Text2text refactor (#78)

* text2text refactor and update robin_hood version

THIRD-PARTY-LICENSES.txt

@@ -68,7 +68,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 ------
 
-** Robin Hood Hashing; version 3.9.0 --
+** Robin Hood Hashing; version 3.11.3 --
 https://github.com/martinus/robin-hood-hashing
 Copyright (c) 2018-2019 Martin Ankerl
 

pecos/apps/text2text/README.md

@@ -105,6 +105,27 @@ python3 -m pecos.apps.text2text.predict \
   --predicted-output-item-path ./test-prediction.txt
 ```
 The predictions are saved in the `./test-prediction.txt`.
+
+### Advanced Usage: Train/Pred params via JSON input
+`pecos.apps.text2text` supports accepting training and predicting parameters from an input JSON file.
+Moreover, `python3 -m pecos.apps.text2text.train` helpfully provide the option to generate all parameters in JSON format to stdout.
+
+You can generate train/pred parameter to `.json` files
+with all of the parameters that you can edit and fill in.
+```bash
+  > python3 -m pecos.apps.text2text.train --generate-train-params-skeleton &> train_params.json
+  > python3 -m pecos.apps.text2text.train --generate-pred-params-skeleton &> pred_params.json
+```
+After editing the `train_params.json` and `pred_params.json` files, you can do training via:
+```bash
+  > python3 -m pecos.apps.text2text.train \
+	--input-text-path ./training-data.txt \
+	--vectorizer-config-path ./config.json \
+	--output-item-path ./output-labels.txt \
+	--model-folder ./pecos-text2text-model
+	--train-params-path train_params.json \
+	--pred-params-path pred_params.json
+```
 ***
 
 Copyright (2021) Amazon.com, Inc.