RuntimeError: The size of tensor a (1024) must match the size of tensor b (1060) at non-singleton dimension 3

[mominabbass]

For DBpedia 8-shot on GPT-2, I incur a warning "token indices sequence length is longer than the specified maximum sequence length" along with an error "RuntimeError: The size of tensor a (1024) must match the size of tensor b (1060) at non-singleton dimension 3" on line 81 of the utils.py file where the function gpt2_model.generate() is called.

One possible Solution: In the current version of the code, we do not have a check when we're encoding a sequence that is larger than the max sequence the GPT-2 model can handle (which is 1024 tokens). If you pass that sequence to the model it will crash as it cannot handle such a long sequence.

You can truncate the sequence: seq = seq[:1023] or use the max_length tokenizer parameter so that it handles it on its own.:
if (len(input_ids['input_ids'][0]) > 1023):
input_ids['input_ids'] = input_ids['input_ids'][:, :1023]
input_ids['attention_mask'] = input_ids['attention_mask'][:, :1023]

Not sure if this is an optimal way to circumvent this issue. I'd appreciate if you could help. Thank you.