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..c2f859b3493289335f9db5b6879775d5ad3f2114 --- /dev/null +++ b/README.md @@ -0,0 +1,45 @@ +--- +base_model: SG161222/Realistic_Vision_V6.0_B1_noVAE +library_name: diffusers +license: creativeml-openrail-m +tags: +- stable-diffusion +- stable-diffusion-diffusers +- text-to-image +- diffusers +- controlnet +- diffusers-training +inference: true +--- + + + + +# controlnet-kmaksatk/experiments + +These are controlnet weights trained on SG161222/Realistic_Vision_V6.0_B1_noVAE with new type of conditioning. +You can find some example images below. + +prompt: human +![images_0)](./images_0.png) +prompt: human +![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-10000/controlnet/config.json b/checkpoint-10000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-10000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..03f5bb6cf5cd67649d97c8f141f71154ff690bed --- /dev/null +++ b/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:deb7655a81332a42f2f34449d1a31d5a3f502f29a912d8ccd27dfa6eda6d8820 +size 1445157120 diff --git a/checkpoint-10000/optimizer.bin b/checkpoint-10000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..a9f13f3c339552957c4eeaebbeb2c5fdead5e8eb --- /dev/null +++ b/checkpoint-10000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1189c1175dd4ae0d6e225b6b0827eef8c049a83aa7b4508953080c7a7358552f +size 2890518041 diff --git a/checkpoint-10000/random_states_0.pkl b/checkpoint-10000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..b66cd1e001b796a7133ddf0d9a903a8d5b59b656 --- /dev/null +++ b/checkpoint-10000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4af859b194e3c3e68d83f6fae5654996f1369278a8b80871aef56548b170f5ce +size 15691 diff --git a/checkpoint-10000/scaler.pt b/checkpoint-10000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b345659e084dfa08f03a221d79b2e302a4748dfe --- /dev/null +++ b/checkpoint-10000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9be2ad76dcbc923b00e6a142f6db62aad4a46c47bb83864ccb68ddc899d0ce78 +size 557 diff --git a/checkpoint-10000/scheduler.bin b/checkpoint-10000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..dfea8b103660873f57a0b99214cb9acb641ced7b --- /dev/null +++ b/checkpoint-10000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6bcd4cb37625cab1478c5f801ecf8dcfe8513c2acdf383cd99f5bc953b5e8a3 +size 563 diff --git a/checkpoint-100000/controlnet/config.json b/checkpoint-100000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-100000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-100000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-100000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ceec6b15aa743a06053baea12588717984112d0c --- /dev/null +++ b/checkpoint-100000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ac41dbceb31b850ec006715258cf9358d132093840b3add666db370ad4a3d06 +size 1445157120 diff --git a/checkpoint-100000/optimizer.bin b/checkpoint-100000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..156a8e11397692fd32e171f33c4896c48c16bf7b --- /dev/null +++ b/checkpoint-100000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:570be8cbc3be9c51571f37dc33039af6dcdd263d280cb719717bc42c2bb8b42c +size 2890518041 diff --git a/checkpoint-100000/random_states_0.pkl b/checkpoint-100000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..ea3bc2b9e8b1e0799f504f389a1e1ca540994fe0 --- /dev/null +++ b/checkpoint-100000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5bbc01bc2ce24eb1749642e5d166af205df2c53e52c656634f1634a32bb62b8 +size 15691 diff --git a/checkpoint-100000/scaler.pt b/checkpoint-100000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..17a9c4fbc9b244989ce412183ecb99b9b8e6db6e --- /dev/null +++ b/checkpoint-100000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41829e2c4f7df3c5fa6234a2e732fd1e48a2898de063ee21016fa3644b9acdf3 +size 557 diff --git a/checkpoint-100000/scheduler.bin b/checkpoint-100000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..bd8d336d5833d94339c2b260ae33e88f9b838b24 --- /dev/null +++ b/checkpoint-100000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82b0a8bc231b5e5e64fb22ea7198300cadcc40bdf2985229ae13f830a931bd7b +size 563 diff --git a/checkpoint-20000/controlnet/config.json b/checkpoint-20000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /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": null, + "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..ff1a80caa9fbcd5bb22d5578e207722f330b0bcb --- /dev/null +++ b/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a35f2a62f24b72dec6b82ba480a0179a8c4ae13b7f2c79de44057598609e6c6 +size 1445157120 diff --git a/checkpoint-20000/optimizer.bin b/checkpoint-20000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..be04c3688396599e9f8d35cb6c9c7a41812457c9 --- /dev/null +++ b/checkpoint-20000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3671e26d073e2bd8a395292b57e11743ddb4b9614074e9e0f00b8126e56dd2b7 +size 2890518041 diff --git a/checkpoint-20000/random_states_0.pkl b/checkpoint-20000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..d49146583196181fe5ffb58e7702cd9755688d79 --- /dev/null +++ b/checkpoint-20000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc02fd116bf3d2cd0309bfe1405a069603eda000e9695f6a08b1fff5851bf214 +size 15691 diff --git a/checkpoint-20000/scaler.pt b/checkpoint-20000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..4338be0dcab582fa3f12b919bf288f8541b81c65 --- /dev/null +++ b/checkpoint-20000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab00099fef848d9a422a8feada54918aab63a6d579c6ec1752d60e2249431549 +size 557 diff --git a/checkpoint-20000/scheduler.bin b/checkpoint-20000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..f5f384006297bac12235be37aaffdf65d823b113 --- /dev/null +++ b/checkpoint-20000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6af40fe1d26f16a16476a6515b0cb875814a602dabb7d83b4f2ad749c139b74 +size 563 diff --git a/checkpoint-30000/controlnet/config.json b/checkpoint-30000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-30000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cbfc46b97bc963a575948255eee3fbd120d109a4 --- /dev/null +++ b/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1de4959ad8bbfcadcfb7ad14c981434b9aac3abc710de617e089d71111bad0a +size 1445157120 diff --git a/checkpoint-30000/optimizer.bin b/checkpoint-30000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..0012d5b6484a40f7ca5126023be43ca046e9aa6f --- /dev/null +++ b/checkpoint-30000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fff9d70c045e038929350d143d22c01ae7c0b9b230dbae4cfa71a6ece74e8deb +size 2890518041 diff --git a/checkpoint-30000/random_states_0.pkl b/checkpoint-30000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..10b57cc71b87dccbfd7b2f60669ddf87467fdb1d --- /dev/null +++ b/checkpoint-30000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e2e60e07565d9014eb3614abd218ffa3012431089c782dc533cc18986b28ef1 +size 15691 diff --git a/checkpoint-30000/scaler.pt b/checkpoint-30000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..5f40e5e9adf8f34c103e877d99bb86992b3a5bf6 --- /dev/null +++ b/checkpoint-30000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a72403f39056a733e01bf1286a7f53ead8ff19baf1868d3b21024542372b54e +size 557 diff --git a/checkpoint-30000/scheduler.bin b/checkpoint-30000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..edd9607a06430b18e8ea60007f86b1e4b3c05dbd --- /dev/null +++ b/checkpoint-30000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dfb657437f2ee916b0df053ec6bf4d0b9df91f3e502cb70ba83e737ad8523ff +size 563 diff --git a/checkpoint-40000/controlnet/config.json b/checkpoint-40000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-40000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fb2ec222c2f43a71a5dcaae1e04630625e7ccdda --- /dev/null +++ b/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d34effc2d27e66ffb3bdc0562f1ef2752ca9b37011c3d345eb321fec0cc6059d +size 1445157120 diff --git a/checkpoint-40000/optimizer.bin b/checkpoint-40000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..2bceee4412198fa224687fc83cb4e194ce6da6df --- /dev/null +++ b/checkpoint-40000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7f3e5186c81170fa68332f8f143dde45d43113aa475e023b385ac7def1efb3c +size 2890518041 diff --git a/checkpoint-40000/random_states_0.pkl b/checkpoint-40000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..d8a43112a2e0d63aa2e3cd708056119ae9cea151 --- /dev/null +++ b/checkpoint-40000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7bf630212b7afefc0d17054bd8909fc3c85add2126125ef288888630ec559d2 +size 15627 diff --git a/checkpoint-40000/scaler.pt b/checkpoint-40000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ca7dfbd31a534c5819bd61009439fcc486cdc8a8 --- /dev/null +++ b/checkpoint-40000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5893a144fe381d6d669cde20146a9de16a1063230a91b61c2e9512c066f6a386 +size 557 diff --git a/checkpoint-40000/scheduler.bin b/checkpoint-40000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..ca4152e8e0f9145df33da8b3c1b3d0788540506e --- /dev/null +++ b/checkpoint-40000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83b4880f7dc713542e0a6fb6d68d48ed1185303dab25b043fda15066b2f9d813 +size 563 diff --git a/checkpoint-50000/controlnet/config.json b/checkpoint-50000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-50000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cc41b73ded62c43b6ee9ef7dde0358f9a72c1c43 --- /dev/null +++ b/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4181e2af76453895247585c41fb662c32b1fe44c14b4d7ce09fd03bbb34ff661 +size 1445157120 diff --git a/checkpoint-50000/optimizer.bin b/checkpoint-50000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..9dda1c1579b277db494e0cba3567dd2b465cf9bd --- /dev/null +++ b/checkpoint-50000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c1d0a03ad58a7877309fc0401e3ffbe061709b04814987974487a8361b42614 +size 2890518041 diff --git a/checkpoint-50000/random_states_0.pkl b/checkpoint-50000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..f053d5bd5ce72d225e89043920376a37e54ddc1a --- /dev/null +++ b/checkpoint-50000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb17d6c2b4b8ec26fdf2d6103234fdf70d416a4a8244e0ba814fe08a003b7e87 +size 15691 diff --git a/checkpoint-50000/scaler.pt b/checkpoint-50000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7a6c22180393d1e60054fa97ac8176a9fb1b841 --- /dev/null +++ b/checkpoint-50000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfa439dae4911bc42e50303d91370d016f1abea5b6ea2987e21979d6c9de5108 +size 557 diff --git a/checkpoint-50000/scheduler.bin b/checkpoint-50000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..e49971e359cb48e005b6c88fb05a9e49daab3e0b --- /dev/null +++ b/checkpoint-50000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bdd90074fe223aac2ce4ea30712cb637ec3a88750609b5185e44c3960989909f +size 563 diff --git a/checkpoint-60000/controlnet/config.json b/checkpoint-60000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-60000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-60000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-60000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2a79602be1ef3fa9c2444e7094c1169d677028cc --- /dev/null +++ b/checkpoint-60000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:744e168a57957ff2c4fe382bdabfd2a625d8ad13956b6b0208cfe4545f29c8bb +size 1445157120 diff --git a/checkpoint-60000/optimizer.bin b/checkpoint-60000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..33bb25fe0767d66b5c513379aa6e7e69d72c2bdf --- /dev/null +++ b/checkpoint-60000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c32151df0e0177e70b409d973b25c263c405a0c689265794439acde962400506 +size 2890518041 diff --git a/checkpoint-60000/random_states_0.pkl b/checkpoint-60000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..a828a4551c8c5c5663b2e03d563235bf155570fd --- /dev/null +++ b/checkpoint-60000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff95720fcab3fb0b189e77e7fd945dbf02c7cf441cb6d9bbce4c7da7fd5c2607 +size 15691 diff --git a/checkpoint-60000/scaler.pt b/checkpoint-60000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ea1385f0968a37266577f1fb719439797d21bb9b --- /dev/null +++ b/checkpoint-60000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b380a1a2ff2b31c17eccfb39037fa0a1d4d97c8f3ac319c5a8e825a011d8f45 +size 557 diff --git a/checkpoint-60000/scheduler.bin b/checkpoint-60000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..0ded1ae72a7ed5a2532663ccaf2e212d1b1361c3 --- /dev/null +++ b/checkpoint-60000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c82275293fb5ba57ee99ba0ee07f100f4c872636c9624fb116c7e582218c49d +size 563 diff --git a/checkpoint-70000/controlnet/config.json b/checkpoint-70000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-70000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-70000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-70000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..551f4a61ee4130b6e072a8c9f5292c13ec962b82 --- /dev/null +++ b/checkpoint-70000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4c0bb151aac527ed6f5a8180687bb09e45ea6084854864f0e2033fc96232a80 +size 1445157120 diff --git a/checkpoint-70000/optimizer.bin b/checkpoint-70000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..5df403988c106ec22ce1f3951f52d90f9ed1bb6a --- /dev/null +++ b/checkpoint-70000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61beb8e2f2042180e14683121bffe7d4fdaa0507a07d450358378905c1cf3d3e +size 2890518041 diff --git a/checkpoint-70000/random_states_0.pkl b/checkpoint-70000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..4c098e2b481de5881ae8782211fd3da96521ad58 --- /dev/null +++ b/checkpoint-70000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f554b6a9749d4a46eefbf6683e13f074ca93a5d16a35b371c80a6bba58a3b84 +size 15691 diff --git a/checkpoint-70000/scaler.pt b/checkpoint-70000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..df45114deccbf85b7f3fd81a6615dc69a906e3b7 --- /dev/null +++ b/checkpoint-70000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e647841d81db524d6cc2b5ac8774b80ce93dc54dd0b70aea5b8b521b128b715 +size 557 diff --git a/checkpoint-70000/scheduler.bin b/checkpoint-70000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..683afed12a54814a0f3a614c34c10e718d076a9f --- /dev/null +++ b/checkpoint-70000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcffbb8d94f1754b42d6ceab8c93161b32689a4077ba8e9c4a4530df14658beb +size 563 diff --git a/checkpoint-80000/controlnet/config.json b/checkpoint-80000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-80000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-80000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-80000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c8a65beb9ea47258711e009eb8b7fd83a6b2a01e --- /dev/null +++ b/checkpoint-80000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f198d3019a466f87901a2844e30fc0c2bd89525339b11880cac125ac9b162fc4 +size 1445157120 diff --git a/checkpoint-80000/optimizer.bin b/checkpoint-80000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..81b141b6970b0a8abbbe654937edac034c6a3dd9 --- /dev/null +++ b/checkpoint-80000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a88645bbe486dee22377674875a57c5ec509ebebd332727921d085af343f0b62 +size 2890518041 diff --git a/checkpoint-80000/random_states_0.pkl b/checkpoint-80000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..e5cee66978ff7287879f3b6d9947845c2dae4287 --- /dev/null +++ b/checkpoint-80000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f24182e362859b6a87ba347f1d36661a0226851ab120bf8aa130f91810b6afd +size 15691 diff --git a/checkpoint-80000/scaler.pt b/checkpoint-80000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..f78e3179ec263d564874f85a3ae43ea3513dee30 --- /dev/null +++ b/checkpoint-80000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f0ce2dbedb34844b799bc6bccde865103b173519439671b90e18cc60f64a0ce +size 557 diff --git a/checkpoint-80000/scheduler.bin b/checkpoint-80000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..90c91defbc62bd1e11fc6176d8d90449d41af543 --- /dev/null +++ b/checkpoint-80000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b212a983dd8da62507afde845b289216e4b866c07ebf9e18716a2fa357728516 +size 563 diff --git a/checkpoint-90000/controlnet/config.json b/checkpoint-90000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /dev/null +++ b/checkpoint-90000/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": null, + "use_linear_projection": false +} diff --git a/checkpoint-90000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-90000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cb7b4b392f7ed44823379503e7f0ef4a17de2226 --- /dev/null +++ b/checkpoint-90000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96b50bdba51071a30c5a1f901d5f0fd75ba262e290179735614fa08abd17020b +size 1445157120 diff --git a/checkpoint-90000/optimizer.bin b/checkpoint-90000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..2caf6a9e6ac2d42a6c428eaf0251be15e5688395 --- /dev/null +++ b/checkpoint-90000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa62fa0a98d382f1fa7a14a68e1f4cb4b639be6b5bbf4d45d6c4f10817fca05a +size 2890518041 diff --git a/checkpoint-90000/random_states_0.pkl b/checkpoint-90000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..b9c10c0c6cd59a3a850f625db7ba5ac6ece18a46 --- /dev/null +++ b/checkpoint-90000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0428fd2aafef272098111e74b072094821688f4cd5068fd96ee87d6451aa718d +size 15627 diff --git a/checkpoint-90000/scaler.pt b/checkpoint-90000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7376ba455065efef802192b4c26fdd43fa6bc79 --- /dev/null +++ b/checkpoint-90000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcd834b275ffa79cbf37efaffa7830164d6064ade5750b48808a79721ff3aea2 +size 557 diff --git a/checkpoint-90000/scheduler.bin b/checkpoint-90000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..fefcb508e0d1294dc8e0d7b4d1c581d19af2306c --- /dev/null +++ b/checkpoint-90000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9ed2ca67b10bc526be88d3b6fa68042663841062ccf1e1598fa1ca3203a584d +size 563 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ebf4218829d788bb3480905af21c32fea60efd43 --- /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": null, + "use_linear_projection": false +} diff --git a/diffusion_pytorch_model.safetensors b/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..99739256a3d8f5215bd221d7ef3df0eaf368a727 --- /dev/null +++ b/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66ff4d8352762ba8be32f27a00210144e04f73d978092e8363a520fbe09e0f96 +size 1445157120 diff --git a/image_control.png b/image_control.png new file mode 100644 index 0000000000000000000000000000000000000000..1d31b90c0476693982213aa90f73bd31bf9496ad 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..260b3659a8bbe6d7b6aaeddcae2ffd316d516947 --- /dev/null +++ b/images_0.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10746525429ac7bf4ee66109780cdc6856d247dd85523593ea342050b62643f7 +size 2044921 diff --git a/images_1.png b/images_1.png new file mode 100644 index 0000000000000000000000000000000000000000..d6f5769e481f823ec62acdf6cf92102bd5744382 --- /dev/null +++ b/images_1.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9729cda4e55e892f3f6731739dcf7086e959409105fbc4b1b2c2f8e0273cac1c +size 1544300