search

rllab-snu / TopologicalSemanticGraphMemory

Topological Semantic Graph Memory for Image Goal Navigation (CoRL 2022 oral)
MIT License
109 stars 8 forks source link

Questions about the code of cross graph mixer. #13

Open cxr0726 opened 12 months ago

cxr0726 commented 12 months ago

Hi,

Thanks for the excellent work!

I'm a little confused about the code of the cross graph mixer part. According to my understanding, the core code of this part is in model/policy/TSGMPolicy/perception.py, as shown below：

屏幕截图 2023-12-07 151129 In the code, I fail to find the cross update function C (corresponding to equations 8 and 9 of the supplementary material). In my opinion, lines 86 and 87 in the code correspond to the self-update function S using Graph Attention Network. Lines 90 and 92 seem to correspond to the update function U (equations 10 and 11 in the supplementary material). And the update process goest through 2 iterations.

Therefore I am confused about the implementation of cross update function C. I don't know if my understanding is correct. If there are any mistakes, please help me point them out.

Thanks for your help!

image