: 0.0001