We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 3ec9305 commit 19d17c3Copy full SHA for 19d17c3
CONTRIBUTING.md
@@ -0,0 +1,10 @@
1
+# Code conventions
2
+
3
+* use black to lint the code
4
+* default values for Part/Assembly config:
5
+ * defaults for common hyper-parameters (e.g. number of layers): no
6
+ * defaults for rarely/never tuned hyper-parameters (e.g. optimizer constants): maybe
7
+ * defaults for structural elements (parts / assemblies / PyTorch modules/functions): yes
8
+ * if a PR contains a default (not covered by the third point) there should be a reason given for it in the PR / code
9
+* prefer to use torch.nn.functional over equivalent module unless it’s required for torch.nn.Sequential
10
+* prefer to do checks in the config class if feasible, if checks are not in the config there should be a reason given in the PR / code
0 commit comments