OldestSalt commited on
Commit
5deade6
·
verified ·
1 Parent(s): 74b80c8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -21,7 +21,7 @@ Bimba is *almost* linear SimulMT model trained with wait-k policy (k = 3, 5, 7,
21
 
22
  The model has encoder-decoder architecture, where self-attention blocks are Mamba-2 blocks instead. It means that encoder is linear, but cross-attention's input is all outputs of encoder, and this means that complexity of Bimba is O(S * T), which is not *exactly* linear
23
 
24
- ![Bimba inference](https://huggingface.co/OldestSalt/Bimba/resolve/main/hybrid.png)
25
 
26
  Bimba was developed and trained as a part of master's thesis, and I hope that I will continue research in the Linear SimulMT field.
27
 
 
21
 
22
  The model has encoder-decoder architecture, where self-attention blocks are Mamba-2 blocks instead. It means that encoder is linear, but cross-attention's input is all outputs of encoder, and this means that complexity of Bimba is O(S * T), which is not *exactly* linear
23
 
24
+ ![Bimba inference](https://huggingface.co/OldestSalt/Bimba/resolve/main/bimba.png)
25
 
26
  Bimba was developed and trained as a part of master's thesis, and I hope that I will continue research in the Linear SimulMT field.
27