just quickly tidying up READMEs, notebooks and adds zenodo links of d…

…emodata and merge #15

.gitignore

@@ -35,8 +35,8 @@
 **/main.pdf
 
 ## temp models
-**/best_model.h5
-**/best_transforer_model.h5
+**/best_model*.h5
+**/best_transforer_model*.h5
 
 ## others
 *.ipynb_checkpoints

data/README.md

@@ -1,5 +1,9 @@
 # Data
 
+Demo datasets (1.3 GB) can be downloaded from zenodo: https://doi.org/10.5281/zenodo.10775099.
+See further details [here](../docs/protocols/experiment-24-aug-2023/README.md) on its collection.
+Notes. you can host up to 50GB of data in zenodo. 
+
 ## Demo data Thu-24-Aug-2023
 Tree of demo dataset. Each pair of video and time-series were recorded for approximately 5 minutes. 
 ```
@@ -92,3 +96,7 @@ Tree of demo dataset. Each pair of video and time-series were recorded for appro
 
 2 directories, 20 files
 ```
+
+## Citations 
+
+Hekim, Sujon, and Miguel Xochicale. ‘Demo Dataset for Multimodal Real-time Ai V.1.0.0’. Zenodo, 3 March 2024. https://doi.org/10.5281/zenodo.10775099.

rtt4ssa/models/example-models/README.md

@@ -0,0 +1,96 @@
+# Example models for timeseries 
+
+## `timeseries_classification_from_scratch.ipynb`
+
+epochs = 500 ## ORIGINAL
+batch_size = 32 ##ORIGINAL
+
+
+```
+=================================================================
+Total params: 25,858
+Trainable params: 25,474
+Non-trainable params: 384
+```
+
+```
+Epoch 256/500
+90/90 [==============================] - 1s 11ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0865 - val_sparse_categorical_accuracy: 0.9722 - lr: 1.0000e-04
+Epoch 256: early stopping
+Execution time: 4.0526354789733885 minutes 
+```
+
+![fig](Screenshot%20from%202023-10-10%2002-20-17.png)
+
+
+## `timeseries_transformer_classification.ipynb`
+
+### head_size=256; num_heads=4; num_transformer_blocks = 2
+```
+==================================================================================================
+Total params: 78,758
+Trainable params: 78,758
+Non-trainable params: 0
+```
+
+```
+Epoch 77/200
+360/360 [==============================] - 14s 38ms/step - loss: 0.0783 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.4438 - val_sparse_categorical_accuracy: 0.8350
+Execution time: 16.701254685719807 minutes 
+```
+
+![fig](Screenshot%20from%202023-10-10%2002-53-14.png)
+
+
+### head_size=256; num_heads=4; num_transformer_blocks = 4 ##ORIGINAL
+
+```
+=================================================================
+Total params: 93,130
+Trainable params: 93,130
+Non-trainable params: 0
+```
+
+```
+Epoch 106/200
+Execution time: 43.552278610070545 minutes 
+```
+
+![fig](Screenshot%20from%202023-10-10%2001-54-35.png)
+
+
+### head_size=256; num_heads=4; num_transformer_blocks = 8
+
+```
+=================================================================
+Total params: 121,874
+Trainable params: 121,874
+Non-trainable params: 0
+```
+
+```
+Epoch 97/200
+360/360 [==============================] - 48s 134ms/step - loss: 0.0737 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.3841 - val_sparse_categorical_accuracy: 0.8530
+Execution time: 78.16839495897293 minutes 
+```
+
+![fig](Screenshot%20from%202023-10-10%2001-54-43.png)
+
+
+### head_size=512; num_heads=4; num_transformer_blocks = 4
+```
+
+=================================================================
+Total params: 121,802
+Trainable params: 121,802
+Non-trainable params: 0
+```
+
+```
+Epoch 84/200
+360/360 [==============================] - 38s 107ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.4588 - val_sparse_categorical_accuracy: 0.8516
+Execution time: 54.10847853024801 minutes 
+```
+
+![fig](Screenshot%20from%202023-10-10%2001-54-57.png)
+