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
662 stars 62 forks source link

Update test_sagemaker.yaml #73

Closed wanmeihuali closed 1 year ago

github-actions[bot] commented 1 year ago

Running experiment on sagemaker with git sha 6e4f7e13aceeed456273ecbbefd56341b4748b41

github-actions[bot] commented 1 year ago

Training job densify-policy-2-6e4f7e1-230701-193258-tat-truck created

github-actions[bot] commented 1 year ago

Training job densify-policy-2-6e4f7e1-230701-193258-tat-truck completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-2-6e4f7e1-230701-193258-tat-truck/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck/densify-policy-2-6e4f7e1-230701-193258-tat-truck/output/output.tar.gz

github-actions[bot] commented 1 year ago

Training job densify-policy-2-6e4f7e1-230701-193258-tat-truck final metrics:

Latest Metrics

train:ssimloss train:l1loss train:iteration train:loss train:ssim val:psnr val:ssim val:loss train:num_valid_points train:psnr
0.14359962940216064 0.0318073146045208 30000.0 0.054165780544281006 0.8564003705978394 24.462080001831055 0.840058445930481 0.060352277010679245 436556.0 24.648658752441406

Max Metrics

train:7kssim train:5kssim val:5kpsnr val:7kssim val:7kpsnr train:ssim val:psnr val:5kssim val:ssim train:7kpsnr train:psnr train:5kpsnr
0.8383926749229431 0.7934287786483765 22.262521743774414 0.8018091320991516 23.027528762817383 0.8564003705978394 24.462080001831055 0.7748708128929138 0.840058445930481 23.939111709594727 24.648658752441406 23.63104248046875