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 #75

Closed wanmeihuali closed 1 year ago

github-actions[bot] commented 1 year ago

Running experiment on sagemaker with git sha ca4e153f39f3a0e7b007646dd60faf5e32fa0cd5

github-actions[bot] commented 1 year ago

Training job densify-policy-4-ca4e153-230701-193333-tat-truck created

github-actions[bot] commented 1 year ago

Training job densify-policy-4-ca4e153-230701-193333-tat-truck completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-4-ca4e153-230701-193333-tat-truck/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-4-ca4e153-230701-193333-tat-truck/output/output.tar.gz

github-actions[bot] commented 1 year ago

Training job densify-policy-4-ca4e153-230701-193333-tat-truck final metrics:

Latest Metrics

train:num_valid_points train:iteration val:psnr val:loss train:l1loss train:psnr train:ssimloss val:ssim train:ssim train:loss
194615.0 30000.0 23.97435760498047 0.0667610689997673 0.02998616173863411 25.505786895751953 0.15720754861831665 0.8176519870758057 0.8427924513816833 0.055430442094802856

Max Metrics

val:5kssim train:5kssim val:7kssim val:7kpsnr val:psnr train:7kssim train:5kpsnr val:5kpsnr val:ssim train:ssim train:psnr train:7kpsnr
0.7694705724716187 0.7486960291862488 0.7948331832885742 22.923336029052734 23.97435760498047 0.7911872863769531 22.607128143310547 22.146446228027344 0.8176519870758057 0.8427924513816833 25.505786895751953 22.80929183959961