Ftr/text vectorization ext

[tommylau-exe]

Fixes #36

Perform text vectorization processes externally from Tensorflow.

[tommylau-exe]

Prematurely updated model, undone with latest force-push

[tommylau-exe]

Updated model so that Tokenization file (tokenizer.json) is accurate.

Metrics

• Categorical Accuracy: 83%
• Precision: 0.84
• Recall: 0.82

Confusion Matrix on held-out set

	Predicted Low-Calorie	Predicted Average-Calorie	Predicted High-Calorie
Actual Low-Calorie	11096	1601	648
Actual Average-Calorie	1673	12788	2264
Actual High-Calorie	782	2298	21196

[tommylau-exe]

Updated model due to further tokenization changes.

Metrics

• Categorical Accuracy: 83%
• Precision: 0.84
• Recall: 0.82

Confusion Matrix on held-out set

	Predicted Low-Calorie	Predicted Average-Calorie	Predicted High-Calorie
Actual Low-Calorie	11185	1590	609
Actual Average-Calorie	1612	13053	2166
Actual High-Calorie	657	2471	21003