Modify HFTokenzer to enable tokenzie with truncation and padding.
Custmize tokenize function of both HFTokenzer and HFGPT2Tokenizer to enable text_token ids followed with eos_id and padding id. If the test_tokens ids length reach out seq_length, then replace the last id with eos_id
Move seq_length from variable exposed to tokenzer.tokenze() function to tokenzer class attribute self.seq_length
Add support to HFTokenzer with following changes: