| pretrained_model_path: checkpoints/CompVis/stable-diffusion-v1-4 | |
| output_dir: /home/user/app/experiments/talking-man | |
| train_data: | |
| video_path: /tmp/gradio/f79f070ba10b1b7872faa157ada00e09644ef2f2/Untitled Video.mp4 | |
| prompt: A person of East Asian ethnicity standing in a well-lit office environment | |
| with large windows, speaking directly to the camera, wearing a white shirt, and | |
| conveying a calm and professional demeanor. The background is slightly blurred, | |
| showcasing greenery and modern office decor, with minimal distractions | |
| n_sample_frames: 8 | |
| width: 512 | |
| height: 512 | |
| sample_start_idx: 0 | |
| sample_frame_rate: 1 | |
| validation_data: | |
| prompts: | |
| - A person of East Asian descent is standing in a bright office setting with large | |
| windows, addressing the camera directly. They are dressed in a white shirt and | |
| project a calm, professional demeanor. The background is slightly out of focus, | |
| highlighting greenery and modern office furnishings, with minimal visual distractions | |
| video_length: 8 | |
| width: 512 | |
| height: 512 | |
| num_inference_steps: 50 | |
| guidance_scale: 7.5 | |
| validation_steps: 100 | |
| trainable_modules: | |
| - attn1.to_q | |
| - attn2.to_q | |
| - attn_temp | |
| train_batch_size: 1 | |
| max_train_steps: 300 | |
| learning_rate: 3.5e-05 | |
| scale_lr: false | |
| lr_scheduler: constant | |
| lr_warmup_steps: 0 | |
| adam_beta1: 0.9 | |
| adam_beta2: 0.999 | |
| adam_weight_decay: 0.01 | |
| adam_epsilon: 1.0e-08 | |
| max_grad_norm: 1.0 | |
| gradient_accumulation_steps: 1 | |
| gradient_checkpointing: true | |
| checkpointing_steps: 1000 | |
| resume_from_checkpoint: null | |
| mixed_precision: fp16 | |
| use_8bit_adam: false | |
| enable_xformers_memory_efficient_attention: true | |
| seed: 31118 | |