Setting

README.md

@@ -1,8 +1,8 @@
 ---
-datasets:
-- ylecun/mnist
 language:
 - en
 library_name: diffusers
+datasets:
+- ylecun/mnist
 pipeline_tag: image-to-image
----
+---

logs/train_example/{events.out.tfevents.1717883137.federatedlearning.3604610.0 → events.out.tfevents.1717957992.federatedlearning.3567175.0}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7059d895d52183be4c5673846a54856ab4b8b4c8e4970b5e7832e97d15fd691
-size 118824
+oid sha256:e74fd609038695d7ce5858ced8689f9ba7e04c5a4f8823a5d98fd5f135e72a7d
+size 595328

model_index.json

@@ -6,7 +6,7 @@
     "DDPMScheduler"
   ],
   "unet": [
-    null,
-    null
+    "diffusers",
+    "UNet2DModel"
   ]
 }

pipeline.py

@@ -1,62 +0,0 @@
-from typing import List, Optional, Tuple, Union
-from diffusers import DiffusionPipeline, ImagePipelineOutput
-from diffusers.utils.torch_utils import randn_tensor
-
-import torch
-
-
-class DDPMConditionalPipeline(DiffusionPipeline):
-    model_cpu_offload_seq = "unet"
-
-    def __init__(self, unet, scheduler):
-        super().__init__()
-        self.register_modules(unet=unet, scheduler=scheduler)
-
-    @torch.no_grad()
-    def __call__(
-        self,
-        label,
-        batch_size: int = 1,
-        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        num_inference_steps: int = 1000,
-        output_type: Optional[str] = "pil",
-        return_dict: bool = True,
-    ) -> Union[ImagePipelineOutput, Tuple]:
-        # Sample gaussian noise to begin loop
-        if isinstance(self.unet.model.sample_size, int):
-            image_shape = (
-                batch_size,
-                self.unet.model.in_channels,
-                self.unet.model.sample_size,
-                self.unet.model.sample_size,
-            )
-        else:
-            image_shape = (
-                batch_size,
-                self.unet.model.in_channels,
-                *self.unet.model.sample_size,
-            )
-
-        image = randn_tensor(image_shape, generator=generator)
-
-        # set step values
-        self.scheduler.set_timesteps(num_inference_steps)
-
-        for t in self.progress_bar(self.scheduler.timesteps):
-            # 1. predict noise model_output
-            model_output = self.unet(image, t, label).sample
-
-            # 2. compute previous image: x_t -> x_t-1
-            image = self.scheduler.step(
-                model_output, t, image, generator=generator
-            ).prev_sample
-
-        image = (image / 2 + 0.5).clamp(0, 1)
-        image = image.cpu().permute(0, 2, 3, 1).numpy()
-        if output_type == "pil":
-            image = self.numpy_to_pil(image)
-
-        if not return_dict:
-            return (image,)
-
-        return ImagePipelineOutput(images=image)

unet/config.json

@@ -0,0 +1,42 @@
+{
+  "_class_name": "UNet2DModel",
+  "_diffusers_version": "0.28.0",
+  "act_fn": "silu",
+  "add_attention": true,
+  "attention_head_dim": 8,
+  "attn_norm_num_groups": null,
+  "block_out_channels": [
+    32,
+    64,
+    64
+  ],
+  "center_input_sample": false,
+  "class_embed_type": null,
+  "down_block_types": [
+    "DownBlock2D",
+    "AttnDownBlock2D",
+    "AttnDownBlock2D"
+  ],
+  "downsample_padding": 1,
+  "downsample_type": "conv",
+  "dropout": 0.0,
+  "flip_sin_to_cos": true,
+  "freq_shift": 0,
+  "in_channels": 1,
+  "layers_per_block": 2,
+  "mid_block_scale_factor": 1,
+  "norm_eps": 1e-05,
+  "norm_num_groups": 32,
+  "num_class_embeds": 10,
+  "num_train_timesteps": null,
+  "out_channels": 1,
+  "resnet_time_scale_shift": "default",
+  "sample_size": 32,
+  "time_embedding_type": "positional",
+  "up_block_types": [
+    "AttnUpBlock2D",
+    "AttnUpBlock2D",
+    "UpBlock2D"
+  ],
+  "upsample_type": "conv"
+}

unet/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a58e881a5db3333ead19d140b55f8744ab337934c05d4cd30ddc3dc7a5734415
+size 6865780