inference fonctionnelle sans load de fichier

src/inference.py

@@ -8,7 +8,7 @@ import torch
 from src import dataloader
 from src.model import Decoder, Encoder, EncoderDecoderModel
 
-with open ("model/vocab.pkl", 'rb') as vocab:
+with open("model/vocab.pkl", "rb") as vocab:
     words = pickle.load(vocab)
 vectoriser = dataloader.Vectoriser(words)
 
@@ -27,12 +27,10 @@ def inferenceAPI(text: str) -> str:
     text = text.split()
     # On défini les paramètres d'entrée pour le modèle
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-    encoder = Encoder(len(vectoriser.idx_to_token) + 1, 256, 512, 0.5, device).to(
-        device
-    )
-    decoder = Decoder(len(vectoriser.idx_to_token) + 1, 256, 512, 0.5, device).to(
-        device
-    )
+    encoder = Encoder(len(vectoriser.idx_to_token) + 1, 256, 512, 0.5, device)
+    encoder.to(device)
+    decoder = Decoder(len(vectoriser.idx_to_token) + 1, 256, 512, 0.5, device)
+    decoder.to(device)
 
     # On instancie le modèle
     model = EncoderDecoderModel(encoder, decoder, vectoriser, device)
@@ -49,7 +47,7 @@ def inferenceAPI(text: str) -> str:
     with torch.no_grad():
         output = model(source).to(device)
         output.to(device)
-        output=output.argmax(dim=-1)
+        output = output.argmax(dim=-1)
     return vectoriser.decode(output)