| job: extension | |
| config: | |
| name: Alessa_wan2.2_low | |
| process: | |
| - type: sd_trainer | |
| training_folder: /out | |
| device: cuda:0 | |
| network: | |
| type: lora | |
| linear: 32 | |
| linear_alpha: 32 | |
| save: | |
| dtype: float16 | |
| save_every: 500 | |
| max_step_saves_to_keep: 4 | |
| datasets: | |
| - folder_path: /alessafiore_dataset | |
| caption_ext: txt | |
| caption_dropout_rate: 0.05 | |
| num_frames: 1 | |
| resolution: | |
| - 1024 | |
| train: | |
| batch_size: 1 | |
| steps: 4000 | |
| gradient_accumulation: 1 | |
| train_unet: true | |
| train_text_encoder: false | |
| gradient_checkpointing: true | |
| noise_scheduler: flowmatch | |
| timestep_type: linear | |
| optimizer: adamw8bit | |
| lr: 0.0001 | |
| optimizer_params: | |
| weight_decay: 0.0001 | |
| skip_first_sample: true | |
| dtype: bf16 | |
| switch_boundary_every: 10 | |
| cache_text_embeddings: true | |
| model: | |
| name_or_path: ai-toolkit/Wan2.2-T2V-A14B-Diffusers-bf16 | |
| arch: wan22_14b | |
| quantize: false | |
| quantize_te: false | |
| qtype_te: qfloat8 | |
| low_vram: false | |
| model_kwargs: | |
| train_high_noise: false | |
| train_low_noise: true | |
| sample: | |
| sampler: flowmatch | |
| sample_every: 1000 | |
| width: 832 | |
| height: 1216 | |
| num_frames: 1 | |
| fps: 16 | |
| prompts: | |
| - tik tok video of an ohwx woman with shoulder length hair is kneeling on floor | |
| near a bed, She gathers her hair up slowly, She is looking at viewer with | |
| a playful expression on her face, | |
| neg: '' | |
| seed: 42 | |
| walk_seed: true | |
| guidance_scale: 3.5 | |
| sample_steps: 25 | |
| meta: | |
| name: Alessa_wan2.2_low | |
| version: '1.0' | |