Update README.md

README.md

@@ -18,7 +18,7 @@ example_title : "A simple sentence"
 
 # t2p-t5-large-orféo
 
-T2P-t5-large-orféo is a text-to-pictograms translation model built by fine-tuning the [t5-large](https://huggingface.co/google-t5/t5-large) model on a dataset of pairs of transcriptions / pictogram token sequence (each token is linked to a pictogram image from [ARASAAC](https://arasaac.org/)).
+*t2p-t5-large-orféo* is a text-to-pictograms translation model built by fine-tuning the [t5-large](https://huggingface.co/google-t5/t5-large) model on a dataset of pairs of transcriptions / pictogram token sequence (each token is linked to a pictogram image from [ARASAAC](https://arasaac.org/)).
 
 ## Training details
 
@@ -37,6 +37,25 @@ T2P-t5-large-orféo is a text-to-pictograms translation model built by fine-tuni
 ## Using t2p-t5-large-orféo model with HuggingFace transformers
 
 ```python
+import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import numpy as np
+
+source_lang = "fr"
+target_lang = "frp"
+max_input_length = 128
+max_target_length = 128
+
+def load_model(checkpoint):
+    tokenizer = AutoTokenizer.from_pretrained(checkpoint)
+    model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint)
+    model = model.to("cuda:0")
+    return tokenizer, model
+
+def generate(sentence, tokenizer, model):
+    inputs = tokenizer("Je mange une pomme", return_tensors="pt").input_ids
+    outputs = model.generate(inputs.to("cuda:0"), max_new_tokens=40, do_sample=True, top_k=30, top_p=0.95)
+    pred = tokenizer.decode(outputs[0], skip_special_tokens=True)
 ```
 
 - **Language(s):** French