← Run Index
Run summary · P1 Phase G.5 final Pareto integration
lego-phase-h-pareto-10000
Phase H Pareto sweep @ stop_iter=10000 variant (early stop + SH progressive)
partial pass
01
設定
- dataset
- /Users/otkrickey/dev/3dgs-workspace/datasets/nerf_synthetic/lego
- gt_convention
- Premultiplied
- iterations
- 10,000
- seed
- 42
- capacity
- 1,000,000 splats
- sh_degree
- 3
- loss
- L1Ssim
- lambda
- 0.200
- ssim
- window=7 sigma=1
- backend.backward
- Simd
- backend.loss_path
- Gpu
02
結果
Val PSNR
35.93dB
Wallclock
8m 42s
Final splats
385,007
Final loss
5.759161e-3
03
3D ビューワー
学習結果 PLY を .splat 形式に変換し、antimatter15/splat の WebGL ビューワーで対話表示。
04
推移
Loss curve — 反復ごとの training loss
Splat count — refine の発火と最終構成
05
中間ログ抜粋
| iter | 本 run loss |
|---|---|
| 1 | 5.9385e-1 |
| 1,000 | 5.5153e-2 |
| 2,000 | 1.4354e-2 |
| 3,000 | 8.6712e-3 |
| 4,000 | 7.7325e-3 |
| 5,000 | 6.8934e-3 |
| 6,000 | 6.4195e-3 |
| 7,000 | 6.1993e-3 |
| 8,000 | 6.0080e-3 |
| 9,000 | 5.8778e-3 |
| 10,000 | 5.7592e-3 |