Update README.md
Browse files
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 |
-
, which is not *exactly* linear
|
| 23 |
|
| 24 |
+

|
| 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 |
|