GitLab

-added the binary and fp tflite model used in the demonstration

-Changed the iridis setup file to see if later versions of tensorflow work

-I can generate a tflite model! you just need to build a new environment with TF 2.5 because nothing else works apart from that... HORAH! Such Tensorflow. Much Unreliable.

-Also removed the early stopping from the training (Model_CH_12.py)
-some alterations with Model_CH_13.py which didnt work

-Added the Val and Test loss into the metrics

-Added model 13 which is a FP CNN with custom adam optimiser

-Updated the Evaluate_TUT2017.py file with local file path parameters

-Model 12 is suppose to reproduce Jing's binary improvements.

-Bug in Model_CH_08.py - the FP CNN pool size was not taken into account in some of the layer size calculations which were used later on in the model description

- Binarised CNN
- Updated run_on_iridis.sh script

-Bug in Model_CH_08.py - the FP CNN pool size was not taken into account in some of the layer size calculations which were used later on in the model description

- Binarised CNN
- Updated run_on_iridis.sh script

-Bug in Model_CH_08.py - the FP CNN pool size was not taken into account in some of the layer size calculations which were used later on in the model description