diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..eea1d8e4da6083525243bbdb899f6bae1ce3d57c 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,5 @@ 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 +images_0.png filter=lfs diff=lfs merge=lfs -text +images_1.png filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..07d4a38abc163ca1d5655f57cd40a0d1d93201d6 --- /dev/null +++ b/README.md @@ -0,0 +1,45 @@ +--- +base_model: runwayml/stable-diffusion-v1-5 +library_name: diffusers +license: creativeml-openrail-m +tags: +- stable-diffusion +- stable-diffusion-diffusers +- text-to-image +- diffusers +- controlnet +- diffusers-training +inference: true +--- + + + + +# controlnet-pribadihcr/controlnet_mvtec + +These are controlnet weights trained on runwayml/stable-diffusion-v1-5 with new type of conditioning. +You can find some example images below. + +prompt: a metal_nut with a bent +![images_0)](./images_0.png) +prompt: a metal_nut with a bent +![images_1)](./images_1.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-12000/controlnet/config.json b/checkpoint-12000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-12000/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-12000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-12000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..acb101d08c0e556464a508d060bbad70cc6ead6b --- /dev/null +++ b/checkpoint-12000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4701b799f603d74658920bf4d892398dce1755f4f728c0816669897493edff3b +size 1445157120 diff --git a/checkpoint-12000/optimizer.bin b/checkpoint-12000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..22548346a69de3bc5038d940e613da349a5b366c --- /dev/null +++ b/checkpoint-12000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebe768d6faf9ca3eca4fe23a00242a83baa7307c837e91ac325c27932528601d +size 2890518478 diff --git a/checkpoint-12000/random_states_0.pkl b/checkpoint-12000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..cd256e927c3b42e8caf69324da644dd5611c7a5d --- /dev/null +++ b/checkpoint-12000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:134c73212136fb27ecf74f36d012bc8932e820cbbf38fbd2b6f3b49dd1265263 +size 14344 diff --git a/checkpoint-12000/scaler.pt b/checkpoint-12000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d508dd2c8992e64f37781434bb6bb2c7f70d2aab --- /dev/null +++ b/checkpoint-12000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d1f9d6a63312bf6188d8844a7d621f2ecda20b6c0aee9e19ddcea1c480954ba +size 988 diff --git a/checkpoint-12000/scheduler.bin b/checkpoint-12000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..6bdf17af1f99c81f133601e635d46318d3a4eee8 --- /dev/null +++ b/checkpoint-12000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62fa42bc0f349bf17974b501b2e04e150f941778516a7dc2dafa712c6dbb5f91 +size 1000 diff --git a/checkpoint-12500/controlnet/config.json b/checkpoint-12500/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-12500/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-12500/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-12500/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..90ab86daea05f77f0d9f67ce889c34f2519863a7 --- /dev/null +++ b/checkpoint-12500/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aeaa647004d1171fa8abaedb06dd735dca4b075dc71749f5e15ac6388ced7a24 +size 1445157120 diff --git a/checkpoint-12500/optimizer.bin b/checkpoint-12500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..247ee615b51688eadd4f94d73d33f83144ca5d42 --- /dev/null +++ b/checkpoint-12500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fcf0ca97ae2971ab78bfaeaa21baeeca801c3ae77052af044ab7c800bccfdd23 +size 2890518478 diff --git a/checkpoint-12500/random_states_0.pkl b/checkpoint-12500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..442a85ee4b3319242c7371c2a95d640de5bf0485 --- /dev/null +++ b/checkpoint-12500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1be8a92f0ad4059953d665a9efa296c657d1890809f276960167993f87c5fd3 +size 14344 diff --git a/checkpoint-12500/scaler.pt b/checkpoint-12500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d0c8abeebda41aa041cedaafab5e1b4e6b849b3 --- /dev/null +++ b/checkpoint-12500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:971b044be39c256f04af2a12fd12d4dcf9ae5e5d310c0887db9e15dbce71ba9b +size 988 diff --git a/checkpoint-12500/scheduler.bin b/checkpoint-12500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..6b406997e928d5db5b45b965f7683f757c078a89 --- /dev/null +++ b/checkpoint-12500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5019925475444f379a3b5e75caf7d43f13436cd993e7d17355fb0399f280dcbe +size 1000 diff --git a/checkpoint-13000/controlnet/config.json b/checkpoint-13000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-13000/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-13000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-13000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6be48ed1342dce21c792a870d0612a2cbca25aa6 --- /dev/null +++ b/checkpoint-13000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0673953e3e0a03c3defd706611c3449cd96cc66a93631d21aa4358b960f57ac4 +size 1445157120 diff --git a/checkpoint-13000/optimizer.bin b/checkpoint-13000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..ea7d09db1bbd21d2ce3f0bd3a4633159b72d8c54 --- /dev/null +++ b/checkpoint-13000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1dfb0a2d5a8de4221bfc63ef0679e5a6d1d8946574eadba2bd645353226a6cf9 +size 2890518478 diff --git a/checkpoint-13000/random_states_0.pkl b/checkpoint-13000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..3b88c92234c821b156bdd352734c0d33b31041a4 --- /dev/null +++ b/checkpoint-13000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67770bfee63821d271a888a3e5061afac0a76f397a2c6be915eb5eb5481a406e +size 14344 diff --git a/checkpoint-13000/scaler.pt b/checkpoint-13000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..4ced4b569621b283ab409fa5e41736fc0a8f43e6 --- /dev/null +++ b/checkpoint-13000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e77f98967818acefa7b6b48d86d0635b780ab57badf525d74c7b5b70c8009b64 +size 988 diff --git a/checkpoint-13000/scheduler.bin b/checkpoint-13000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..4bdd1a2f834a10d180f0aa35e0a9c0065002ea57 --- /dev/null +++ b/checkpoint-13000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72cb1b03d647881aad39d7d0a71d358611b03dd7c270ac648fccc6829b38899d +size 1000 diff --git a/checkpoint-13500/controlnet/config.json b/checkpoint-13500/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-13500/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-13500/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-13500/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..17ac8d89adbf56ad97524618abad59b911617e54 --- /dev/null +++ b/checkpoint-13500/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75edbe7e940ba2edc3a48f3f45ae0c32ae763a9b56ad15f34ee85173344f7917 +size 1445157120 diff --git a/checkpoint-13500/optimizer.bin b/checkpoint-13500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..fda1e15140fb43b7d619dacad4ebf0eb21d64f10 --- /dev/null +++ b/checkpoint-13500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:215884ff49df45b62f736126fd2c527e3f76b3d8fbe32a8c169aa8979a8a5c79 +size 2890518478 diff --git a/checkpoint-13500/random_states_0.pkl b/checkpoint-13500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..5b8d42907c382763e0660cbec310a974c11ec244 --- /dev/null +++ b/checkpoint-13500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:956c1eb80fc6adbd9299a9c7b79006b0da23ff44b4e1be87b97d2565bdafa0e8 +size 14408 diff --git a/checkpoint-13500/scaler.pt b/checkpoint-13500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..025e7ede23687ba6f87d08eab5f099a2f269162d --- /dev/null +++ b/checkpoint-13500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bb4ee3801b11e2b95773f86abaa2d1e4b1da1e9da1588595ee5a4248d6798c1 +size 988 diff --git a/checkpoint-13500/scheduler.bin b/checkpoint-13500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..b213c0a8a1e1862a5eb42be0d0fc617705ccce36 --- /dev/null +++ b/checkpoint-13500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:480ace677a4104a5bef0ca188f3ad85916106b493db364c7bdb8150a47558b9f +size 1000 diff --git a/checkpoint-14000/controlnet/config.json b/checkpoint-14000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-14000/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-14000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-14000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ef6b45b505b2579403b0d4451af25815604e2422 --- /dev/null +++ b/checkpoint-14000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:56f8174572bd40fb45f0cb8efe22d08d2ca9397f5e47c17c35728c7d5865c989 +size 1445157120 diff --git a/checkpoint-14000/optimizer.bin b/checkpoint-14000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..936641bdf8b4140b879699e2e3617207aad2dc2d --- /dev/null +++ b/checkpoint-14000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bbbaffd8f7a40b579731186163800e477ed908ed5404ec31c8ee4010ac08f10 +size 2890518478 diff --git a/checkpoint-14000/random_states_0.pkl b/checkpoint-14000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..59d0f95ac94eea21a43e5711f3fdd65b68c65b55 --- /dev/null +++ b/checkpoint-14000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed7ca707803b61f1b52ef90fde6c7921de036a319fb87700bb37f9e20c2e5f7e +size 14344 diff --git a/checkpoint-14000/scaler.pt b/checkpoint-14000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..1b662b51676a156b87defef3451e7cf318983cd7 --- /dev/null +++ b/checkpoint-14000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8972e137dff582ed156c48a434d8d5304443a88bca38ef8e7f0e80298e6f0533 +size 988 diff --git a/checkpoint-14000/scheduler.bin b/checkpoint-14000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..49efa70310463d886c40ae19ae1a0f74c0a6b89f --- /dev/null +++ b/checkpoint-14000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1693773abc7f45155c36387554802d6c4617d4ef15aeaae90156c786e525bc93 +size 1000 diff --git a/checkpoint-14500/controlnet/config.json b/checkpoint-14500/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-14500/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-14500/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-14500/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..04634fd21697b4e76c3b56ec17d5f2d92fcacdf1 --- /dev/null +++ b/checkpoint-14500/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4225c434f9e3be74afa04cfce292786afc91cb5118ae3b7f6d0098ae884279b5 +size 1445157120 diff --git a/checkpoint-14500/optimizer.bin b/checkpoint-14500/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..95e96744c6e97afe900a11d6dd3e8bd7960d2055 --- /dev/null +++ b/checkpoint-14500/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:044990c462dab2aa17e027d8b9c197a6c4d92137bc0efddf2401ef4c4644fc8c +size 2890518478 diff --git a/checkpoint-14500/random_states_0.pkl b/checkpoint-14500/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..57ba6e62b23801377d8db128db2fd0238c38556a --- /dev/null +++ b/checkpoint-14500/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6948962a8073bcf51fdd514aa08500a80ffa73d9f1b880252b6050c6a5d6f23f +size 14344 diff --git a/checkpoint-14500/scaler.pt b/checkpoint-14500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..92e6df6461226c7c7dd6dd7a0004d1b6591aacbd --- /dev/null +++ b/checkpoint-14500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fa669548d7d0cea54c34bbdeb3defa489898ad7ca32a778f0f5d4793a93394b +size 988 diff --git a/checkpoint-14500/scheduler.bin b/checkpoint-14500/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..b701068ec2ead7b49d587864674836b52e317785 --- /dev/null +++ b/checkpoint-14500/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a689e735f57fd2077fcd0b978bf0fdd82476c3bfa187c6f6f29c2bf20fbc8377 +size 1000 diff --git a/checkpoint-15000/controlnet/config.json b/checkpoint-15000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-15000/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bde7232d7801792a79a0ebed36c76c6ae9125668 --- /dev/null +++ b/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b383efc000ee52fd850a73fd47e0e8602f8471e2d2ad504e1ff8339578d4bdf +size 1445157120 diff --git a/checkpoint-15000/optimizer.bin b/checkpoint-15000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..a6ab0a2ff03bc42c263dd648ea3cef618801f10d --- /dev/null +++ b/checkpoint-15000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d95a0f5ad8ac1e37a9cfa5c9635df93499e10d8132c7ab3723e8894eea490983 +size 2890518478 diff --git a/checkpoint-15000/random_states_0.pkl b/checkpoint-15000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..2b2b42ed020ae400d4cd25d1ae4a6658ce07f2d2 --- /dev/null +++ b/checkpoint-15000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e821e51d6b720e950611162cbf368d0bd319d5460cd2eeedfdc745a4e7c9bf3b +size 14344 diff --git a/checkpoint-15000/scaler.pt b/checkpoint-15000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..365015aeee3b31cbc062de74953db1994e36a224 --- /dev/null +++ b/checkpoint-15000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2491380f0b55e3e8cbfec1654220a94cf6d79d50cd1b1e51dfe593c04a274cb9 +size 988 diff --git a/checkpoint-15000/scheduler.bin b/checkpoint-15000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..6a2a4070a52ccbed96da55a7070bc6563a7f86b5 --- /dev/null +++ b/checkpoint-15000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb5488e48edf134f11b30bc97f9db87ef384ca8eea75a0471520f372e45b82a0 +size 1000 diff --git a/checkpoint-20000/controlnet/config.json b/checkpoint-20000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/checkpoint-20000/controlnet/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..00118813d66f0585654cad628563dffac90ef02c --- /dev/null +++ b/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:baacf7f57c2ab3a1a27481b1368e29a1494839de1b9901c32722f37aa1cdec1f +size 1445157120 diff --git a/checkpoint-20000/optimizer.bin b/checkpoint-20000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..d9463c737cb2150964b255ded28d8335edab8fff --- /dev/null +++ b/checkpoint-20000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:933de07f0ce3f36340b15de1d6a021a1f545bd2755f67b986f9a9b9e4c5fb4e3 +size 2890518478 diff --git a/checkpoint-20000/random_states_0.pkl b/checkpoint-20000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..b75b760756e3751328d6723f7d90a62bf337c972 --- /dev/null +++ b/checkpoint-20000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7327ee0280e04f65d00e173624d4a1fbeb2eccfbe71c36433bb621cd309fa8f6 +size 14344 diff --git a/checkpoint-20000/scheduler.bin b/checkpoint-20000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..86c15d06a20918c0bcdf8341881c0f1294ac62e3 --- /dev/null +++ b/checkpoint-20000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51e0d744a95b954d01691d43399197ed4341f27678cc9bb35b811a7c14d85702 +size 1000 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7812aca132ee32bbe7167871d0ced6512d6535b3 --- /dev/null +++ b/config.json @@ -0,0 +1,51 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.30.0.dev0", + "act_fn": "silu", + "addition_embed_type": null, + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": null, + "attention_head_dim": 8, + "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": 768, + "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": false +} diff --git a/diffusion_pytorch_model.safetensors b/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bde7232d7801792a79a0ebed36c76c6ae9125668 --- /dev/null +++ b/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b383efc000ee52fd850a73fd47e0e8602f8471e2d2ad504e1ff8339578d4bdf +size 1445157120 diff --git a/image_control.png b/image_control.png new file mode 100644 index 0000000000000000000000000000000000000000..52dc2ef479834de91ae9a8aa64653ad8a81160fe Binary files /dev/null and b/image_control.png differ diff --git a/images_0.png b/images_0.png new file mode 100644 index 0000000000000000000000000000000000000000..3146daa5a88edd88ae5d271aeb7c48d9ebf22adc --- /dev/null +++ b/images_0.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:beca19f67ad8435cce83618f07367e137cfe81c0f79d9b7058f535378d4272a3 +size 2999307 diff --git a/images_1.png b/images_1.png new file mode 100644 index 0000000000000000000000000000000000000000..010b97326c5aacc3df6ef55cc943de3948628802 --- /dev/null +++ b/images_1.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80b179d4d36550238d969bdb15306c39c63e92e49e031e885b024328f40a676b +size 2916604