XL

app.py

@@ -7,15 +7,13 @@ device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cp
 
 class Interactive:
     def __init__(self):
-        print('Init invoked')
         self.tokenizer = transformers.AutoTokenizer.from_pretrained('liujch1998/cd-pi', use_auth_token=os.environ['HF_TOKEN_DOWNLOAD'])
-        self.model = transformers.AutoModelForSeq2SeqLM.from_pretrained('liujch1998/cd-pi', use_auth_token=os.environ['HF_TOKEN_DOWNLOAD']).to(device)
+        self.model = transformers.T5EncoderModel.from_pretrained('liujch1998/cd-pi', use_auth_token=os.environ['HF_TOKEN_DOWNLOAD']).to(device)
         self.linear = torch.nn.Linear(self.model.shared.embedding_dim, 1).to(device)
-        # self.linear.weight = torch.nn.Parameter(self.model.shared.weight[32099, :].unsqueeze(0)) # (1, D)
-        # self.linear.bias = torch.nn.Parameter(self.model.shared.weight[32098, 0].unsqueeze(0)) # (1)
+        self.linear.weight = torch.nn.Parameter(self.model.shared.weight[32099, :].unsqueeze(0)) # (1, D)
+        self.linear.bias = torch.nn.Parameter(self.model.shared.weight[32098, 0].unsqueeze(0)) # (1)
         self.model.eval()
         self.t = 2.2247
-        print('Init finished')
 
     def run(self, statement):
         input_ids = self.tokenizer.batch_encode_plus([statement], return_tensors='pt', padding='longest').input_ids.to(device)