| task: t2v |
| work_dir: output/sana_video2_5b_720p |
| report_to: none |
| tracker_project_name: sana-video-2 |
| data: |
| data_dir: |
| video: data/video_toy_data |
| image_size: 720 |
| type: SanaZipDataset |
| transform: default_train_video |
| aspect_ratio_type: ASPECT_RATIO_VIDEO_720_MS_DIV32 |
| caption_selection_type: proportion |
| caption_proportion: |
| prompt: 1 |
| num_frames: 193 |
| target_fps: 24 |
| resample_fps: false |
| model: |
| model: SanaVideo2_5B |
| image_size: 720 |
| mixed_precision: bf16 |
| fp32_attention: true |
| load_from: null |
| multi_scale: false |
| linear_head_dim: 128 |
| softmax_head_dim: 256 |
| softmax_ratio: 0.25 |
| ffn_type: SwiGLU |
| mlp_ratio: 4.0 |
| use_pe: true |
| pos_embed_type: wan_rope |
| qk_norm: true |
| cross_norm: true |
| class_dropout_prob: 0.1 |
| attn_res_block_size: 8 |
| vae: |
| vae_type: LTX2VAE_diffusers |
| vae_pretrained: output/pretrained_models/LTX-2.3-Diffusers |
| weight_dtype: bfloat16 |
| vae_latent_dim: 128 |
| vae_downsample_rate: 32 |
| vae_stride: [8, 32, 32] |
| sample_posterior: false |
| use_causal_encode: true |
| text_encoder: |
| text_encoder_name: gemma-2-2b-it |
| caption_channels: 2304 |
| y_norm: true |
| y_norm_scale_factor: 0.01 |
| model_max_length: 300 |
| chi_prompt: |
| - 'Given a user prompt, generate an "Enhanced prompt" that provides detailed visual descriptions suitable for image generation. Evaluate the level of detail in the user prompt:' |
| - '- If the prompt is simple, focus on adding specifics about colors, shapes, sizes, textures, and spatial relationships to create vivid and concrete scenes.' |
| - '- If the prompt is already detailed, refine and enhance the existing details slightly without overcomplicating.' |
| - 'Here are examples of how to transform or refine prompts:' |
| - '- User Prompt: A cat sleeping -> Enhanced: A small, fluffy white cat curled up in a round shape, sleeping peacefully on a warm sunny windowsill, surrounded by pots of blooming red flowers.' |
| - '- User Prompt: A busy city street -> Enhanced: A bustling city street scene at dusk, featuring glowing street lamps, a diverse crowd of people in colorful clothing, and a double-decker bus passing by towering glass skyscrapers.' |
| - 'Please generate only the enhanced description for the prompt below and avoid including any additional commentary or evaluations:' |
| - 'User Prompt: ' |
| scheduler: |
| train_sampling_steps: 1000 |
| predict_flow_v: true |
| noise_schedule: linear_flow |
| pred_sigma: false |
| learn_sigma: false |
| flow_shift: 6.0 |
| inference_flow_shift: 12.0 |
| weighting_scheme: logit_normal |
| logit_mean: 0.0 |
| logit_std: 0.95 |
| vis_sampler: flow_dpm-solver |
| train: |
| use_fsdp: true |
| num_workers: 8 |
| seed: 1 |
| train_batch_size: 1 |
| num_epochs: 10 |
| gradient_accumulation_steps: 1 |
| grad_checkpointing: true |
| gradient_clip: 0.1 |
| joint_training_interval: 0 |
| optimizer: |
| type: AdamW |
| lr: 5.0e-5 |
| betas: [0.9, 0.999] |
| eps: 1.0e-10 |
| weight_decay: 0.0 |
| auto_lr: null |
| lr_schedule: constant |
| lr_schedule_args: |
| num_warmup_steps: 500 |
| log_interval: 10 |
| save_model_steps: 1000 |
| save_model_epochs: 1 |
|
|
