diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..0aa25ddadbb9e4de91db84d7856a9e80e65bbb41 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text *.zip filter=lfs diff=lfs merge=lfs -text *.zst filter=lfs diff=lfs merge=lfs -text *tfevents* filter=lfs diff=lfs merge=lfs -text +image_control.png filter=lfs diff=lfs merge=lfs -text +images_0.png filter=lfs diff=lfs merge=lfs -text +images_1.png filter=lfs diff=lfs merge=lfs -text +images_2.png filter=lfs diff=lfs merge=lfs -text +images_3.png filter=lfs diff=lfs merge=lfs -text +images_4.png filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..869e1cb7c3b7898a92b0024de748c4c886467ee3 --- /dev/null +++ b/README.md @@ -0,0 +1,51 @@ +--- +base_model: stabilityai/stable-diffusion-2-1-base +library_name: diffusers +license: creativeml-openrail-m +inference: true +tags: +- stable-diffusion +- stable-diffusion-diffusers +- text-to-image +- diffusers +- controlnet +- diffusers-training +--- + + + + +# controlnet-Amitz244/output_dir_controlnet + +These are controlnet weights trained on stabilityai/stable-diffusion-2-1-base with new type of conditioning. +You can find some example images below. + +prompt: Woman in blue and black on a large plaza. +![images_0)](./images_0.png) +prompt: A men's restroom showcasing the toilet through an open door. +![images_1)](./images_1.png) +prompt: A man riding a kiteboard over the ocean under a cloudy sky. +![images_2)](./images_2.png) +prompt: Two skiers stand on their skis in the snow. +![images_3)](./images_3.png) +prompt: A meal of cheese toast, spaghetti, and broccoli on a white plate. +![images_4)](./images_4.png) + + + +## Intended uses & limitations + +#### How to use + +```python +# TODO: add an example code snippet for running this diffusion pipeline +``` + +#### Limitations and bias + +[TODO: provide examples of latent issues and potential remediations] + +## Training details + +[TODO: describe the data used to train the model] \ No newline at end of file diff --git a/checkpoint-10360/controlnet/config.json b/checkpoint-10360/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-10360/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-10360/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-10360/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0b5b2333f2dbdaa950f0fc33a2a38f5a0f04b1f5 --- /dev/null +++ b/checkpoint-10360/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c07ed1e7cdf01df24aaa75be0143d227d9226cb9974f1be5b6e381917790b03 +size 1456953560 diff --git a/checkpoint-10360/optimizer.bin b/checkpoint-10360/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..5b51d3a8bbe827f60f73e3293899d021d62de93a --- /dev/null +++ b/checkpoint-10360/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c958ef4b62c41129e95b70fbd2ec471e054cb4510be9053525f637d91a99549 +size 2914111182 diff --git a/checkpoint-10360/random_states_0.pkl b/checkpoint-10360/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..b5a5b723284f90ae14bd5694146dff156121394a --- /dev/null +++ b/checkpoint-10360/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ee16c8732289151838fc3cd8cc181b4f98420f94b2bff3b553ce559a2472de9 +size 14408 diff --git a/checkpoint-10360/scaler.pt b/checkpoint-10360/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..68f6361c6c8d2834e7cc26de7ca56856dbffaa13 --- /dev/null +++ b/checkpoint-10360/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3773287d9fcf328897ef980c39661889bbebea3f4fe85e16a0fe2a3bd04dfa19 +size 988 diff --git a/checkpoint-10360/scheduler.bin b/checkpoint-10360/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..a0feb24294147470b7b21c15f3312247f55cf955 --- /dev/null +++ b/checkpoint-10360/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c6debe9a480eb3313ff62e5c1c32551163ac09b4f1da91c440cbd193314696c +size 1000 diff --git a/checkpoint-12432/controlnet/config.json b/checkpoint-12432/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-12432/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-12432/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-12432/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..476a6575dd273820ad869648a62518f19208e3ea --- /dev/null +++ b/checkpoint-12432/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69b95241ac59c001203ad289b4b031ecf9f248cd606ca02c0a0753bb7fe27e6d +size 1456953560 diff --git a/checkpoint-12432/optimizer.bin b/checkpoint-12432/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..9c6ae833d69b35811fe003f999f95a6714810c0f --- /dev/null +++ b/checkpoint-12432/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d121f8383c3a6423ea795c486d5d62c829fb052a222152d1836658cc28680f4e +size 2914111182 diff --git a/checkpoint-12432/random_states_0.pkl b/checkpoint-12432/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..31c1aa5bb9040b3abba06f38f091d25af38fa868 --- /dev/null +++ b/checkpoint-12432/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e941b60917d177670efc4eca71ffa80c006b835868aa981103c741f057df6897 +size 14408 diff --git a/checkpoint-12432/scaler.pt b/checkpoint-12432/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..4a00453db238ec066a6a6a0f4caa0fc5cf388a40 --- /dev/null +++ b/checkpoint-12432/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5c528c51f1e929a14ce059fc408e7e7c49c660ad323792a4fc6bf41de0c113f +size 988 diff --git a/checkpoint-12432/scheduler.bin b/checkpoint-12432/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..ca142de039f49605e8d88f130b9ef11f7aea27c5 --- /dev/null +++ b/checkpoint-12432/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f20bddb1c54f7c1c8540fee10b71c2cbd2ee3a943b7c25215ac911729faa9616 +size 1000 diff --git a/checkpoint-2072/controlnet/config.json b/checkpoint-2072/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-2072/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-2072/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-2072/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..584746816f29c344db8c57932add64795a213f5a --- /dev/null +++ b/checkpoint-2072/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef40ce4f85bee69126b20aa16d10382976f1974028d0e68ab48752c697bf1f64 +size 1456953560 diff --git a/checkpoint-2072/optimizer.bin b/checkpoint-2072/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..868f405be39dfa762343065c5ec8404406fe4ef8 --- /dev/null +++ b/checkpoint-2072/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7497263c88ff5d631f79c1e38b93db24b9a8979d712263529edebb0569012cb1 +size 2914111182 diff --git a/checkpoint-2072/random_states_0.pkl b/checkpoint-2072/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..8bf600d3afe1253e05b8deddf67ead284b64c8f3 --- /dev/null +++ b/checkpoint-2072/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53614ea51971c4f50120132b84bff318de082108b9416978b6f79d1ae0029d03 +size 14408 diff --git a/checkpoint-2072/scaler.pt b/checkpoint-2072/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1f013bb65615e1bf26012f5f9d81c46fbb1aabe --- /dev/null +++ b/checkpoint-2072/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f0d2b6bdfffc43eec111e578b6b2f09a13229e8d7d7d44af5c8d3709726691a +size 988 diff --git a/checkpoint-2072/scheduler.bin b/checkpoint-2072/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..68dab3c97b6d690317f9b86f34ab999dedd0d75b --- /dev/null +++ b/checkpoint-2072/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a54133d5741a6bc0c257a0e37edcb466d65e55ec8fa751742769f48a1a78605 +size 1000 diff --git a/checkpoint-4144/controlnet/config.json b/checkpoint-4144/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-4144/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-4144/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-4144/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b077730898611e411c3203eb5cc593ca081e9992 --- /dev/null +++ b/checkpoint-4144/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd43e8b096b5213b0d1ad2fc725627c357892ee2e18f8d160088da996c42dc50 +size 1456953560 diff --git a/checkpoint-4144/optimizer.bin b/checkpoint-4144/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..505964362a2bf7b465d25e17f0bf1b0cdf9a9bdd --- /dev/null +++ b/checkpoint-4144/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:764732243a5bc58db9e67b8dd9575f36b73bc05806f3f01f895d70ffbde64e21 +size 2914111182 diff --git a/checkpoint-4144/random_states_0.pkl b/checkpoint-4144/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..ca913e5c82959a48fa2af8ffaf20d5b1bb13ac67 --- /dev/null +++ b/checkpoint-4144/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1eacca36f3e3f64c972a7dcd20f850d98e368542584235f3d05fccfb7f3c77fa +size 14408 diff --git a/checkpoint-4144/scaler.pt b/checkpoint-4144/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ba93df847b2548de3504e03df26bc7b82c34c8c4 --- /dev/null +++ b/checkpoint-4144/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c94e5078b1e82c1200457e1eac97cbc88da8deb856053356bf11270d3b4ef85c +size 988 diff --git a/checkpoint-4144/scheduler.bin b/checkpoint-4144/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..22b4c19ce23ea85734b83f8377448a2ae40d9d13 --- /dev/null +++ b/checkpoint-4144/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62a678f252988df6ff432b51ece6f62d1595e5f093edb3dcbda6690adcedf910 +size 1000 diff --git a/checkpoint-5000/controlnet/config.json b/checkpoint-5000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-5000/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-5000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-5000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ac0b062540ea9bb5fb5c3d2640e1890b37b9bfb8 --- /dev/null +++ b/checkpoint-5000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc812a92c089217cd33e8fc053c458f63c7b9d6be6644c20b29bac668fd5020d +size 1456953560 diff --git a/checkpoint-5000/optimizer.bin b/checkpoint-5000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b57175ebcfe1c4a8aecc26e99e27f0b84ed3771 --- /dev/null +++ b/checkpoint-5000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:641b7330c769bb3e3800b16d4f2cfdb0eb1f6c8607b2b388b6958b302cb78342 +size 2914111182 diff --git a/checkpoint-5000/random_states_0.pkl b/checkpoint-5000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..a19678e8e11c197286b5ee62e4885361ae4212bf --- /dev/null +++ b/checkpoint-5000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94925e4ff44b31f798aa5de4f0a0d79c8c6d23fd41758f8b30fd145b4e83225a +size 14344 diff --git a/checkpoint-5000/scaler.pt b/checkpoint-5000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..da81849ef9b20a8870285eaa92c68ae84344e397 --- /dev/null +++ b/checkpoint-5000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41216328ab75de937007afb04d76156949bedb908461001a451c8991c2ba8cca +size 988 diff --git a/checkpoint-5000/scheduler.bin b/checkpoint-5000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..976a27de201da5488bc7b42ce6d3ddff38320cae --- /dev/null +++ b/checkpoint-5000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:920333155b08e76154b89a85170c320833074db7b84649a9cfe79fd10b6bf2d9 +size 1000 diff --git a/checkpoint-6216/controlnet/config.json b/checkpoint-6216/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-6216/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-6216/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-6216/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..37d32019e72381b34596f6b6a6eb33a8cbf802e4 --- /dev/null +++ b/checkpoint-6216/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9961eb0b34a3d403786ddeb81c6cb600f706ff0ddac46410c122b7891d431b99 +size 1456953560 diff --git a/checkpoint-6216/optimizer.bin b/checkpoint-6216/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..4a0e0545c3975c66013006036aa4889b588ac257 --- /dev/null +++ b/checkpoint-6216/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:326bc9e7f3cde19bae257188196f237f5fe7e5c77561cacde19dea0bed09b4c0 +size 2914111182 diff --git a/checkpoint-6216/random_states_0.pkl b/checkpoint-6216/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..aa41ddcf3ea11a8516ab5026fc143f31809d5315 --- /dev/null +++ b/checkpoint-6216/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dde78fc9f17b0e7c1a08adf4986acac9125b3ecffeb9ae59551fc045f48c7fda +size 14472 diff --git a/checkpoint-6216/scaler.pt b/checkpoint-6216/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..e3e2c98159105f04820574d811c5e37a67df0d14 --- /dev/null +++ b/checkpoint-6216/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:faa4e6e1fd533a919c432d351eb6ab7372184928ef4295bf1f1c6689dc90045b +size 988 diff --git a/checkpoint-6216/scheduler.bin b/checkpoint-6216/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..715ef891af5064c950b245295d4c4d5d6d430c43 --- /dev/null +++ b/checkpoint-6216/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:022a4a9af12dab34df0cc618a91afbca7460f63ff24f21753fb37a1fa8d00a45 +size 1000 diff --git a/checkpoint-8288/controlnet/config.json b/checkpoint-8288/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/checkpoint-8288/controlnet/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-8288/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-8288/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e370bab7e69954e1fc01254e90421720ff8c9501 --- /dev/null +++ b/checkpoint-8288/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48a92b6a81fc49011e573e3ab25398a7528e5b2171f6b4cb2ddd707dabbc4747 +size 1456953560 diff --git a/checkpoint-8288/optimizer.bin b/checkpoint-8288/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..3ecaf8902c624979162176db130ad02ccf77e925 --- /dev/null +++ b/checkpoint-8288/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7bfee117ade12d6f8e3ea8ab031bde401ab9529ae80ce8371225d8c3f5f8462 +size 2914111182 diff --git a/checkpoint-8288/random_states_0.pkl b/checkpoint-8288/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..a7a60b6d322e9b554e639f4fa990fb33c31fb21b --- /dev/null +++ b/checkpoint-8288/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d8798085083f6870930e7e09e5bbfab72502028a22c7d0195d619cf4f86436b +size 14408 diff --git a/checkpoint-8288/scaler.pt b/checkpoint-8288/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c2859d700624a19dcadd28b5d31655d8eca5c3f --- /dev/null +++ b/checkpoint-8288/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:560da6388f984b2777b6f909d626f62e0cc824710b7f900824608aded0c1e790 +size 988 diff --git a/checkpoint-8288/scheduler.bin b/checkpoint-8288/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..1fff38fd6ec375dec84a6002d891996953a110a9 --- /dev/null +++ b/checkpoint-8288/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae48b7d09f81e407f55de6ee44ea66a781ecffe79ee123c2ac5bfc338f7eacb7 +size 1000 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8e70435dd284808d369ee492a2ff7a5e0aa9bc --- /dev/null +++ b/config.json @@ -0,0 +1,56 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": [ + 5, + 10, + 20, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 1024, + "down_block_types": [ + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D", + "DownBlock2D" + ], + "downsample_padding": 1, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "global_pool_conditions": false, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "projection_class_embeddings_input_dim": null, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": 1, + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/diffusion_pytorch_model.safetensors b/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..476a6575dd273820ad869648a62518f19208e3ea --- /dev/null +++ b/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69b95241ac59c001203ad289b4b031ecf9f248cd606ca02c0a0753bb7fe27e6d +size 1456953560 diff --git a/image_control.png b/image_control.png new file mode 100644 index 0000000000000000000000000000000000000000..f5f6a9ad3e724402d728dc2aa5edfb60763140c8 --- /dev/null +++ b/image_control.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8848d2d2d644b4a3323304e77ab0b0d4cf2a4e5a533ea9323c5e611d6e5a925 +size 237019 diff --git a/images_0.png b/images_0.png new file mode 100644 index 0000000000000000000000000000000000000000..a1f5d67eebcfb6660df59738c09300e5d8445028 --- /dev/null +++ b/images_0.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5e3a6704c7c9eb52ec28a6e89161f30ffa0eb557c124328c8b526e22a33e0a9 +size 1370670 diff --git a/images_1.png b/images_1.png new file mode 100644 index 0000000000000000000000000000000000000000..d0b41cbdcc1fcc3113ffeb36333db00c7ca86a7b --- /dev/null +++ b/images_1.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b8397826413f71c83d935141b80557c25715943cbbe75362d073f64c576bcf0 +size 1114110 diff --git a/images_2.png b/images_2.png new file mode 100644 index 0000000000000000000000000000000000000000..9785dd4f97f8c1f68a7bc516099f208229958f97 --- /dev/null +++ b/images_2.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c87811d08c5fb0c536d57e6a2020e4a05084f0dde82ae1ad82c4d235331e50a6 +size 1178427 diff --git a/images_3.png b/images_3.png new file mode 100644 index 0000000000000000000000000000000000000000..5ba78c17762a41ce70c0b7838b2870a6bb75e22f --- /dev/null +++ b/images_3.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b9c558abc7a5130f6583b03422d89f91af2d8cae5e3eb6f5f06aff504452de5 +size 1386932 diff --git a/images_4.png b/images_4.png new file mode 100644 index 0000000000000000000000000000000000000000..3cff86b067b8f918ead4432f8b826115d40770da --- /dev/null +++ b/images_4.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33ca5cbb243c370fc190e8ea12830399a6af73d6f853623332da906f0abc49aa +size 1507509 diff --git a/val_spiga_cond_0.jpg b/val_spiga_cond_0.jpg new file mode 100644 index 0000000000000000000000000000000000000000..7f61e05da341d7b6046dc3e16ce18b61d250a3e0 Binary files /dev/null and b/val_spiga_cond_0.jpg differ diff --git a/val_spiga_cond_1.jpg b/val_spiga_cond_1.jpg new file mode 100644 index 0000000000000000000000000000000000000000..7677d62ccbe793a4b212840cc47cd1c0a72ec03d Binary files /dev/null and b/val_spiga_cond_1.jpg differ