diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..d6cd0731891cf0a93eb9ad60749912bc6cff4104 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,13 @@ 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 +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 +images_5.png filter=lfs diff=lfs merge=lfs -text +images_6.png filter=lfs diff=lfs merge=lfs -text +images_7.png filter=lfs diff=lfs merge=lfs -text +images_8.png filter=lfs diff=lfs merge=lfs -text +images_9.png filter=lfs diff=lfs merge=lfs -text diff --git a/.ipynb_checkpoints/README-checkpoint.md b/.ipynb_checkpoints/README-checkpoint.md new file mode 100644 index 0000000000000000000000000000000000000000..38072239dbe210aeb3c0bae85783cc6c6056ee58 --- /dev/null +++ b/.ipynb_checkpoints/README-checkpoint.md @@ -0,0 +1,62 @@ +--- +library_name: diffusers +license: openrail++ +inference: true +tags: +- stable-diffusion-xl +- stable-diffusion-xl-diffusers +- text-to-image +- diffusers +- controlnet +- diffusers-training +--- + +base_model: waiNSFWIllustrious_v80_VRoidstyle + + + + +# controlnet-Visual-Bank/CN_ControlBody_SD_v1_diffusers_train_v1 + +These are controlnet weights trained on /cn_train/waiNSFWIllustrious_v80_VRoidstyle with new type of conditioning. +You can find some example images below. + +prompt: 3D, 1girl, solo, white background, simple background, green shorts, from behind, green pants, brown hair, shoes, sweater, long sleeves, red sweater, short hair, black footwear, shorts, hands up, pants, standing on one leg, facing away, standing +![images_0)](./images_0.png) +prompt: 3D, 1girl, solo, brown eyes, brown hair, white background, bangs, simple background, open mouth, chibi, upper teeth only, full body, long sleeves, pants, white footwear, teeth, green pants, green shorts, smile, shirt, shoes, blunt bangs, sweater, shorts +![images_1)](./images_1.png) +prompt: 3D, 1girl, solo, brown hair, brown eyes, white background, simple background, standing on one leg, green pants, pants, sandals, standing, full body, long sleeves, open mouth, shirt, sweater, looking at viewer, bangs, upper teeth only, teeth, green shorts, white footwear +![images_2)](./images_2.png) +prompt: 3D, 1girl, solo, blue hair, long hair, chibi, white background, black footwear, purple eyes, boots, bangs, necktie, sitting, black necktie, twintails, simple background, from side, long sleeves, black pants, pantyhose, pants, full body, shirt, low twintails +![images_3)](./images_3.png) +prompt: 3D, 1girl, solo, long hair, green eyes, skirt, brown hair, white background, shirt, black skirt, simple background, black footwear, white shirt, long sleeves, full body, standing, profile, shoes +![images_4)](./images_4.png) +prompt: 3D, 1girl, long hair, solo, white background, brown hair, very long hair, simple background, facing away, long sleeves, sandals, from behind, squatting, dress, shirt, full body, black dress, black skirt, standing +![images_5)](./images_5.png) +prompt: 3D, 1girl, solo, green eyes, long hair, skirt, brown hair, white background, black skirt, full body, shirt, white shirt, simple background, pointing, sandals, long sleeves, white footwear, bangs, crop top +![images_6)](./images_6.png) +prompt: 3D, 1girl, solo, shirt, green hair, pants, white shirt, plaid, plaid pants, glasses, bow, white background, full body, simple background, shoes, bowtie, open mouth, brown footwear, short sleeves, bangs, collared shirt, hands up, purple eyes, black bow, standing on one leg, standing, short hair, smile, looking at viewer, red-framed eyewear, green eyes, dress shirt, school uniform +![images_7)](./images_7.png) +prompt: 3D, 1girl, solo, green hair, plaid pants, glasses, pants, shirt, chibi, white shirt, plaid, white background, purple eyes, green pants, looking at viewer, full body, simple background, brown footwear, shoes, arms up, bangs, closed mouth, kneeling +![images_8)](./images_8.png) +prompt: 3D, 1girl, solo, green hair, shirt, pants, white shirt, plaid pants, plaid, glasses, short hair, white background, bow, bowtie, full body, shoes, simple background, short sleeves, gumi, arm up, standing, brown footwear, green pants, smile, bangs, closed mouth +![images_9)](./images_9.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/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..38072239dbe210aeb3c0bae85783cc6c6056ee58 --- /dev/null +++ b/README.md @@ -0,0 +1,62 @@ +--- +library_name: diffusers +license: openrail++ +inference: true +tags: +- stable-diffusion-xl +- stable-diffusion-xl-diffusers +- text-to-image +- diffusers +- controlnet +- diffusers-training +--- + +base_model: waiNSFWIllustrious_v80_VRoidstyle + + + + +# controlnet-Visual-Bank/CN_ControlBody_SD_v1_diffusers_train_v1 + +These are controlnet weights trained on /cn_train/waiNSFWIllustrious_v80_VRoidstyle with new type of conditioning. +You can find some example images below. + +prompt: 3D, 1girl, solo, white background, simple background, green shorts, from behind, green pants, brown hair, shoes, sweater, long sleeves, red sweater, short hair, black footwear, shorts, hands up, pants, standing on one leg, facing away, standing +![images_0)](./images_0.png) +prompt: 3D, 1girl, solo, brown eyes, brown hair, white background, bangs, simple background, open mouth, chibi, upper teeth only, full body, long sleeves, pants, white footwear, teeth, green pants, green shorts, smile, shirt, shoes, blunt bangs, sweater, shorts +![images_1)](./images_1.png) +prompt: 3D, 1girl, solo, brown hair, brown eyes, white background, simple background, standing on one leg, green pants, pants, sandals, standing, full body, long sleeves, open mouth, shirt, sweater, looking at viewer, bangs, upper teeth only, teeth, green shorts, white footwear +![images_2)](./images_2.png) +prompt: 3D, 1girl, solo, blue hair, long hair, chibi, white background, black footwear, purple eyes, boots, bangs, necktie, sitting, black necktie, twintails, simple background, from side, long sleeves, black pants, pantyhose, pants, full body, shirt, low twintails +![images_3)](./images_3.png) +prompt: 3D, 1girl, solo, long hair, green eyes, skirt, brown hair, white background, shirt, black skirt, simple background, black footwear, white shirt, long sleeves, full body, standing, profile, shoes +![images_4)](./images_4.png) +prompt: 3D, 1girl, long hair, solo, white background, brown hair, very long hair, simple background, facing away, long sleeves, sandals, from behind, squatting, dress, shirt, full body, black dress, black skirt, standing +![images_5)](./images_5.png) +prompt: 3D, 1girl, solo, green eyes, long hair, skirt, brown hair, white background, black skirt, full body, shirt, white shirt, simple background, pointing, sandals, long sleeves, white footwear, bangs, crop top +![images_6)](./images_6.png) +prompt: 3D, 1girl, solo, shirt, green hair, pants, white shirt, plaid, plaid pants, glasses, bow, white background, full body, simple background, shoes, bowtie, open mouth, brown footwear, short sleeves, bangs, collared shirt, hands up, purple eyes, black bow, standing on one leg, standing, short hair, smile, looking at viewer, red-framed eyewear, green eyes, dress shirt, school uniform +![images_7)](./images_7.png) +prompt: 3D, 1girl, solo, green hair, plaid pants, glasses, pants, shirt, chibi, white shirt, plaid, white background, purple eyes, green pants, looking at viewer, full body, simple background, brown footwear, shoes, arms up, bangs, closed mouth, kneeling +![images_8)](./images_8.png) +prompt: 3D, 1girl, solo, green hair, shirt, pants, white shirt, plaid pants, plaid, glasses, short hair, white background, bow, bowtie, full body, shoes, simple background, short sleeves, gumi, arm up, standing, brown footwear, green pants, smile, bangs, closed mouth +![images_9)](./images_9.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..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-10000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..89b3c320f5b30ad1ae2fa82684176e35005ae001 --- /dev/null +++ b/checkpoint-10000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94c447fd9f8368bfc12f24479b55e1e54b516f3db93f52d24602bb06e0bc62bb +size 5004167864 diff --git a/checkpoint-10000/optimizer.bin b/checkpoint-10000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..0067db165b471f90839507d3697e7b2588dbc967 --- /dev/null +++ b/checkpoint-10000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f6b55367bc5b8787858f06346338213e71526f5d766c599c9ae0fb50b3cf20e +size 10008841510 diff --git a/checkpoint-10000/random_states_0.pkl b/checkpoint-10000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..0ff25e2d63b856799c19e4aa5cd1fad6e5d5fc59 --- /dev/null +++ b/checkpoint-10000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ee9253b733104b7558460e4dd018c8c53aa00b7cb212bc4a7618a819afdab03 +size 14344 diff --git a/checkpoint-10000/scheduler.bin b/checkpoint-10000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..62171498f3fc3798d395650e016e3a8a63d8c1c7 --- /dev/null +++ b/checkpoint-10000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f521328d01649fbd5f2dd87990339cd73e72f126e734b8cfd90598c1181069e2 +size 1000 diff --git a/checkpoint-15000/controlnet/config.json b/checkpoint-15000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-15000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a4a95e865437932768b7248abf9ce77312f93e85 --- /dev/null +++ b/checkpoint-15000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:174c86d26fc98580d8011c3a43a08a440921ee690877265d49d873a210393722 +size 5004167864 diff --git a/checkpoint-15000/optimizer.bin b/checkpoint-15000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..c18754d0a03d2743be0ddcaf5f692a284b6e787c --- /dev/null +++ b/checkpoint-15000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35edfd5dd1c3bd96e9fe69a0aa81f9ff145482c2c53c23cbb1f56ee0fe7a91e1 +size 10008841510 diff --git a/checkpoint-15000/random_states_0.pkl b/checkpoint-15000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..1bd2fc1696e0f4a649f7aae9a1f3416fe5682318 --- /dev/null +++ b/checkpoint-15000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b23ba419676d094cba98b9adc2d6d7dd8d7f8704b760b8bebf2c98fbb9573735 +size 14344 diff --git a/checkpoint-15000/scheduler.bin b/checkpoint-15000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..d62c2e9a21c8b32753f377583fd71df4bd0b2db1 --- /dev/null +++ b/checkpoint-15000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7590cb441f3094aed2a51fdf90eed4667df80938a3b08297706789ad8895e4d9 +size 1000 diff --git a/checkpoint-20000/controlnet/config.json b/checkpoint-20000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-20000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2f092f032e9dac2f54b0b7e881ef41649b7f7d45 --- /dev/null +++ b/checkpoint-20000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b851955d228107c742bac84e9451b6292a0b7936db8c328b72e468e1a1113613 +size 5004167864 diff --git a/checkpoint-20000/optimizer.bin b/checkpoint-20000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..22b2ec99787752cc4369345f1a149cf38466fdad --- /dev/null +++ b/checkpoint-20000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcd8cbe358808a56633f25ac8eb590480f98aa9b204196faeac95f915ea72047 +size 10008841510 diff --git a/checkpoint-20000/random_states_0.pkl b/checkpoint-20000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..0ddb63c1f6f01e3fe34b89a9adc8fa545a56833c --- /dev/null +++ b/checkpoint-20000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3253df3d43a3863c13bc9b6c9072f25fb32d0363195c943565f3a68354899fe +size 14344 diff --git a/checkpoint-20000/scheduler.bin b/checkpoint-20000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..d74988761e14c5dde3a1236b54e09e73769bf3d2 --- /dev/null +++ b/checkpoint-20000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a81ff9d6b2bbeb0c6c3c98580999952ee828678240d2891b3412c0b4e02c4c2b +size 1000 diff --git a/checkpoint-25000/controlnet/config.json b/checkpoint-25000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-25000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-25000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-25000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ae4768d1d04412d70933e2406df3f955d2a159bc --- /dev/null +++ b/checkpoint-25000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2950d6519cf5d14e765168dbc0161cee617ab78195136c2456cdec8e163cdd8 +size 5004167864 diff --git a/checkpoint-25000/optimizer.bin b/checkpoint-25000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..1bfae51bd11f6a1cae49be1a21ba860b35def017 --- /dev/null +++ b/checkpoint-25000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:881ead4af89f7a9206e69891e4d8e898780065a105a958d595959dd48d788b31 +size 10008841510 diff --git a/checkpoint-25000/random_states_0.pkl b/checkpoint-25000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..12794a00199771cc83ca556a448b49b8abba9ee5 --- /dev/null +++ b/checkpoint-25000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d66b04b6a21641c90a5ace83613c22eaf00643a6973dfb761becd2a21e387961 +size 14344 diff --git a/checkpoint-25000/scheduler.bin b/checkpoint-25000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..7e194b2dd73e49fd8f43482be95851151e33bce2 --- /dev/null +++ b/checkpoint-25000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac2bba63a3285cbf3b0e07fe2f8b27620281ac9240269169f3997bd54890720c +size 1000 diff --git a/checkpoint-30000/controlnet/config.json b/checkpoint-30000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-30000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..887d8061dfff2208314e77de807d6f7bceff5873 --- /dev/null +++ b/checkpoint-30000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44dd69334c4590fd04ed538306a583a43f7d8d5404c1d2bfdec74dfb89efba6d +size 5004167864 diff --git a/checkpoint-30000/optimizer.bin b/checkpoint-30000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..e63cb76057cfb36736f1a48e57f5655d2198fea0 --- /dev/null +++ b/checkpoint-30000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65ea4f43b1cc1f6be13d11aa582fb006653f1141d393d2abc072ef92182244f2 +size 10008841510 diff --git a/checkpoint-30000/random_states_0.pkl b/checkpoint-30000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..5cf106c43ee0fe89e26b21f1f531538ba9c2b63b --- /dev/null +++ b/checkpoint-30000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6288165fcd2679961bdb81105ac642dc5c2bfa504ee5e935fc15b28ad170389d +size 14344 diff --git a/checkpoint-30000/scheduler.bin b/checkpoint-30000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..561c8024d66589200d9c89abcc42b954af012e38 --- /dev/null +++ b/checkpoint-30000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:264c61967c5683c11d783c73aceadb4e26d49cf98d971575e68ae8346da147df +size 1000 diff --git a/checkpoint-35000/controlnet/config.json b/checkpoint-35000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-35000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-35000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-35000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3ba0e26fa09c6b762dc4a375a10a975e7d1f7fad --- /dev/null +++ b/checkpoint-35000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7806ec60ee821f003ab5197d31cdb3ee727683d86d832e74478b805830a71101 +size 5004167864 diff --git a/checkpoint-35000/optimizer.bin b/checkpoint-35000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..34b924b8d2b4a7c5e494984c9405ab908d089352 --- /dev/null +++ b/checkpoint-35000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ec699ea416915df001e67fbb007b5e4bd4dde27b5cf9ff6e2aceed00f0bc07a +size 10008841510 diff --git a/checkpoint-35000/random_states_0.pkl b/checkpoint-35000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..cd3949fec6dfa40a399aeea9e388e3e852865316 --- /dev/null +++ b/checkpoint-35000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fd02d425fed2fe49a1cd062d768864da97a545b16a51d77408d1ed3aaebe221 +size 14344 diff --git a/checkpoint-35000/scheduler.bin b/checkpoint-35000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..f2abe8cd38c740bfc24c76867381fbea76b37c80 --- /dev/null +++ b/checkpoint-35000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad5305dfcd2d8a8d911b8f29c4a709ae96d2ed2755fe5da4a758f6c2711faf78 +size 1000 diff --git a/checkpoint-40000/controlnet/config.json b/checkpoint-40000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-40000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d8f0086febe8c4a68a5e8d4b51f0575bc5470b94 --- /dev/null +++ b/checkpoint-40000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:105055c529fbcd296ea2ec7eb7b136609ed119ef8b3afa8ee1ad478f906f9eae +size 5004167864 diff --git a/checkpoint-40000/optimizer.bin b/checkpoint-40000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..e3f688de6572d0da1ca53aa5e421d633c8cbde5b --- /dev/null +++ b/checkpoint-40000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd065c906d82ae5d43d357b8a2f28c459d3df0dc1c0e74f976abf29a665082c0 +size 10008841510 diff --git a/checkpoint-40000/random_states_0.pkl b/checkpoint-40000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..1fe1c3f0322f7a86e21133d5102a5ddfbacce0dc --- /dev/null +++ b/checkpoint-40000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c4091635c2b228aaead526afd387598c1ea5c108d60142f25af1df8d14b895a +size 14344 diff --git a/checkpoint-40000/scheduler.bin b/checkpoint-40000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..2b5911d1ec25c0339115e0809bdc1e64195cb056 --- /dev/null +++ b/checkpoint-40000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37bbc55e0c7f879268b391c09aa55d50369cd4d5e36607772200b83e0f19ff19 +size 1000 diff --git a/checkpoint-45000/controlnet/config.json b/checkpoint-45000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-45000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-45000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-45000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6cfb939d849eb5fced59dfc636f389ef773c3a9a --- /dev/null +++ b/checkpoint-45000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f305c4b5bef6ac0e81964c93b79a576546bc0d4822a64440d5a2235d242d8ff7 +size 5004167864 diff --git a/checkpoint-45000/optimizer.bin b/checkpoint-45000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..0665d6b3496783ba0d0405a944d18f400e02d1ad --- /dev/null +++ b/checkpoint-45000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b406175e413377189ebf226be843636e959f2600f13a56759bc5a8a7b113625 +size 10008841510 diff --git a/checkpoint-45000/random_states_0.pkl b/checkpoint-45000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..499c254a40af754dfa3f224f459a554ebcb0e765 --- /dev/null +++ b/checkpoint-45000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bde766279222ce566245385f6f92e623cad9baadc2cbc2cb5895c812c339a9e +size 14344 diff --git a/checkpoint-45000/scheduler.bin b/checkpoint-45000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..4ccac498e51f0fe288aa6f33e1dc58ec82ab843f --- /dev/null +++ b/checkpoint-45000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:055bf3e8c7f4b620b1f9723db9fb2657c737caab9bb56fd4607b04f5a49b186a +size 1000 diff --git a/checkpoint-5000/controlnet/config.json b/checkpoint-5000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-5000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "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..2b888c13b04c4b66fcf726bc4a877104d69b85ac --- /dev/null +++ b/checkpoint-5000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6443d6b05382d22b6a91175b1bb2a43d1dc60cb5d9641899904eec791175363 +size 5004167864 diff --git a/checkpoint-5000/optimizer.bin b/checkpoint-5000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..4a82d5e4344f90651e6e02ff3b2993a2998191cf --- /dev/null +++ b/checkpoint-5000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59b34a8d9a1024eebe7ece8bd0138b67c45987def00167f13ffc76a8587d5ddf +size 10008841510 diff --git a/checkpoint-5000/random_states_0.pkl b/checkpoint-5000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..e978af49d043a2422c2f1828190f3e6841d8268e --- /dev/null +++ b/checkpoint-5000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:692a31b34e9cff79a8efeb471842f40f540f7c8ac171da0c15bc26dbbdc02984 +size 14344 diff --git a/checkpoint-5000/scheduler.bin b/checkpoint-5000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b33560b2b640d6ae8c1ae2635506d4ed943b476 --- /dev/null +++ b/checkpoint-5000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d349ef653c04614ca5d0690e9b2ec58287a2df6227cd08ef013b175365d41f2 +size 1000 diff --git a/checkpoint-50000/controlnet/config.json b/checkpoint-50000/controlnet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/checkpoint-50000/controlnet/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "upcast_attention": false, + "use_linear_projection": true +} diff --git a/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors b/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..feb2b2d7ad939bdac141e48fd861156ff8a90902 --- /dev/null +++ b/checkpoint-50000/controlnet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74a211ecced9905f74b4f2400a54e8e8b8517074e32edf062c8a878a21c9d82b +size 5004167864 diff --git a/checkpoint-50000/optimizer.bin b/checkpoint-50000/optimizer.bin new file mode 100644 index 0000000000000000000000000000000000000000..7fbf4a94997af51049f221efadd807d9bd048ec1 --- /dev/null +++ b/checkpoint-50000/optimizer.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e6eb3bfc801d4f45da389c0a6d35e1610c6748e8fb666bd4d0d957463fd7ce6 +size 10008841510 diff --git a/checkpoint-50000/random_states_0.pkl b/checkpoint-50000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..cd02163cde3b2b89b900e187285616fd80074342 --- /dev/null +++ b/checkpoint-50000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b38fe6fe1fd10c8f89eb783f28b7ac69c70699cfcd85c00a899291807eeb0b6e +size 14344 diff --git a/checkpoint-50000/scheduler.bin b/checkpoint-50000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..e0e88ca7c9b09ae5c420449dfce891c4b09ae62c --- /dev/null +++ b/checkpoint-50000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c71d3e0d0229fa013b22250308b11b6535aad0bc6202d8f6a0d3023c6ea2a484 +size 1000 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8839e3cae1ba35480936ebbb607239bf2917970 --- /dev/null +++ b/config.json @@ -0,0 +1,57 @@ +{ + "_class_name": "ControlNetModel", + "_diffusers_version": "0.33.0.dev0", + "act_fn": "silu", + "addition_embed_type": "text_time", + "addition_embed_type_num_heads": 64, + "addition_time_embed_dim": 256, + "attention_head_dim": [ + 5, + 10, + 20 + ], + "block_out_channels": [ + 320, + 640, + 1280 + ], + "class_embed_type": null, + "conditioning_channels": 3, + "conditioning_embedding_out_channels": [ + 16, + 32, + 96, + 256 + ], + "controlnet_conditioning_channel_order": "rgb", + "cross_attention_dim": 2048, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "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": 2816, + "resnet_time_scale_shift": "default", + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "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..feb2b2d7ad939bdac141e48fd861156ff8a90902 --- /dev/null +++ b/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74a211ecced9905f74b4f2400a54e8e8b8517074e32edf062c8a878a21c9d82b +size 5004167864 diff --git a/image_control.png b/image_control.png new file mode 100644 index 0000000000000000000000000000000000000000..c1e1f98268ad1d0bace569afeeb327049cdca8b6 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..e8c07da3cb27972ed84c573c7221374a7765610a --- /dev/null +++ b/images_0.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:951af1163383c9bc5815bb4f61ee93434c5dc555dd6ee688aa4a9b847bbd9041 +size 2140065 diff --git a/images_1.png b/images_1.png new file mode 100644 index 0000000000000000000000000000000000000000..fe80d981f33f626d3cc19e8b86d05e92166499fb --- /dev/null +++ b/images_1.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cda9fa0ae34dfb6908e90cd3157292438bb9bf977bd253ef589ddd4120a3882f +size 703320 diff --git a/images_2.png b/images_2.png new file mode 100644 index 0000000000000000000000000000000000000000..df611639aade5a8fd8644bdf5785f9f3d43194ad --- /dev/null +++ b/images_2.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a030923d61f3cc6201df64b9fea50fbae32bead5f6b9ef74e5e6d1e48e6c4fa4 +size 888904 diff --git a/images_3.png b/images_3.png new file mode 100644 index 0000000000000000000000000000000000000000..9173be31a9bbb874f3a16c30967382056ce45f03 --- /dev/null +++ b/images_3.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ba22a6a92b267c43d29847c2a5561fdb583574fb89bd74e9b993058437082bf +size 1829269 diff --git a/images_4.png b/images_4.png new file mode 100644 index 0000000000000000000000000000000000000000..4254252abe9269696e64ad10aeea647942d6e130 --- /dev/null +++ b/images_4.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8972da0da0cd84c61d77a9c287308a849f1108d7f852de6999b8673dc89d1f32 +size 905002 diff --git a/images_5.png b/images_5.png new file mode 100644 index 0000000000000000000000000000000000000000..3d476ce9198aaa17f35fb9743baabbfd5f7a9460 --- /dev/null +++ b/images_5.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91816ed9e06326789d43930bb60d0e19466a439c38c8d0ef897180af136061d3 +size 901810 diff --git a/images_6.png b/images_6.png new file mode 100644 index 0000000000000000000000000000000000000000..b0d82f6a70db2102f956cb3590c86146f1173cef --- /dev/null +++ b/images_6.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27f6ac4830a6ab6aaeee73d45b01e32c095cd361750e7b5784a2fd008c50167f +size 2220185 diff --git a/images_7.png b/images_7.png new file mode 100644 index 0000000000000000000000000000000000000000..66de57e59a8ee1200857ce29e4699ffd2ca6d1d8 --- /dev/null +++ b/images_7.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f32dcfe7ec9b4a2c791803a95fbb560cf6322fffb2bc8874210a4468e583c3aa +size 971678 diff --git a/images_8.png b/images_8.png new file mode 100644 index 0000000000000000000000000000000000000000..f47a7d2ee573eff88a23e621f46697e1f67689d3 --- /dev/null +++ b/images_8.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4717da7e83903b92a1eaae098321a0088aaee78ef8c002b519676352c9eda37 +size 1025306 diff --git a/images_9.png b/images_9.png new file mode 100644 index 0000000000000000000000000000000000000000..ed520d3cd4027b84e84615b8a555ce1d99c55e39 --- /dev/null +++ b/images_9.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9898293912709884ab0f23e43ef6721f2c8f3133b5709d0fa3689a6fe32ce2a6 +size 933496 diff --git a/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.4110186/events.out.tfevents.1744306934.03c963ec07e4.1179.1 b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.4110186/events.out.tfevents.1744306934.03c963ec07e4.1179.1 new file mode 100644 index 0000000000000000000000000000000000000000..3f5bbe2b4080a34a9a82a76c9214196235cd8e4b --- /dev/null +++ b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.4110186/events.out.tfevents.1744306934.03c963ec07e4.1179.1 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49af7b6b2701f2e3dc2c84ef4aebb2a7068c6bd0b2730ccf481045d5b094143f +size 2849 diff --git a/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.412721/hparams.yml b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.412721/hparams.yml new file mode 100644 index 0000000000000000000000000000000000000000..1e2813c75436b7670fea1f1dbb5a846b0ca49e19 --- /dev/null +++ b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/1744306934.412721/hparams.yml @@ -0,0 +1,54 @@ +adam_beta1: 0.9 +adam_beta2: 0.999 +adam_epsilon: 1.0e-08 +adam_weight_decay: 0.01 +allow_tf32: false +cache_dir: /workspace/cn_output/cache +caption_column: text +checkpointing_steps: 5000 +checkpoints_total_limit: null +conditioning_image_column: conditioning_image +controlnet_model_name_or_path: null +crops_coords_top_left_h: 0 +crops_coords_top_left_w: 0 +dataloader_num_workers: 0 +dataset_config_name: null +dataset_name: Visual-Bank/CN_ControlBody_SD_v1_diffusers +enable_npu_flash_attention: false +enable_xformers_memory_efficient_attention: true +gradient_accumulation_steps: 1 +gradient_checkpointing: false +hub_model_id: Visual-Bank/CN_ControlBody_SD_v1_diffusers_train_v1 +hub_token: null +image_column: image +learning_rate: 1.0e-05 +logging_dir: logs +lr_num_cycles: 1 +lr_power: 1.0 +lr_scheduler: cosine +lr_warmup_steps: 500 +max_grad_norm: 1.0 +max_train_samples: null +max_train_steps: 50000 +mixed_precision: null +num_train_epochs: 4 +num_validation_images: 2 +output_dir: /workspace/cn_output/2025-04-11 +pretrained_model_name_or_path: /cn_train/waiNSFWIllustrious_v80_VRoidstyle +pretrained_vae_model_name_or_path: madebyollin/sdxl-vae-fp16-fix +proportion_empty_prompts: 0 +push_to_hub: true +report_to: tensorboard +resolution: 1024 +resume_from_checkpoint: null +revision: null +scale_lr: false +seed: 42 +set_grads_to_none: false +tokenizer_name: null +tracker_project_name: CN_ControlBody_SD_v1_diffusers_train_v1 +train_batch_size: 2 +train_data_dir: null +use_8bit_adam: false +validation_steps: 1000 +variant: null diff --git a/logs/CN_ControlBody_SD_v1_diffusers_train_v1/events.out.tfevents.1744306934.03c963ec07e4.1179.0 b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/events.out.tfevents.1744306934.03c963ec07e4.1179.0 new file mode 100644 index 0000000000000000000000000000000000000000..5faffc148a37e8fa5520d9ea2612ef53889d80ce --- /dev/null +++ b/logs/CN_ControlBody_SD_v1_diffusers_train_v1/events.out.tfevents.1744306934.03c963ec07e4.1179.0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7061ded1bbbf6842fa5fe9afaae142b792e4d967739fe59be2fe89b866d3bc84 +size 761612510