Awni Hannun be98f4ab6b Reduce a little overhead (#871) ...

* some small overhead improvements

* use result_type in rms_norm

* remove release force

* fix + use non-vector version

* revert compile change

* fix ops

* a little more overhead

* a little more cleanup and overhead

2024-03-22 17:29:36 -07:00

..

__init__.py

Implement RNN, GRU, LSTM (#268)

2024-03-11 21:14:44 -07:00

activations.py

Add back compiled function signatures and docstrings (#749)

2024-02-27 13:18:59 -08:00

base.py

Reduce a little overhead (#871)

2024-03-22 17:29:36 -07:00

containers.py

copyright + ack

2023-11-30 11:12:53 -08:00

convolution.py

Dilation for convolutional layers (#766)

2024-03-04 06:43:00 -08:00

dropout.py

Make shape a tuple (#591)

2024-01-30 13:11:01 -08:00

embedding.py

updated calls to use loc &scale (#643)

2024-02-08 09:01:59 -08:00

linear.py

Reduce a little overhead (#871)

2024-03-22 17:29:36 -07:00

normalization.py

Reduce a little overhead (#871)

2024-03-22 17:29:36 -07:00

pooling.py

Use reshape and transpose for non-overlapping pooling windows (#867)

2024-03-21 10:21:03 -07:00

positional_encoding.py

No reshape rope (#838)

2024-03-18 17:03:07 -07:00

quantized.py

Reduce a little overhead (#871)

2024-03-22 17:29:36 -07:00

recurrent.py

Implement RNN, GRU, LSTM (#268)

2024-03-11 21:14:44 -07:00

transformer.py

Change the transformer to norm_first by default (#599)

2024-01-31 12:55:30 -08:00

upsample.py

Upsample2d (#414)

2024-02-23 09:55:04 -08:00