replace torch.triu with onnx compatible code (#6929)

src/transformers/modeling_bart.py

@@ -154,9 +154,10 @@ def _prepare_bart_decoder_inputs(
     if decoder_padding_mask is not None and decoder_padding_mask.shape[1] > 1:
         # never mask leading token, even if it is pad
         decoder_padding_mask[:, 0] = decoder_padding_mask[:, 1]
-    causal_mask = torch.triu(fill_with_neg_inf(torch.zeros(tgt_len, tgt_len)), 1).to(
-        dtype=causal_mask_dtype, device=decoder_input_ids.device
-    )
+    tmp = fill_with_neg_inf(torch.zeros(tgt_len, tgt_len))
+    mask = torch.arange(tmp.size(-1))
+    tmp.masked_fill_(mask < (mask + 1).view(tmp.size(-1), 1), 0)
+    causal_mask = tmp.to(dtype=causal_mask_dtype, device=decoder_input_ids.device)
     return decoder_input_ids, decoder_padding_mask, causal_mask