Update README.md

README.md

@@ -9,7 +9,7 @@ tags:
 - PyTorch
 - Safetensors
 widget:
-- text: 'translate spanish to nahuatl: Quiero agua. or translate nahuatl to spanish: Nimitstlazohkamate.'
+- text: 'translate nahuatl to spanish: Nimitstlazohkamate'
 ---
 
 # mt5-large-spanish-nahuatl
@@ -19,6 +19,14 @@ Nahuatl is the most widely spoken indigenous language in Mexico, yet training a
 ## Model description
 This model is an MT5 Transformer ([mt5-large](https://huggingface.co/google/mt5-large)) fine-tuned on Spanish and Nahuatl sentences collected from diverse places online. The dataset is normalized using 'inali' normalization from [py-elotl](https://github.com/ElotlMX/py-elotl).
 
+## Inference API use
+You can change and translate to nahuatl or spanish just do:
+
+translate spanish to nahuatl: Quiero agua
+
+or 
+
+translate nahuatl to spanish: Nimitstlazohkamate
 
 ## Usage
 ```python
@@ -71,12 +79,12 @@ Also, additional 30,000 samples were collected from the web to enhance the data.
 The employed method uses a single training stage using the mt5. This model was leveraged given that it can handle different vocabularies and prefixes.
 
 ### Training
-The model is trained till convergence, adding the prefixes "translate spanish to nahuatl: + word" and "translate nahuatl to spanish: + word".
+The model is trained bidirectionally till convergence, adding the prefixes "translate spanish to nahuatl: + word" and "translate nahuatl to spanish: + word".
+This is meant as an improvement of [previous models](https://huggingface.co/hackathon-pln-es/t5-small-spanish-nahuatl).
 
 ### Training setup
 The model uses the same dataset for 77,500 steps using batch size = 4 and a learning rate of 1e-4.
 
-
 ## Evaluation results
 The models are evaluated on 2 different datasets: 
 1. First on the test sentences similar to the evaluation ones.