commit 753fe9c2f5519447a7c9ed5b19596a1af9751314
parent 1ecbeed0232d9c556ca455273a051eb37d94979b
Author: Keith Bloemer <32459398+GuitarML@users.noreply.github.com>
Date: Sat, 5 Dec 2020 08:08:12 -0600
Update README.md
Diffstat:
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/README.md b/README.md
@@ -88,9 +88,10 @@ Helpful tips on training models:
This implementation of the LSTM model uses a high amount of
RAM to preprocess wav data. If you experience crashes due to
limited memory, reduce the "input_size" parameter by using
-the "--input_size=" flag with train.py. The default setting is 100.
-Increasing this setting will improve training accuracy, but the size
-of the preprocessed wav data in RAM will increase as well.
+the "--input_size=" flag with train.py. The default setting is 100,
+which requires about 8GB of RAM. Increasing this setting will improve
+training accuracy, but the size of the preprocessed wav data in
+RAM will increase as well.
Adding a custom dataloader would reduce RAM usage at the cost of
training speed, and will be a focus of future work.