Spaces:
Running on Zero
Running on Zero
File size: 2,422 Bytes
3718945 98e7453 3718945 98e7453 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 | exp_root_dir: "outputs"
name: "diffusion_gs_scene_re10k_256_stage1_eval_test"
tag: "${rmspace:${system.shape_model_type}+lr${system.optimizer.args.lr},_}"
seed: 0
# resume: outputs/image-to-shape-diffusion/clip-dino-rgb-pixart-lr2e4-ddim-refine-sum-scaleup-from-190k-norm-scale-up-1024-new-vae/michelangelo-autoencoder+n12288+noise0.0+pfeat3+normembFalse+lr0.0002+qkvbiasFalse+nfreq8+ln_postTrue/ckpts/last.ckpt
data_type: "Re10k-datamodule"
data:
local_dir: "DATAPATH_TO_RE10K_TRAIN_FULLLIST"
local_eval_dir: "DATAPATH_TO_RE10K_EVAL_FULLLIST"
sel_views: 3
eval_subset: 16
sel_views_train: 4
training_res: [256, 256]
batch_size: 16
eval_batch_size: 1
num_workers: 4
num_workers_val: 4
system_type: "diffusion-gs-scene-system"
system:
num_inference_steps: 30
save_intermediate_video: True
save_result_for_eval: True
shape_model_type: "diffusion-gs-model-scene"
shape_model:
pretrained_model_name_or_path: PATH_TO_CKPT
width: 1024
in_channels: 9 #rgb+plucker
patch_size: 8
n_gaussians: 2
dim_heads: 64
num_layers: 24
range_setting_near: 0
range_setting_far: 500
prior_distribution: 'gaussian' #sphere_gaussian
ray_pe_type: 'plk'
#ray_pe_type: str = "relative_plk"
use_flash: true
use_checkpoint: true
noise_scheduler_type: "diffusionGS.models.scheduler.ddim_scheduler.DDIMScheduler"
noise_scheduler:
num_train_timesteps: 1000
prediction_type: "sample"
loggers:
wandb:
enable: false
project: "diffusionGS"
name: image-to-shape-diffusion+${name}+${tag}
loss:
loss_type: "mse"
lambda_diffusion: 1. #[150, 0., 1., 151] #1.
lambda_lpips: 0.1 #[150, 0., 0.5, 151]
lambda_ssim: 0.0
lambda_pointsdist: 0.0 #[150, 1., 0., 151]
lambda_xyz: 0. #[100, 0., 1., 101]
lambda_depth: 0. #[100, 0., 0.25, 101]
optimizer:
name: AdamW
args:
lr: 1.e-4
betas: [0.9, 0.99]
eps: 1.e-6
scheduler:
name: CosineAnnealingLR
args:
T_max: 500000
eta_min: 1e-6
trainer:
num_nodes: 1
max_epochs: 1000000
log_every_n_steps: 1
num_sanity_val_steps: 1
check_val_every_n_epoch: 1
accumulate_grad_batches: 1
gradient_clip_val: 0.5
enable_progress_bar: true
precision: 16-mixed
strategy: 'ddp_find_unused_parameters_true' #_find_unused_parameters_true
checkpoint:
save_last: true
save_top_k: -1
every_n_train_steps: 5000
|