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
659 stars 62 forks source link

Experiment to add entropy to grad #140

Closed wanmeihuali closed 12 months ago

github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha c60a9f57c13f6747c48842a1d78c499673ed047f

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010942-tat-train-baseline created

github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha c60a9f57c13f6747c48842a1d78c499673ed047f

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010957-tat-truck-baseline created

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010957-tat-truck-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-c60a9f5-231009-010957-tat-truck-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-c60a9f5-231009-010957-tat-truck-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010957-tat-truck-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.02750481851398945 0.04671567678451538 441216.0 26.80449676513672 0.8764408826828003 0.12355911731719971 0.05515303835272789 24.966114044189453 0.8640872836112976

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
23.637409210205078 0.8565502166748047 22.780866622924805 0.8191241025924683 26.80449676513672 0.8764408826828003 23.13558006286621 0.8160795569419861 23.8129940032959 0.8376764059066772 24.966114044189453 0.8640872836112976
github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha 5f41a93228830cf8e5c8b4f8dbdb8a2ded01b161

github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031403-tat-truck-baseline created

github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha 5f41a93228830cf8e5c8b4f8dbdb8a2ded01b161

github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031411-tat-train-baseline created

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010942-tat-train-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-c60a9f5-231009-010942-tat-train-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-c60a9f5-231009-010942-tat-train-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-c60a9f5-231009-010942-tat-train-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.0550207756459713 0.07832827419042587 680845.0 22.1871395111084 0.8284417390823364 0.17155826091766357 0.10075955837965012 20.922143936157227 0.7998666763305664

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
25.24920654296875 0.8906899690628052 23.606063842773438 0.8213552236557007 22.1871395111084 0.8284417390823364 19.12961196899414 0.731582760810852 19.73741912841797 0.7576631903648376 20.922143936157227 0.7998666763305664
github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031403-tat-truck-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-5f41a93-231009-031403-tat-truck-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-5f41a93-231009-031403-tat-truck-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031403-tat-truck-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.025792447850108147 0.0428241491317749 441987.0 26.606510162353516 0.8890490531921387 0.11095094680786133 0.05506135895848274 25.01679229736328 0.8643010854721069

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
24.413545608520508 0.8359179496765137 25.65595245361328 0.8772885203361511 26.606510162353516 0.8890490531921387 23.028295516967773 0.8129215240478516 23.705944061279297 0.8349975943565369 25.01679229736328 0.8643010854721069
github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031411-tat-train-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-5f41a93-231009-031411-tat-train-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-5f41a93-231009-031411-tat-train-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-5f41a93-231009-031411-tat-train-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.03879787027835846 0.049730051308870316 669202.0 25.410064697265625 0.9065412282943726 0.09345877170562744 0.10145072638988495 20.791149139404297 0.7993519902229309

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
23.40985107421875 0.8211886286735535 22.658485412597656 0.8382388949394226 25.410064697265625 0.9065412282943726 19.18515968322754 0.7303782105445862 19.890708923339844 0.7592393755912781 20.791149139404297 0.7993519902229309
github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha a7eef8b6d25e00250ee63d1bc3b5672e187a6bde

github-actions[bot] commented 12 months ago

Training job try-entropy-a7eef8b-231009-064406-tat-truck-baseline created

github-actions[bot] commented 12 months ago

Training job try-entropy-a7eef8b-231009-064406-tat-truck-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-a7eef8b-231009-064406-tat-truck-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-a7eef8b-231009-064406-tat-truck-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-a7eef8b-231009-064406-tat-truck-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.20388361811637878 0.2839539647102356 214512.0 11.89755630493164 0.39576470851898193 0.6042352914810181 0.29992449283599854 11.164863586425781 0.43217572569847107

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
12.535301208496094 0.4729728400707245 13.79015064239502 0.5134201049804688 11.89755630493164 0.39576470851898193 11.84522819519043 0.4488368034362793 11.241450309753418 0.439075767993927 11.164863586425781 0.43217572569847107
github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha 333e891802240c0fb883ac3e2136ef2e96dc68e8

github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182606-tat-train-baseline created

github-actions[bot] commented 12 months ago

Running experiment on sagemaker with git sha 333e891802240c0fb883ac3e2136ef2e96dc68e8

github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182715-tat-truck-baseline created

github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182715-tat-truck-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-333e891-231009-182715-tat-truck-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-truck-baseline/try-entropy-333e891-231009-182715-tat-truck-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182715-tat-truck-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.5383625030517578 0.6304363012313843 146929.0 4.4726481437683105 0.001268533756956458 0.9987314939498901 0.6232029795646667 4.728336334228516 0.0011190521763637662

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
4.828002452850342 0.0007100386428646743 4.9570088386535645 0.0007174116326496005 4.4726481437683105 0.001268533756956458 4.728366851806641 0.0011238715378567576 4.728342056274414 0.0011196979321539402 4.728336334228516 0.0011190521763637662
github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182606-tat-train-baseline completed.

Model url: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-333e891-231009-182606-tat-train-baseline/output/model.tar.gz,

tensorboard output path: s3://taichi-3d-gaussian-splatting-log/tat-train-baseline/try-entropy-333e891-231009-182606-tat-train-baseline/output/output.tar.gz

github-actions[bot] commented 12 months ago

Training job try-entropy-333e891-231009-182606-tat-train-baseline final metrics:

Latest Metrics

train:iteration train:l1loss train:loss train:num_valid_points train:psnr train:ssim train:ssimloss val:loss val:psnr val:ssim
30000.0 0.5874654650688171 0.6697753071784973 147870.0 3.447625160217285 0.0009854701347649097 0.9990145564079285 0.5711046457290649 5.487584114074707 0.007891391403973103

Max Metrics

train:5kpsnr train:5kssim train:7kpsnr train:7kssim train:psnr train:ssim val:5kpsnr val:5kssim val:7kpsnr val:7kssim val:psnr val:ssim
4.375644207000732 0.000313255877699703 5.90975284576416 0.009472731500864029 3.447625160217285 0.0009854701347649097 5.487605094909668 0.007896960712969303 5.487586498260498 0.007891782559454441 5.487584114074707 0.007891391403973103
wanmeihuali commented 12 months ago

Seem not working at all