Thesis/pytorch-dnc
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

update readme

Browse Source
This commit is contained in:
ixaxaar 2017-12-17 18:33:53 +05:30
parent c7812efe49
commit 6517fb9585
1 changed files with 1 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@ -312,9 +312,7 @@ sudo make install
```
2. An alternative to FLANN is [FAISS](https://github.com/facebookresearch/faiss), which is much faster and interoperable with torch cuda tensors (but is difficult to distribute, see [dnc/faiss_index.py](dnc/faiss_index.py)).
3. DNCs converge faster with Adam and RMSProp learning rules, SGD generally converges extremely slowly.
The copy task, for example, takes 25k iterations on SGD with lr 1 compared to 3.5k for adam with lr 0.01.
4. `nan`s in the gradients are common, try with different batch sizes
3. `nan`s in the gradients are common, try with different batch sizes
Repos referred to for creation of this repo: