search

wanmeihuali / taichi_3d_gaussian_splatting

An unofficial implementation of paper 3D Gaussian Splatting for Real-Time Radiance Field Rendering by taichi lang.
Apache License 2.0
648 stars 61 forks source link

Update test_sagemaker.yaml #74

Closed wanmeihuali closed 1 year ago

github-actions[bot] commented 1 year ago

Running experiment on sagemaker with git sha 70c7acd05c31334ea2ef30bf03fe90f0e58ebee1

github-actions[bot] commented 1 year ago

Training job densify-policy-3-70c7acd-230701-193325-tat-truck created

github-actions[bot] commented 1 year ago

Training job densify-policy-3-70c7acd-230701-193325-tat-truck completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-3-70c7acd-230701-193325-tat-truck/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-3-70c7acd-230701-193325-tat-truck/output/output.tar.gz

github-actions[bot] commented 1 year ago

Training job densify-policy-3-70c7acd-230701-193325-tat-truck final metrics:

Latest Metrics

train:psnr val:ssim train:loss val:loss train:l1loss train:iteration val:psnr train:num_valid_points train:ssimloss train:ssim
25.020227432250977 0.8403215408325195 0.052141644060611725 0.06021135672926903 0.030450956895947456 30000.0 24.51424789428711 440635.0 0.1389043927192688 0.8610956072807312

Max Metrics

train:psnr val:5kssim val:ssim val:5kpsnr train:5kssim val:7kpsnr val:psnr train:7kpsnr val:7kssim train:5kpsnr train:ssim train:7kssim
25.020227432250977 0.7835337519645691 0.8403215408325195 22.49529457092285 0.8012645244598389 23.1370792388916 24.51424789428711 23.383872985839844 0.8053874969482422 22.735769271850586 0.8610956072807312 0.7979914546012878