Instructions to use aipracticecafe/dual-stream-dit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use aipracticecafe/dual-stream-dit with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("aipracticecafe/dual-stream-dit", dtype=torch.bfloat16, device_map="cuda") prompt = "1girl, souryuu asuka langley, neon genesis evangelion, rebuild of evangelion, evangelion: 3.0 you can \\(not\\) redo, momoko \\(momopoco\\), toosaka asagi, ebifurya, sensitive, masterpiece, good score, absurdres, highres, best quality, amazing quality, great quality, solo, long hair, blue eyes, orange hair, breasts, parted lips, official alternate costume, pilot suit, red bodysuit, black eye patch, animal hat, cabbie hat, cat hat, black hat, gloves, hat, hat horns, jacket, mecha pilot suit, bodysuit under clothes, red gloves, red jacket, track jacket, turtleneck, turtleneck jacket, interface headset, multicolored clothes, interface headset \\(evangelion\\), upper body, simple background" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Upload folder using huggingface_hub
Browse files- epoch_209_step_292500/config.json +110 -0
- epoch_209_step_292500/optimizer.pt +3 -0
- epoch_209_step_292500/scheduler.pt +3 -0
- epoch_209_step_292500/training_state.pt +3 -0
- epoch_209_step_292500/unet.safetensors +3 -0
- epoch_210_step_296400/config.json +110 -0
- epoch_210_step_296400/optimizer.pt +3 -0
- epoch_210_step_296400/scheduler.pt +3 -0
- epoch_210_step_296400/training_state.pt +3 -0
- epoch_210_step_296400/unet.safetensors +3 -0
epoch_209_step_292500/config.json
ADDED
|
@@ -0,0 +1,110 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "sprint_dual",
|
| 3 |
+
"model_type": "sprint_dual",
|
| 4 |
+
"in_channels": 32,
|
| 5 |
+
"hidden_size": 768,
|
| 6 |
+
"depth": 13,
|
| 7 |
+
"num_heads": 12,
|
| 8 |
+
"encoder_depth": 2,
|
| 9 |
+
"decoder_depth": 2,
|
| 10 |
+
"drop_ratio": 0.75,
|
| 11 |
+
"drop_target": "image",
|
| 12 |
+
"residual_type": "concat_linear",
|
| 13 |
+
"hf_text_encoder": "LiquidAI/LFM2.5-230M",
|
| 14 |
+
"hf_vae": "kaiyuyue/FLUX.2-dev-vae",
|
| 15 |
+
"vae_mean": 0.05106564902828182,
|
| 16 |
+
"vae_std": 1.69915383699,
|
| 17 |
+
"resume_from_checkpoint": "results/discrete_rope/epoch_209_step_290199",
|
| 18 |
+
"cfg_mask_prob": 0.0,
|
| 19 |
+
"skip_checkpointing_layers": 0,
|
| 20 |
+
"use_random_drop": false,
|
| 21 |
+
"vae_batch_size": 2,
|
| 22 |
+
"use_calibrated_spatial": true,
|
| 23 |
+
"train": {
|
| 24 |
+
"batch_size": 8,
|
| 25 |
+
"lr": 0.0001,
|
| 26 |
+
"epochs": 215,
|
| 27 |
+
"gradient_accumulation_steps": 8,
|
| 28 |
+
"dtype": "fp32",
|
| 29 |
+
"use_checkpointing": true,
|
| 30 |
+
"train_te": false,
|
| 31 |
+
"wd": 0.01,
|
| 32 |
+
"warmup": 0.04,
|
| 33 |
+
"min_lr_ratio": 0.01,
|
| 34 |
+
"reset_scheduler": true,
|
| 35 |
+
"reset_optimizer": false,
|
| 36 |
+
"use_cos_scheduler": false,
|
| 37 |
+
"objective": "flow_matching",
|
| 38 |
+
"timestep_fn": "logit-normal",
|
| 39 |
+
"shift": 2.0,
|
| 40 |
+
"snr_gamma": 5.0,
|
| 41 |
+
"input_perturb": 0.1,
|
| 42 |
+
"cfg_dropout_prob": 0.1,
|
| 43 |
+
"seed": 232,
|
| 44 |
+
"use_bitsandbytes": true,
|
| 45 |
+
"train_only_output": false,
|
| 46 |
+
"skip_save_n_times": 4,
|
| 47 |
+
"run_stability_check": false,
|
| 48 |
+
"world_size": 2,
|
| 49 |
+
"use_ot": false,
|
| 50 |
+
"use_ema": false,
|
| 51 |
+
"ema_decay": 0.999,
|
| 52 |
+
"use_kahan_sum": false,
|
| 53 |
+
"use_cpu_accumulator": false,
|
| 54 |
+
"compile_model": true,
|
| 55 |
+
"gpu_peak_tflops": 165.2
|
| 56 |
+
},
|
| 57 |
+
"data": {
|
| 58 |
+
"dataset_type": "streaming",
|
| 59 |
+
"h5_path": "data",
|
| 60 |
+
"streaming_dataset_name": "aipracticecafe/curated-danbooru-2026",
|
| 61 |
+
"num_workers": 4,
|
| 62 |
+
"prefetch_factor": 4,
|
| 63 |
+
"load_into_ram": true,
|
| 64 |
+
"tag_dropout": 0.1,
|
| 65 |
+
"use_shift_crop": true,
|
| 66 |
+
"resolution": 1024,
|
| 67 |
+
"buffer_size": 2500,
|
| 68 |
+
"cache_latents_to_ram": false,
|
| 69 |
+
"is_latent": false,
|
| 70 |
+
"precompute_num_workers": 2,
|
| 71 |
+
"precompute_prefetch_factor": 2
|
| 72 |
+
},
|
| 73 |
+
"models": {
|
| 74 |
+
"model_type": "sprint_dual",
|
| 75 |
+
"resume_from_checkpoint": "results/discrete_rope/epoch_209_step_290199",
|
| 76 |
+
"output_head_path": null,
|
| 77 |
+
"hf_text_encoder": "LiquidAI/LFM2.5-230M",
|
| 78 |
+
"hidden_size": 768,
|
| 79 |
+
"depth": 13,
|
| 80 |
+
"num_heads": 12,
|
| 81 |
+
"encoder_depth": 2,
|
| 82 |
+
"decoder_depth": 2,
|
| 83 |
+
"drop_ratio": 0.75,
|
| 84 |
+
"drop_target": "image",
|
| 85 |
+
"residual_type": "concat_linear",
|
| 86 |
+
"cfg_mask_prob": 0.0,
|
| 87 |
+
"skip_checkpointing_layers": 0,
|
| 88 |
+
"use_random_drop": false,
|
| 89 |
+
"hf_vae": "kaiyuyue/FLUX.2-dev-vae",
|
| 90 |
+
"vae_mean": 0.05106564902828182,
|
| 91 |
+
"vae_std": 1.69915383699,
|
| 92 |
+
"vae_batch_size": 2,
|
| 93 |
+
"in_channels": 32,
|
| 94 |
+
"use_calibrated_spatial": true
|
| 95 |
+
},
|
| 96 |
+
"paths": {
|
| 97 |
+
"models": "models"
|
| 98 |
+
},
|
| 99 |
+
"logging": {
|
| 100 |
+
"project": "dual-dit",
|
| 101 |
+
"wandb_entity": "aipracticecafe-aipracticecafe",
|
| 102 |
+
"save_interval": 6500,
|
| 103 |
+
"hf_repo": "dual-stream-dit-spatial-rope",
|
| 104 |
+
"wandb_run_name_prefix": "4090-sprint-1MP-latents-rope-swap"
|
| 105 |
+
},
|
| 106 |
+
"sampling": {
|
| 107 |
+
"interval": 1300,
|
| 108 |
+
"config_file": "sample.toml"
|
| 109 |
+
}
|
| 110 |
+
}
|
epoch_209_step_292500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e86b507a2935bf90868f63f0bcd57e888ab58e97d3b181589034cc05aaf58d2c
|
| 3 |
+
size 419905931
|
epoch_209_step_292500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d280ed9e6e025994f6f5685461c75347656fda2c654cdd87cefd5f48b8f5d66
|
| 3 |
+
size 1721
|
epoch_209_step_292500/training_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e93d935ced229bac04b5341942983436fb6f907a047df785dd6bdb2564d4bfca
|
| 3 |
+
size 1303
|
epoch_209_step_292500/unet.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0a3b68f4521dbee33047acdd042814d43be19539b02b953305ab518d376e13f
|
| 3 |
+
size 425266512
|
epoch_210_step_296400/config.json
ADDED
|
@@ -0,0 +1,110 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "sprint_dual",
|
| 3 |
+
"model_type": "sprint_dual",
|
| 4 |
+
"in_channels": 32,
|
| 5 |
+
"hidden_size": 768,
|
| 6 |
+
"depth": 13,
|
| 7 |
+
"num_heads": 12,
|
| 8 |
+
"encoder_depth": 2,
|
| 9 |
+
"decoder_depth": 2,
|
| 10 |
+
"drop_ratio": 0.75,
|
| 11 |
+
"drop_target": "image",
|
| 12 |
+
"residual_type": "concat_linear",
|
| 13 |
+
"hf_text_encoder": "LiquidAI/LFM2.5-230M",
|
| 14 |
+
"hf_vae": "kaiyuyue/FLUX.2-dev-vae",
|
| 15 |
+
"vae_mean": 0.05106564902828182,
|
| 16 |
+
"vae_std": 1.69915383699,
|
| 17 |
+
"resume_from_checkpoint": "results/discrete_rope/epoch_209_step_292500",
|
| 18 |
+
"cfg_mask_prob": 0.0,
|
| 19 |
+
"skip_checkpointing_layers": 0,
|
| 20 |
+
"use_random_drop": false,
|
| 21 |
+
"vae_batch_size": 2,
|
| 22 |
+
"use_calibrated_spatial": true,
|
| 23 |
+
"train": {
|
| 24 |
+
"batch_size": 16,
|
| 25 |
+
"lr": 8e-05,
|
| 26 |
+
"epochs": 215,
|
| 27 |
+
"gradient_accumulation_steps": 8,
|
| 28 |
+
"dtype": "fp32",
|
| 29 |
+
"use_checkpointing": true,
|
| 30 |
+
"train_te": false,
|
| 31 |
+
"wd": 0.01,
|
| 32 |
+
"warmup": 0.04,
|
| 33 |
+
"min_lr_ratio": 0.01,
|
| 34 |
+
"reset_scheduler": true,
|
| 35 |
+
"reset_optimizer": false,
|
| 36 |
+
"use_cos_scheduler": false,
|
| 37 |
+
"objective": "flow_matching",
|
| 38 |
+
"timestep_fn": "logit-normal",
|
| 39 |
+
"shift": 3.0,
|
| 40 |
+
"snr_gamma": 5.0,
|
| 41 |
+
"input_perturb": 0.1,
|
| 42 |
+
"cfg_dropout_prob": 0.1,
|
| 43 |
+
"seed": 232,
|
| 44 |
+
"use_bitsandbytes": true,
|
| 45 |
+
"train_only_output": false,
|
| 46 |
+
"skip_save_n_times": 4,
|
| 47 |
+
"run_stability_check": false,
|
| 48 |
+
"world_size": 2,
|
| 49 |
+
"use_ot": false,
|
| 50 |
+
"use_ema": false,
|
| 51 |
+
"ema_decay": 0.999,
|
| 52 |
+
"use_kahan_sum": false,
|
| 53 |
+
"use_cpu_accumulator": false,
|
| 54 |
+
"compile_model": true,
|
| 55 |
+
"gpu_peak_tflops": 165.2
|
| 56 |
+
},
|
| 57 |
+
"data": {
|
| 58 |
+
"dataset_type": "streaming",
|
| 59 |
+
"h5_path": "data",
|
| 60 |
+
"streaming_dataset_name": "aipracticecafe/curated-danbooru-2026",
|
| 61 |
+
"num_workers": 4,
|
| 62 |
+
"prefetch_factor": 4,
|
| 63 |
+
"load_into_ram": true,
|
| 64 |
+
"tag_dropout": 0.1,
|
| 65 |
+
"use_shift_crop": true,
|
| 66 |
+
"resolution": 1024,
|
| 67 |
+
"buffer_size": 2500,
|
| 68 |
+
"cache_latents_to_ram": false,
|
| 69 |
+
"is_latent": false,
|
| 70 |
+
"precompute_num_workers": 2,
|
| 71 |
+
"precompute_prefetch_factor": 2
|
| 72 |
+
},
|
| 73 |
+
"models": {
|
| 74 |
+
"model_type": "sprint_dual",
|
| 75 |
+
"resume_from_checkpoint": "results/discrete_rope/epoch_209_step_292500",
|
| 76 |
+
"output_head_path": null,
|
| 77 |
+
"hf_text_encoder": "LiquidAI/LFM2.5-230M",
|
| 78 |
+
"hidden_size": 768,
|
| 79 |
+
"depth": 13,
|
| 80 |
+
"num_heads": 12,
|
| 81 |
+
"encoder_depth": 2,
|
| 82 |
+
"decoder_depth": 2,
|
| 83 |
+
"drop_ratio": 0.75,
|
| 84 |
+
"drop_target": "image",
|
| 85 |
+
"residual_type": "concat_linear",
|
| 86 |
+
"cfg_mask_prob": 0.0,
|
| 87 |
+
"skip_checkpointing_layers": 0,
|
| 88 |
+
"use_random_drop": false,
|
| 89 |
+
"hf_vae": "kaiyuyue/FLUX.2-dev-vae",
|
| 90 |
+
"vae_mean": 0.05106564902828182,
|
| 91 |
+
"vae_std": 1.69915383699,
|
| 92 |
+
"vae_batch_size": 2,
|
| 93 |
+
"in_channels": 32,
|
| 94 |
+
"use_calibrated_spatial": true
|
| 95 |
+
},
|
| 96 |
+
"paths": {
|
| 97 |
+
"models": "models"
|
| 98 |
+
},
|
| 99 |
+
"logging": {
|
| 100 |
+
"project": "dual-dit",
|
| 101 |
+
"wandb_entity": "aipracticecafe-aipracticecafe",
|
| 102 |
+
"save_interval": 3900,
|
| 103 |
+
"hf_repo": "dual-stream-dit",
|
| 104 |
+
"wandb_run_name_prefix": "4090-sprint-1MP-latents-rope-swap"
|
| 105 |
+
},
|
| 106 |
+
"sampling": {
|
| 107 |
+
"interval": 1300,
|
| 108 |
+
"config_file": "sample.toml"
|
| 109 |
+
}
|
| 110 |
+
}
|
epoch_210_step_296400/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ce3ae7b9e1c3fc28c10ae3f1bca8df4b40713c7453ae709b275542ab26de193c
|
| 3 |
+
size 419905931
|
epoch_210_step_296400/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:033f61e79f53b5efe9f10e2394873c6b1e79d69aee56b760a778da0e253397ff
|
| 3 |
+
size 1721
|
epoch_210_step_296400/training_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:04899cbd75dfc8e86b5d8f1dd457eceeed59595ce30d6a9ddd8c510334e833f7
|
| 3 |
+
size 1303
|
epoch_210_step_296400/unet.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5259d52b33af97381006c507f37dfef332797d9e579e794fe4def030c6861702
|
| 3 |
+
size 425266512
|