diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000000000000000000000000000000000000..79d1facbe641aa7da396a0875b650788e3edef65 --- /dev/null +++ b/.gitignore @@ -0,0 +1,2 @@ +step_* +epoch_* diff --git a/checkpoint-1000/optimizer.bin b/checkpoint-1000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..9d73e814f06e8d66402d957190708bf83dcfcab5 --- /dev/null +++ b/checkpoint-1000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a77936f971d8d2ea1acd4422d4db8c237425dbc060d925dd62900f4163bd96a9 +size 73251413 diff --git a/checkpoint-1000/random_states_0.pkl b/checkpoint-1000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..398fff6a3db20b312530d43f48ec2a5d19541486 --- /dev/null +++ b/checkpoint-1000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de63daf99e1e85577a98eb464946bfdf57af9cf3d994a1890c117f900500e031 +size 14663 diff --git a/checkpoint-1000/scaler.pt b/checkpoint-1000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..efdbd3c795f6b0d4144e68355e99c220ccdedd09 --- /dev/null +++ b/checkpoint-1000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68cff80b680ddf6e7abbef98b5f336b97f9b5963e2209307f639383870e8cc71 +size 557 diff --git a/checkpoint-1000/scheduler.bin b/checkpoint-1000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..5e2c3fc9b66bd9689084dcba42130c1cd7dfdfbc --- /dev/null +++ b/checkpoint-1000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7223915606ffeb4740373e378fc6ee25a8063899d81b73e9d9630a18bf386915 +size 563 diff --git a/checkpoint-1000/unet/config.json b/checkpoint-1000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-1000/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-1000/unet/diffusion_pytorch_model.bin b/checkpoint-1000/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..3db9fbee49298aaeb86978ffcbee3bdca63a9188 --- /dev/null +++ b/checkpoint-1000/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0477eb3c3da58436fe7693878f450a3e69f7433db4b6d64148040d747932e0a4 +size 36621969 diff --git a/checkpoint-1000/unet_ema/config.json b/checkpoint-1000/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..195e58a87e92c2be5e179310db3313d96a2bcffd --- /dev/null +++ b/checkpoint-1000/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 1000, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-1000/unet_ema/diffusion_pytorch_model.bin b/checkpoint-1000/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..b4df1f34e28c998a324d5ad20fb628b615a47f26 --- /dev/null +++ b/checkpoint-1000/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6fc32e94b6f9c420a0f6bedd1b118084b0245e5eb2a629bb0a42890f0fcb290e +size 36616081 diff --git a/checkpoint-1500/optimizer.bin b/checkpoint-1500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..8c34335bf3325c87b98b7378636a1d7c0c5de0c6 --- /dev/null +++ b/checkpoint-1500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:388a25582110c6010f1d4b7fbda1d958a222789efcac65af04579d360123fe67 +size 73251413 diff --git a/checkpoint-1500/random_states_0.pkl b/checkpoint-1500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..45de92c5faa4c6bc19c4a7ad3a009c806bab2ddb --- /dev/null +++ b/checkpoint-1500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86ed5ce96de2e3c8754ac07d03c8821c114ad84b467e60f3b26e38ffa8f2f3a6 +size 14663 diff --git a/checkpoint-1500/scaler.pt b/checkpoint-1500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..69c76b8f27f096d1d1a3d9d0e387af1a9cf5028d --- /dev/null +++ b/checkpoint-1500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:203a72d6c29f42a0e2964fdddc8d7a98df1eccee78fea9de0fa416613390f5c6 +size 557 diff --git a/checkpoint-1500/scheduler.bin b/checkpoint-1500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..220a92c33a912b1260e3731b7c31bb7946990e68 --- /dev/null +++ b/checkpoint-1500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1bc53d21d5be22242495d40019e29e447be310bc47a92764b1589cb8c4e4483 +size 563 diff --git a/checkpoint-1500/unet/config.json b/checkpoint-1500/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-1500/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-1500/unet/diffusion_pytorch_model.bin b/checkpoint-1500/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..cb1a28b640f4dc773f753aeb4e05e9eef9712b8c --- /dev/null +++ b/checkpoint-1500/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8add4a43c1a1db5684f0d2cd55111c1913da47423f7cee6b3a20332752863cf +size 36621969 diff --git a/checkpoint-1500/unet_ema/config.json b/checkpoint-1500/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5e5d9d9548c5361ed8b5f4d7b22d33b346179397 --- /dev/null +++ b/checkpoint-1500/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 1500, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-1500/unet_ema/diffusion_pytorch_model.bin b/checkpoint-1500/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5897a7dfccbbf1687a1b2bae5d8097849e23fc6 --- /dev/null +++ b/checkpoint-1500/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e74396c9b33addec059b775ab45cd1cf68977d48595193d5f5bc1050374b3c83 +size 36616081 diff --git a/checkpoint-2000/optimizer.bin b/checkpoint-2000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..43f1c8e3c913a19d110769378b1e1fa94497cfc1 --- /dev/null +++ b/checkpoint-2000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa19110b484d3e8835f8af390579d747dc676c6d6fb88470b470a9868d876067 +size 73251413 diff --git a/checkpoint-2000/random_states_0.pkl b/checkpoint-2000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..39b2478d505d2ac5dd3f2f9b1f4dd8bb0a103e94 --- /dev/null +++ b/checkpoint-2000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e9469ece6dcdb0453cef4fc90f3a4c75c62f81cee97472304b775164a5bcc11 +size 14663 diff --git a/checkpoint-2000/scaler.pt b/checkpoint-2000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..21ece35416ce79724d347155f11cfa297b97cabc --- /dev/null +++ b/checkpoint-2000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd2de9749828adacdf103bf6e9592702bb7067a2c1df27dd62ab38c1eb8c070f +size 557 diff --git a/checkpoint-2000/scheduler.bin b/checkpoint-2000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..e92da66d153ba6569a357066c095d8d16b49da18 --- /dev/null +++ b/checkpoint-2000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cc6a2e9846582647ba829a5cab5e19379ab6169fd378876d9a2c2b5a948f3fb +size 563 diff --git a/checkpoint-2000/unet/config.json b/checkpoint-2000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-2000/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-2000/unet/diffusion_pytorch_model.bin b/checkpoint-2000/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..e2bf6c5bcd0bfb5f04965423f0fdf2b8d79e6627 --- /dev/null +++ b/checkpoint-2000/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f031e9cc3237b6887c68efab79a28c90b749773e6666cdddc8c8baed3252c5c5 +size 36621969 diff --git a/checkpoint-2000/unet_ema/config.json b/checkpoint-2000/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..664b2bf07e188ee627e139dfacfde3c44edef3ed --- /dev/null +++ b/checkpoint-2000/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 2000, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-2000/unet_ema/diffusion_pytorch_model.bin b/checkpoint-2000/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..87ba471166c8423979308bcb75347bdf2a7f41ea --- /dev/null +++ b/checkpoint-2000/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87774ed9e33e0d376e35fce7d12b616f01f6f11465633254672a6d003395b9ce +size 36616081 diff --git a/checkpoint-2500/optimizer.bin b/checkpoint-2500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..50ee829c3b5cc71f68d924823b6395d21f980ca5 --- /dev/null +++ b/checkpoint-2500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33a5906232d8b5340b2798a2386d794822167bb0c50067a05cf640f19fe5fab1 +size 73251413 diff --git a/checkpoint-2500/random_states_0.pkl b/checkpoint-2500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..ecea9d1054149a9db60ae966b1d26edbc3c57dce --- /dev/null +++ b/checkpoint-2500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50c0cb8597bad2e92fd83ebd2e3f8b2a93db3b2c2a08002c5904d457e207a498 +size 14663 diff --git a/checkpoint-2500/scaler.pt b/checkpoint-2500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..9a3e6e139e26b18feddbe2f66ba98b73890dacfa --- /dev/null +++ b/checkpoint-2500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fbcebc8f5487b0c117b5dd47f2ea304af3eebf408d297118d9307e1223927e1 +size 557 diff --git a/checkpoint-2500/scheduler.bin b/checkpoint-2500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..89afd616ca9d02edf85d4437220dd708e67e2536 --- /dev/null +++ b/checkpoint-2500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fdee5d0999ddc7c72ad066f56b8124a7730e2bdf658fa2439ba19b02cd9f42a4 +size 563 diff --git a/checkpoint-2500/unet/config.json b/checkpoint-2500/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-2500/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-2500/unet/diffusion_pytorch_model.bin b/checkpoint-2500/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..2697dcc92c9b006f3fd079aa54f6ad6bcb02c125 --- /dev/null +++ b/checkpoint-2500/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82129e54faee0789504b93182c5be5e4e3b420b5abbb9f06be442fcef356877b +size 36621969 diff --git a/checkpoint-2500/unet_ema/config.json b/checkpoint-2500/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3c9db3cb30b887d9e3e2f34dce3bfbdb32d3f2f1 --- /dev/null +++ b/checkpoint-2500/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 2500, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-2500/unet_ema/diffusion_pytorch_model.bin b/checkpoint-2500/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..b4b56a5e500da281724d843173762689ea4bbf9e --- /dev/null +++ b/checkpoint-2500/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2e218d62ccfb426d69e9fd9cd48f1039f86bd27f0fc5ecd18c4ccb3983ea5b0 +size 36616081 diff --git a/checkpoint-3000/optimizer.bin b/checkpoint-3000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..31245c8ed0547d1e8eae3af0137222fd1afe7ff4 --- /dev/null +++ b/checkpoint-3000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f56cda5f86ba292b1597f9730a62f9cb225895fb8d7a29067a761a00df701ba +size 73251413 diff --git a/checkpoint-3000/random_states_0.pkl b/checkpoint-3000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..eea68b8857d2a0a3beaa5cf3cb5d7f1aad68b57f --- /dev/null +++ b/checkpoint-3000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3068affad1a930678c6735175ca5d3dfa0984dc67a3962b2497b204ebeb3ef5 +size 14663 diff --git a/checkpoint-3000/scaler.pt b/checkpoint-3000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..2888c9ddc2c30e6b47ce0e6abcc8e2d1303aab00 --- /dev/null +++ b/checkpoint-3000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb1f9398b77268202e8e1465734a63d123b1ef11c27f20f2473677e9883a6869 +size 557 diff --git a/checkpoint-3000/scheduler.bin b/checkpoint-3000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..d17e7dcb0ffbeb7dcd292098e8b538e24dd1fbfb --- /dev/null +++ b/checkpoint-3000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c7d765e228e9ca3b9dccd3fd0195cdbf25c24ea82c4bcf3f3ac6bd257583232 +size 563 diff --git a/checkpoint-3000/unet/config.json b/checkpoint-3000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-3000/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-3000/unet/diffusion_pytorch_model.bin b/checkpoint-3000/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..085ab13ae07ae79afdc3e9058e171eee0a3a5d8a --- /dev/null +++ b/checkpoint-3000/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad5eb4c3197ea5ddf0ab55d38cb1e20a8ca3aebf01cb5df5dd940a75bcf2b167 +size 36621969 diff --git a/checkpoint-3000/unet_ema/config.json b/checkpoint-3000/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8a5776629eeef00d3eba68418c5ec8765c215fee --- /dev/null +++ b/checkpoint-3000/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 3000, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-3000/unet_ema/diffusion_pytorch_model.bin b/checkpoint-3000/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..8db0387ab4cb17801617fb79c5e1bae285afb834 --- /dev/null +++ b/checkpoint-3000/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:676b476c0f696e91d8fe7792584f22e2b6d430837768412307a6aacb2b29c576 +size 36616081 diff --git a/checkpoint-500/optimizer.bin b/checkpoint-500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..cce82809f6c4b9630acb5e6e675357922f76150d --- /dev/null +++ b/checkpoint-500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:56d7698aa7b35cbea9ce99010e5f33918567c9e960274198f4228d9fa0fc4aa9 +size 73251413 diff --git a/checkpoint-500/random_states_0.pkl b/checkpoint-500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..0da731cfc96064e9487b3d57204076ec3d48bb48 --- /dev/null +++ b/checkpoint-500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77af94b4d3fc305349206cfc5301339c4d0593d60b943db6d53a25e2d2f57af9 +size 14663 diff --git a/checkpoint-500/scaler.pt b/checkpoint-500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5f2e5ee77c2cef5eff816170558ce9209855685 --- /dev/null +++ b/checkpoint-500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3f196a54202bb4ba1220e8c59f42f9cda0702d68ea83147d814c2fb2f36b8f2 +size 557 diff --git a/checkpoint-500/scheduler.bin b/checkpoint-500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..e6c5b29ae22621e607df84fd6532ef593dede11f --- /dev/null +++ b/checkpoint-500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1eedad58536ac188d101845e13deb36e3aca8514e3298625b0fb265ebd7ea7a +size 563 diff --git a/checkpoint-500/unet/config.json b/checkpoint-500/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/checkpoint-500/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/checkpoint-500/unet/diffusion_pytorch_model.bin b/checkpoint-500/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..b254e0fcb215149afcb708957cec876c79e3de87 --- /dev/null +++ b/checkpoint-500/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:506c50a28416c82c4ed9476e0dfeda6135694134ec4e0ea046aba0b7b719d743 +size 36621969 diff --git a/checkpoint-500/unet_ema/config.json b/checkpoint-500/unet_ema/config.json new file mode 100644 index 0000000000000000000000000000000000000000..1110a0c94071ab30d43f7f2a7fea6fa7989a1ea2 --- /dev/null +++ b/checkpoint-500/unet_ema/config.json @@ -0,0 +1,67 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "decay": 0.9999, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "inv_gamma": 1.0, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "min_decay": 0.0, + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "optimization_step": 500, + "out_channels": 3, + "power": 0.75, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "update_after_step": 0, + "use_ema_warmup": true, + "use_linear_projection": false +} diff --git a/checkpoint-500/unet_ema/diffusion_pytorch_model.bin b/checkpoint-500/unet_ema/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..843b18f756017c49e3a616f354639dec07cb0ebf --- /dev/null +++ b/checkpoint-500/unet_ema/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58d7b1b09e4b5d30175b767403b8ec619ef8d23ef4938369b4a7de08166e68ce +size 36616081 diff --git a/model_index.json b/model_index.json new file mode 100644 index 0000000000000000000000000000000000000000..03e20d2fa1312c95132533293ca493878e184ca3 --- /dev/null +++ b/model_index.json @@ -0,0 +1,12 @@ +{ + "_class_name": "LayoutDDPMPipeline", + "_diffusers_version": "0.16.1", + "scheduler": [ + "diffusers", + "DDPMScheduler" + ], + "unet": [ + "diffusers", + "UNet2DConditionModel" + ] +} diff --git a/scheduler/scheduler_config.json b/scheduler/scheduler_config.json new file mode 100644 index 0000000000000000000000000000000000000000..b7485e2539b46708b51281a0d29449fc08a32b61 --- /dev/null +++ b/scheduler/scheduler_config.json @@ -0,0 +1,16 @@ +{ + "_class_name": "DDPMScheduler", + "_diffusers_version": "0.16.1", + "beta_end": 0.02, + "beta_schedule": "linear", + "beta_start": 0.0001, + "clip_sample": true, + "clip_sample_range": 1.0, + "dynamic_thresholding_ratio": 0.995, + "num_train_timesteps": 1000, + "prediction_type": "epsilon", + "sample_max_value": 1.0, + "thresholding": false, + "trained_betas": null, + "variance_type": "fixed_small" +} diff --git a/unet/config.json b/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb1cd664fc36d4dc821a4f0fdae4b94d7d99b356 --- /dev/null +++ b/unet/config.json @@ -0,0 +1,60 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.16.1", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "attention_head_dim": 8, + "block_out_channels": [ + 64, + 64, + 128, + 128 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 256, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "DownBlock2D", + "DownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 3, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 3, + "projection_class_embeddings_input_dim": null, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "sample_size": 256, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "up_block_types": [ + "CrossAttnUpBlock2D", + "UpBlock2D", + "UpBlock2D", + "UpBlock2D" + ], + "upcast_attention": false, + "use_linear_projection": false +} diff --git a/unet/diffusion_pytorch_model.bin b/unet/diffusion_pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..269ee50053fb6b29f1c63897a1b9b8890ede9067 --- /dev/null +++ b/unet/diffusion_pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5935601c9d8121234a587458ef2e7e6be86b933328c8a740118a50c076c6ee5 +size 36621969