Draft: Implements Encoder-Decoder Attention Model - #28

Open

mmz33 wants to merge 22 commits into

mainfrom

zeineldeen_att_decoder

Commits on Oct 15, 2024

implemented enc-dec-att model

mmz33
authored and
JackTemaki
committed
3476c07
fix docs

mmz33
authored and
JackTemaki
committed
b06aac0
fix docs

mmz33
authored and
JackTemaki
committed
3f5ada0
better check

mmz33
authored and
JackTemaki
committed
4378fb9
better comment

mmz33
authored and
JackTemaki
committed
085f9f3
better comment

mmz33
authored and
JackTemaki
committed
523ce6c
refactor + implement zoneout

mmz33
authored and
JackTemaki
committed
396a664
implement zoneout lstm cell

mmz33
authored and
JackTemaki
committed
9c22aa2
fix

mmz33
authored and
JackTemaki
committed
c4d5710
put tensors on cuda

mmz33
authored and
JackTemaki
committed
cd366ab
make device configurable

mmz33
authored and
JackTemaki
committed
d0ed59b
check for cuda availability

mmz33
authored and
JackTemaki
committed
5a78e40
remove explicit device, fix return value
JackTemaki
committed
163315a
remove device from attention test
JackTemaki
committed
6d200eb
remove leftover print statement
JackTemaki
committed
1b8dd52
Add shift_embeddings flag for Attention Decoder
JackTemaki
committed
dcf0381
fix logits dropout bug

mmz33
authored and
JackTemaki
committed
6a147b7

Commits on Oct 24, 2024

Update i6_models/decoder/attention.py

JackTemaki
and
Atticus1806
authored
a988e85
Disable autocast for ZoneoutLSTM cell
JackTemaki
committed
bb8fa4e

Commits on Jul 9, 2026

Commits on Jul 31, 2026

Formatting
SimBe195
committed
92b7a68