Train encoder/decoder error

I get this error with train() while trying to figure out how to fine tune a translation model.

/usr/local/lib/python3.7/dist-packages/transformers/models/marian/ in forward(self, input_ids, attention_mask, encoder_hidden_states, encoder_attention_mask, head_mask, cross_attn_head_mask, past_key_values, inputs_embeds, use_cache, output_attentions, output_hidden_states, return_dict)
    964             input_shape = inputs_embeds.size()[:-1]
    965         else:
--> 966             raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")
    968         # past_key_values_length

ValueError: You have to specify either decoder_input_ids or decoder_inputs_embeds

Here the googlecolab, can someone please tell me what’s wrong with the code?

Please, someone can help? I really need this :smiling_face_with_tear: