fix: change logits to sigmoid

Author: monologg

run_goemotions.py

@@ -179,10 +179,10 @@ def evaluate(args, model, eval_dataset, mode, global_step=None):
             eval_loss += tmp_eval_loss.mean().item()
         nb_eval_steps += 1
         if preds is None:
-            preds = logits.detach().cpu().numpy()
+            preds = 1 / (1 + np.exp(-logits.detach().cpu().numpy()))  # Sigmoid
             out_label_ids = inputs["labels"].detach().cpu().numpy()
         else:
-            preds = np.append(preds, logits.detach().cpu().numpy(), axis=0)
+            preds = np.append(preds, 1 / (1 + np.exp(-logits.detach().cpu().numpy())), axis=0)  # Sigmoid
             out_label_ids = np.append(out_label_ids, inputs["labels"].detach().cpu().numpy(), axis=0)
 
     eval_loss = eval_loss / nb_eval_steps