 mckay
|
98372f1717
|
优化八叉树性能
|
10 months ago |
|
mckay
|
48755817c0
|
Revert "oct 修改中间版本"
This reverts commit 5bd4a8d866.
|
10 months ago |
|
mckay
|
5bd4a8d866
|
oct 修改中间版本
|
10 months ago |
|
mckay
|
21f8369e20
|
兼容torch.jit.script，load和save方便。但是牺牲了随机访问，性能降了很多
|
10 months ago |
|
mckay
|
2ac55ea3df
|
基本可以跑通，除了八叉树部分保存还有问题
|
10 months ago |
|
mckay
|
26e3984f76
|
加强gen name list功能
|
10 months ago |
|
mckay
|
d5ac19e093
|
调c进行isosurfing
|
10 months ago |
|
mckay
|
cc88dfb798
|
支持normal
|
10 months ago |
|
mckay
|
1e4c360403
|
可以加载和保存
|
11 months ago |
|
mckay
|
32d4e45196
|
修改预处理
|
11 months ago |
|
mckay
|
a7541c9da3
|
优化写法，兼容 torch
|
11 months ago |
|
mckay
|
b33203a7e6
|
fix: logger
|
11 months ago |
|
mckay
|
8fe07b2b3a
|
gitignore 还有config 修改
|
11 months ago |
|
mckay
|
845ad864b5
|
引入 NH-Rep 从 mesh 采样二面角的处理程序
|
11 months ago |
|
mckay
|
2e8dc3b0fe
|
可以直接从 step 开始处理的版本
|
11 months ago |
|
mckay
|
7f2049c4bd
|
加保存checkpoint
|
11 months ago |
|
mckay
|
584354bbf7
|
能在八叉树上面训练，但是速度很慢
|
11 months ago |
|
mckay
|
cabc98090e
|
BREAKPOINT: first attremp based on brepgen encoder and deepsef decoder
|
11 months ago |
|
mckay
|
0b4df54447
|
feat: filter the data if their face_num or edge_num > max_face or max_edge.
|
1 year ago |
|
mckay
|
4dddcdcd56
|
feat: diagnose script
|
1 year ago |
|
mckay
|
7c97b82fdf
|
feat: optimaize loss, by torch.norm and mead
|
1 year ago |
|
mckay
|
66fa9fcc58
|
fix: zero_grad mv afterwards
|
1 year ago |
|
mckay
|
86fb4c31e4
|
config: 修改
|
1 year ago |
|
mckay
|
e34d99a955
|
fix: loss 增加batch归一化
|
1 year ago |
|
mckay
|
6c76b66fb4
|
config: change batch and epoch
|
1 year ago |
|
mckay
|
0fa6a3ac3c
|
fix: neg sdf too small, -> shape error
|
1 year ago |
|
mckay
|
6d5c24c842
|
feat: script read npz
|
1 year ago |
|
mckay
|
de231aeb04
|
refactor: config
|
1 year ago |
|
mckay
|
bcc8ea8cf3
|
style: del log
|
1 year ago |
|
mckay
|
03fa4012a4
|
feat: 统计brep数据
|
1 year ago |
|
mckay
|
f81ca6040e
|
feat: SDF值分布log
|
1 year ago |
|
mckay
|
d9ecfbf4c7
|
style: 注释了不必要log
|
1 year ago |
|
mckay
|
959dafc4b1
|
refctor: print_tensor_stats 移入log
|
1 year ago |
|
mckay
|
41eb043d3f
|
fix: 梯度改在train里面加
|
1 year ago |
|
mckay
|
7f1ee89ba5
|
fix: del collate and load_brep
|
1 year ago |
|
mckay
|
ad58e0c12a
|
fix: 文件格式化器中使用了 caller_info 但没有正确设置
|
1 year ago |
|
mckay
|
e8e7f0e607
|
feat: log for param in debug mode
|
1 year ago |
|
mckay
|
530f680657
|
fix: network 测试脚本 tofix: some
norm =0
|
1 year ago |
|
mckay
|
ca2c3734d5
|
tofix： 概率性有梯度
|
1 year ago |
|
mckay
|
e1a76cf0ae
|
fix: can run, but still no grad
|
1 year ago |
|
mckay
|
7bfe8cad44
|
fix: grad escape
|
1 year ago |
|
mckay
|
b955b7d770
|
fix: git ignore 增加checkpoints/
|
1 year ago |
|
mckay
|
d8b5b545e2
|
fix:修复loss参数
|
1 year ago |
|
mckay
|
bf6f473edd
|
refactor: 在networks文件里面移除loss相关的东西
|
1 year ago |
|
mckay
|
9f9b5816eb
|
feat: test脚本
|
1 year ago |
|
mckay
|
608c6b75bf
|
feat: 完善log记录，可以显示报错位置
|
1 year ago |
|
mckay
|
0eb2063e28
|
refactor: 模型相关函数拆分
|
1 year ago |
|
mckay
|
49e0f8e9bc
|
fix: git ignore add *.pth
|
1 year ago |
|
mckay
|
df186deae6
|
fix: can train
|
1 year ago |
|
mckay
|
1110dc220e
|
fix: 由于设置的采样点太小，取消Encoder1D的池化操作
|
1 year ago |
