Open hyx0324 opened 3 years ago
May the code "new_state = (new_state, new_attns, new_attn_states)" should be written to "new_state = (new_state, output, new_attn_states)"?
May the code "new_state = (new_state, new_attns, new_attn_states)" should be written to "new_state = (new_state, output, new_attn_states)"?