diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..ccc6ec75575b8d7fdfd55a1c9b7e9766f3140f2e --- /dev/null +++ b/README.md @@ -0,0 +1,2 @@ + +More information on all the CLI arguments and the environment are available on your [`wandb` run page](https://wandb.ai/sayakpaul/diffusion-orpo-lora-sdxl/runs/mz6nhlqy). diff --git a/checkpoint-10000/random_states_0.pkl b/checkpoint-10000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..f3282c2ac54b54375a2cadced931553e03f6761d --- /dev/null +++ b/checkpoint-10000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e98b13243e4780a85f9c4abaa09a09262aac1acd9cae50df799419468f273047 +size 16036 diff --git a/checkpoint-10000/scaler.pt b/checkpoint-10000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cd2b3369a707005cefcc0647df96a02cc8f16d2 --- /dev/null +++ b/checkpoint-10000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1cef5f438f202bf558e57f4f3680c2a1840a613bb001b280c74a505963c4f378 +size 988 diff --git a/checkpoint-10000/scheduler.bin b/checkpoint-10000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..5aaa470f2207f9d8976656a91358b09a2cd26cbf --- /dev/null +++ b/checkpoint-10000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:058b3567396721c95682f3f854f20f26834cd8a8331b207e94c1d1639a736213 +size 1000 diff --git a/checkpoint-10000/unet/config.json b/checkpoint-10000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-10000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-10000/unet/diffusion_pytorch_model.safetensors b/checkpoint-10000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..782317df840837a5ebd7e0714555e64113636986 --- /dev/null +++ b/checkpoint-10000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc57900bea6c6f132e037100b07bbb13d31b33b2ead63a38d177b479aec99dad +size 10270077736 diff --git a/checkpoint-15000/random_states_0.pkl b/checkpoint-15000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..7b5b4bde9111c832cc921f39027636135d5dc906 --- /dev/null +++ b/checkpoint-15000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:895c2f7493a95f931ff51a8d7e89dcaabddb1062237a03479464b82a3df2f46f +size 16036 diff --git a/checkpoint-15000/scaler.pt b/checkpoint-15000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..3c8c1b735b202fa81b6b5b90023d1a841d82a478 --- /dev/null +++ b/checkpoint-15000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69922125fbb07e5e8b1e111a409f50ad1eb7f56d00ce858f2776a6ab3df675dd +size 988 diff --git a/checkpoint-15000/scheduler.bin b/checkpoint-15000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..74a641fc9a382dd8aa71b4a5f1201708873e22d7 --- /dev/null +++ b/checkpoint-15000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3903952dccef0ce0ec5797931a20f62e471153db4d6d97e40c406687ba0cdc17 +size 1000 diff --git a/checkpoint-15000/unet/config.json b/checkpoint-15000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-15000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-15000/unet/diffusion_pytorch_model.safetensors b/checkpoint-15000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fb950798594287a8e6f6ba46f34c5cb94cbe74d3 --- /dev/null +++ b/checkpoint-15000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68ffe905f1e4e8483e9d5c542499eb749212391437f6a7883acedc2e12bab08f +size 10270077736 diff --git a/checkpoint-20000/random_states_0.pkl b/checkpoint-20000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..f790e12d30dbe9b0338b21d80431a38874695ac7 --- /dev/null +++ b/checkpoint-20000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:499075326f3fb3e536e27ab341469542eb8cf4f42239e4a9e3e3ea4dcf7569a4 +size 16036 diff --git a/checkpoint-20000/scaler.pt b/checkpoint-20000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4036cea424f0dcddd93f59693beaeb7e4437123 --- /dev/null +++ b/checkpoint-20000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97d4e1c38d4f0421f975753ddf60782bf9d9533e8b06eda67f128dfc8fb7b820 +size 988 diff --git a/checkpoint-20000/scheduler.bin b/checkpoint-20000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..a1d22678dc5cbeab4c32b4bd353d9b2eab74e12a --- /dev/null +++ b/checkpoint-20000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7a23387f2a388acc1a12a024990359fe75b09c636c5bc96db1314af1151677f +size 1000 diff --git a/checkpoint-20000/unet/config.json b/checkpoint-20000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-20000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-20000/unet/diffusion_pytorch_model.safetensors b/checkpoint-20000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..09826f5f585c59f9b5d384f94280db1444ace26f --- /dev/null +++ b/checkpoint-20000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4ee2a58aab641f3dc94d6b8988a92a674a063dcff444c3af159a7f18f46f7e6 +size 10270077736 diff --git a/checkpoint-25000/random_states_0.pkl b/checkpoint-25000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..ca6fc5594f30d9ea310b70d67a429e473dd3a6dc --- /dev/null +++ b/checkpoint-25000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d45e466c97b84c56456211114a767301587ff6f7122c8113f0a967a8b904a9f2 +size 16036 diff --git a/checkpoint-25000/scaler.pt b/checkpoint-25000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2bf1e2f2beb8167a4c239776874fc908ad163ac --- /dev/null +++ b/checkpoint-25000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8bd5f4dc3443a2879b2e99eb74d3936a3cad9c7fa25f64fa4c974498c748209 +size 988 diff --git a/checkpoint-25000/scheduler.bin b/checkpoint-25000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..02a50652e993c5b1716aac409427add4b28e8cea --- /dev/null +++ b/checkpoint-25000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60ed35b01cda610b696704d98f7e23c658f88456dc96e9db879a7b2bf0072391 +size 1000 diff --git a/checkpoint-25000/unet/config.json b/checkpoint-25000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-25000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-25000/unet/diffusion_pytorch_model.safetensors b/checkpoint-25000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2485479b39dda8cc8eb235828d1f0c3f7f55b426 --- /dev/null +++ b/checkpoint-25000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8f3dfa8d88b8abf06482c408aae1f64f41d0bc45ddf72a73cc961f30d33557b +size 10270077736 diff --git a/checkpoint-30000/random_states_0.pkl b/checkpoint-30000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..c458be1800a327146c139a8f47acf89a3a9a2898 --- /dev/null +++ b/checkpoint-30000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71ecc1d45fe5969ec3ddb6470f6dad007545e58f22de065282e93a9c73c9e4c8 +size 16036 diff --git a/checkpoint-30000/scaler.pt b/checkpoint-30000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..430e1c4ab925858e6be65046a8fd44e30eb6293e --- /dev/null +++ b/checkpoint-30000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:360003f823a40049daa0275f08aa5c50eeb500508105c28460680ca3ab204cf9 +size 988 diff --git a/checkpoint-30000/scheduler.bin b/checkpoint-30000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..89e86ca768d727ecf935276e156051dd0f463bc5 --- /dev/null +++ b/checkpoint-30000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bd50c65d05bf606dc53dbbb4d56877bbe34526557ba87fd2da2ada4fc6bea1e +size 1000 diff --git a/checkpoint-30000/unet/config.json b/checkpoint-30000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-30000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-30000/unet/diffusion_pytorch_model.safetensors b/checkpoint-30000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fd2036a5ffc142eed7d582d1f19fbefc2995bd05 --- /dev/null +++ b/checkpoint-30000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d6a2b853d29604bf40467baeff01a26b687c8f59769eb941dbb44f2400d8d74 +size 10270077736 diff --git a/checkpoint-35000/random_states_0.pkl b/checkpoint-35000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..a8f65ecc1756d9153ae765e4dcbfb8f225bc5785 --- /dev/null +++ b/checkpoint-35000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e845ed03d70f650593ec8c78360894eadec5de6a1a69f15cc490857d79eccc3 +size 16036 diff --git a/checkpoint-35000/scaler.pt b/checkpoint-35000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d5b3e64ab90f009eef46798586d79950509ee20 --- /dev/null +++ b/checkpoint-35000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26b8ce69d2bbac3edc8edb7b7f6c259c240f871a3ead48edc03f66c08fc2a67c +size 988 diff --git a/checkpoint-35000/scheduler.bin b/checkpoint-35000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..5c45a6d9d7125a793fb86905408b4acf60d57f24 --- /dev/null +++ b/checkpoint-35000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a86cb0842d41362e55ef1f89679002ba18763ab2a3acb791f5dab413cf826e2 +size 1000 diff --git a/checkpoint-35000/unet/config.json b/checkpoint-35000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-35000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-35000/unet/diffusion_pytorch_model.safetensors b/checkpoint-35000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..25ddf170bfb21a986cacc2b6d2281d5b98ef66a5 --- /dev/null +++ b/checkpoint-35000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bd8a2840f123436db5d49104aa355e6d90bec27cb75cb3a7cc0a7ef520780c6 +size 10270077736 diff --git a/checkpoint-40000/random_states_0.pkl b/checkpoint-40000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..67890add39ca824f3b21266d8065ac1942b3840a --- /dev/null +++ b/checkpoint-40000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f61e4aa9644c8e820beb42106c96d354568451497eb9b7ab6bd515578209c80 +size 16036 diff --git a/checkpoint-40000/scaler.pt b/checkpoint-40000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d2302b93f90e1126a5503aef8563c3ccf179bd1c --- /dev/null +++ b/checkpoint-40000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03cfa14ab3b8c7ee3b70db9d8528c9de3591ff7852b2b33077111289210306b6 +size 988 diff --git a/checkpoint-40000/scheduler.bin b/checkpoint-40000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..e7c98b590d1640b2b8bb3ebf670065a06b224702 --- /dev/null +++ b/checkpoint-40000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6e3424f966c9e042c9c26af312e48f5f79d292b5631023ba1bae8b544202c0b +size 1000 diff --git a/checkpoint-40000/unet/config.json b/checkpoint-40000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-40000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-40000/unet/diffusion_pytorch_model.safetensors b/checkpoint-40000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ffae855dd7a222a3b6982fdfed3777d8af88d0a5 --- /dev/null +++ b/checkpoint-40000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f8634dcecc3890a63daa485eeea718a25a1e326090930c204ada7d782943a34 +size 10270077736 diff --git a/checkpoint-45000/random_states_0.pkl b/checkpoint-45000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..fbc3420c28803b8210eb247adaf45b630fe26865 --- /dev/null +++ b/checkpoint-45000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd9b32ddb1f583ebab40b7737f06b1c7267d6c07cbe0ae22bf85556261e68d7d +size 16036 diff --git a/checkpoint-45000/scaler.pt b/checkpoint-45000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..37c7c8b5943e81547a0f06d7e47467348c5f236a --- /dev/null +++ b/checkpoint-45000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34ca2ab32e6d785538f2d6692a7b0834b28fdc0d7fb95f7780fab28883908b27 +size 988 diff --git a/checkpoint-45000/scheduler.bin b/checkpoint-45000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..54bdce4ba6d911766b9a3b247737017464bcee2b --- /dev/null +++ b/checkpoint-45000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:117417fd5556fcc91598f1ff3db540b4f46180384a167dee989723610e69bb8f +size 1000 diff --git a/checkpoint-45000/unet/config.json b/checkpoint-45000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-45000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-45000/unet/diffusion_pytorch_model.safetensors b/checkpoint-45000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7e35875a62ca7dc1f707d47f070d4fd384339cd8 --- /dev/null +++ b/checkpoint-45000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4b7855ea16ebda7dff73fb479438dad075caa9bf40a64bc9b5fe90c9176e995 +size 10270077736 diff --git a/checkpoint-5000/random_states_0.pkl b/checkpoint-5000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..4875de4a82545b0fe2aca812b4767bc995665373 --- /dev/null +++ b/checkpoint-5000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbbc21579b451da2a0f4bb3bf3f3b657ad0181ea85e124bc8edd5ac442d8a9bf +size 16036 diff --git a/checkpoint-5000/scaler.pt b/checkpoint-5000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..da81849ef9b20a8870285eaa92c68ae84344e397 --- /dev/null +++ b/checkpoint-5000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41216328ab75de937007afb04d76156949bedb908461001a451c8991c2ba8cca +size 988 diff --git a/checkpoint-5000/scheduler.bin b/checkpoint-5000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..d3beff5e1b0451ec90a58321a702bd0183fc21b5 --- /dev/null +++ b/checkpoint-5000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:342fe3a3a88174485c035ff966bc745e22129e5a18598af4cc0235385103124d +size 1000 diff --git a/checkpoint-5000/unet/config.json b/checkpoint-5000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-5000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-5000/unet/diffusion_pytorch_model.safetensors b/checkpoint-5000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f46d9b9d399408e9f61402f8a5552450196d46b8 --- /dev/null +++ b/checkpoint-5000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0943561c0d8ffe9f11ca33eaa714abb607728f1813ca4ba4c8607a9f7f702305 +size 10270077736 diff --git a/checkpoint-50000/random_states_0.pkl b/checkpoint-50000/random_states_0.pkl new file mode 100644 index 0000000000000000000000000000000000000000..7948b7350a2c39d4ef75f741dad85dbf4a5b012b --- /dev/null +++ b/checkpoint-50000/random_states_0.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a79370d420350bb07ecd3ec002744058f2815d22cf0db4739a87b35521df31ab +size 16036 diff --git a/checkpoint-50000/scaler.pt b/checkpoint-50000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ff142173e3d9243a98ff7f5e8a5ea64d417e8063 --- /dev/null +++ b/checkpoint-50000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59a9b3fd995070be4e8e57f0cf07ae30e24b5908cfa2650770cdcd2c93fdac6b +size 988 diff --git a/checkpoint-50000/scheduler.bin b/checkpoint-50000/scheduler.bin new file mode 100644 index 0000000000000000000000000000000000000000..c498dcf28f8e12ecee4a7cf82eedac644794f939 --- /dev/null +++ b/checkpoint-50000/scheduler.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:655c9e0f92a95ea9c285a12bb30639eab13398126984b3baae8117b71e2ac20c +size 1000 diff --git a/checkpoint-50000/unet/config.json b/checkpoint-50000/unet/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/checkpoint-50000/unet/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/checkpoint-50000/unet/diffusion_pytorch_model.safetensors b/checkpoint-50000/unet/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..20d82335d21e3e2bd997798a2a20c35893e5cb68 --- /dev/null +++ b/checkpoint-50000/unet/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea84dd0d799fef0de5eb6a3b475703a876902b8a55e63a3b318c691613574949 +size 10270077736 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..25bdef6f4c3ebef80c6b674e7878f989e76aaee9 --- /dev/null +++ b/config.json @@ -0,0 +1,73 @@ +{ + "_class_name": "UNet2DConditionModel", + "_diffusers_version": "0.28.0.dev0", + "_name_or_path": "stabilityai/stable-diffusion-xl-base-1.0", + "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 + ], + "attention_type": "default", + "block_out_channels": [ + 320, + 640, + 1280 + ], + "center_input_sample": false, + "class_embed_type": null, + "class_embeddings_concat": false, + "conv_in_kernel": 3, + "conv_out_kernel": 3, + "cross_attention_dim": 2048, + "cross_attention_norm": null, + "down_block_types": [ + "DownBlock2D", + "CrossAttnDownBlock2D", + "CrossAttnDownBlock2D" + ], + "downsample_padding": 1, + "dropout": 0.0, + "dual_cross_attention": false, + "encoder_hid_dim": null, + "encoder_hid_dim_type": null, + "flip_sin_to_cos": true, + "freq_shift": 0, + "in_channels": 4, + "layers_per_block": 2, + "mid_block_only_cross_attention": null, + "mid_block_scale_factor": 1, + "mid_block_type": "UNetMidBlock2DCrossAttn", + "norm_eps": 1e-05, + "norm_num_groups": 32, + "num_attention_heads": null, + "num_class_embeds": null, + "only_cross_attention": false, + "out_channels": 4, + "projection_class_embeddings_input_dim": 2816, + "resnet_out_scale_factor": 1.0, + "resnet_skip_time_act": false, + "resnet_time_scale_shift": "default", + "reverse_transformer_layers_per_block": null, + "sample_size": 128, + "time_cond_proj_dim": null, + "time_embedding_act_fn": null, + "time_embedding_dim": null, + "time_embedding_type": "positional", + "timestep_post_act": null, + "transformer_layers_per_block": [ + 1, + 2, + 10 + ], + "up_block_types": [ + "CrossAttnUpBlock2D", + "CrossAttnUpBlock2D", + "UpBlock2D" + ], + "upcast_attention": null, + "use_linear_projection": true +} diff --git a/diffusion_pytorch_model.safetensors b/diffusion_pytorch_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..20d82335d21e3e2bd997798a2a20c35893e5cb68 --- /dev/null +++ b/diffusion_pytorch_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea84dd0d799fef0de5eb6a3b475703a876902b8a55e63a3b318c691613574949 +size 10270077736