Update app.py

app.py

@@ -27,7 +27,7 @@ def inference(text, seq_length=1):
             h = F.linear(h, weight=model.transformer.word_embeddings.weight)  # note: this line takes a while, will also be fixed
             next_token_ix = torch.multinomial((h[0, -1] / 0.8).softmax(-1), 1)
         
-            final_tokens = torch.cat([final_tokens, next_token_ix], dim=-1)
+            final_tokens = torch.cat([final_tokens, next_token_ix.view(1, 1)], dim=-1)
             input_ids = next_token_ix.view(1, 1)
         return tokenizer.decode(final_tokens, skip_special_tokens=False)