 mckay
|
b5928becf7
|
fix: neg sdf too small, -> shape error
|
11 months ago |
|
mckay
|
59f17a7199
|
feat: script read npz
|
11 months ago |
|
mckay
|
130735bae1
|
refactor: config
|
11 months ago |
|
mckay
|
97e0cf9699
|
style: del log
|
11 months ago |
|
mckay
|
f930bc17ad
|
feat: 统计brep数据
|
11 months ago |
|
mckay
|
200c617d86
|
feat: SDF值分布log
|
11 months ago |
|
mckay
|
8f0e108696
|
style: 注释了不必要log
|
11 months ago |
|
mckay
|
5eef6c4290
|
refctor: print_tensor_stats 移入log
|
11 months ago |
|
mckay
|
de0cf41d59
|
fix: 梯度改在train里面加
|
11 months ago |
|
mckay
|
2e124b7193
|
fix: del collate and load_brep
|
11 months ago |
|
mckay
|
feb49e71fb
|
fix: 文件格式化器中使用了 caller_info 但没有正确设置
|
11 months ago |
|
mckay
|
424ae34067
|
feat: log for param in debug mode
|
11 months ago |
|
mckay
|
51db266834
|
fix: network 测试脚本 tofix: some
norm =0
|
11 months ago |
|
mckay
|
596a7b6565
|
tofix： 概率性有梯度
|
11 months ago |
|
mckay
|
fa81eefa54
|
fix: can run, but still no grad
|
11 months ago |
|
mckay
|
b0be9a26f9
|
fix: grad escape
|
11 months ago |
|
mckay
|
5517de4e9a
|
fix: git ignore 增加checkpoints/
|
12 months ago |
|
mckay
|
b2b42918f4
|
fix:修复loss参数
|
12 months ago |
|
mckay
|
8f232a2103
|
refactor: 在networks文件里面移除loss相关的东西
|
12 months ago |
|
mckay
|
5d0d76f5eb
|
feat: test脚本
|
12 months ago |
|
mckay
|
519ad08b67
|
feat: 完善log记录，可以显示报错位置
|
12 months ago |
|
mckay
|
5b98c59270
|
refactor: 模型相关函数拆分
|
12 months ago |
|
mckay
|
0187ae991d
|
fix: git ignore add *.pth
|
12 months ago |
|
mckay
|
229c0aeffc
|
fix: can train
|
12 months ago |
|
mckay
|
c5bbc29f6a
|
fix: 由于设置的采样点太小，取消Encoder1D的池化操作
|
12 months ago |
|
mckay
|
94678c5c17
|
todo: 配置修改
|
12 months ago |
|
mckay
|
2b5860a7e7
|
增加sdf dataloader随机采样
|
12 months ago |
|
mckay
|
b14d7455cb
|
fix :trian
|
12 months ago |
|
mckay
|
eff305e802
|
refactor： config配net
|
12 months ago |
|
mckay
|
2269a77f6a
|
TODO: 太大加载不了
|
12 months ago |
|
mckay
|
8ee97699b7
|
fix ：BRepFeatureEmbedder
|
12 months ago |
|
mckay
|
a586f2e534
|
fix: sdf_xyz和sdf_value 分开加载
|
12 months ago |
|
mckay
|
a2cb2c6e8a
|
fix:使用size()而不是shape来获取维度
|
12 months ago |
|
mckay
|
f3e3886f0e
|
fix :train script
|
12 months ago |
|
mckay
|
a8fdaa9583
|
fix: ignore wandb files
|
12 months ago |
|
mckay
|
24ffaac722
|
fix: config 路径使用绝对路径
|
12 months ago |
|
mckay
|
a3206cb0e8
|
fix : config refactor log & details separate
|
12 months ago |
|
mckay
|
3555cf5359
|
refactor & style： 用config配置train，同时增加保存checkpoint的信息
|
12 months ago |
|
mckay
|
40bb28cfe7
|
fix: net dim问题。num*3,变成[num,3],保留空间特征
|
12 months ago |
|
mckay
|
84fc447adb
|
refactor: 用config替代硬编码
|
12 months ago |
|
mckay
|
c82aef0a33
|
refactor: process_brep，用config替代原来的100边采样
|
12 months ago |
|
mckay
|
8db42d1ef7
|
feat: 增加全局config
|
12 months ago |
|
mckay
|
4180ad0eae
|
feat: train 脚本
|
12 months ago |
|
mckay
|
3053eda7fd
|
test： 修改数据集加载测试
|
12 months ago |
|
mckay
|
85a581be17
|
fix: 修改数据加载process_brep_data，防止维度超过最大（用截取）
|
12 months ago |
|
mckay
|
9e6020adaf
|
修复: 改进数据预处理中的数组填充逻辑
1. 修复了pad_zero函数中的负数维度问题
- 添加了对超长输入的截断处理
- 优化了掩码生成逻辑
- 保持多维数组的维度一致性
2. 主要改动：
- 处理数组长度 > max_len：进行截断
- 处理数组长度 < max_len：进行填充
- 处理数组长度 = max_len：直接返回
- 改进布尔掩码的生成方式
|
12 months ago |
|
mckay
|
c03a08a13e
|
fix: 数据集可以加载
|
12 months ago |
|
mckay
|
ee47a5f682
|
fix: 修改了check_data_format函数，正确检查对象数组类型
在parse_solid函数中确保所有数组都有正确的类型
对点云数据使用对象数组类型
对其他数据使用正确的数值类型(float32或int32)
|
12 months ago |
|
mckay
|
333c2214cb
|
fix: corners维度变成[num_faces, max_edge, 2, 3]
|
12 months ago |
|
mckay
|
7b1484aed4
|
feat: log 不同level报错
|
12 months ago |
