the prompt should be refined

[tongsir]

to you handsome guys.
the prompt in this demo maybe not working as the author said, the output is excessive because of bad format of the prompt, as mentioned here, and i followed that answer, using huggingface-chat-template, and the output is ok.

below is my solution.

>>> messages = [
...     {"role": "system", "content": "Convert the following SQL written in Oracle dialect to MySQL dialect. only SQL is allowed in response"},
...     {"role": "user", "content": "SELECT col1, TRUNC (col2/20) 'c2', TO_DATE(SYSDATE) 'c3' FROM ttt WHERE col3=4"},
... ]
>>>
>>>
>>> tokenized_chat = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt")
>>> print(tokenizer.decode(tokenized_chat[0]))
<s> [INST] <<SYS>>
Convert the following SQL written in Oracle dialect to MySQL dialect. only SQL is allowed in response
<</SYS>>

SELECT col1, TRUNC (col2/20) 'c2', TO_DATE(SYSDATE) 'c3' FROM ttt WHERE col3=4 [/INST]
>>> outputs = model.generate(tokenized_chat, max_new_tokens=1280)
The attention mask and the pad token id were not set. As a consequence, you may observe unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results.
Setting `pad_token_id` to `eos_token_id`:None for open-end generation.
>>> print(tokenizer.decode(outputs[0]))
<s> [INST] <<SYS>>
Convert the following SQL written in Oracle dialect to MySQL dialect. only SQL is allowed in response
<</SYS>>

SELECT col1, TRUNC (col2/20) 'c2', TO_DATE(SYSDATE) 'c3' FROM ttt WHERE col3=4 [/INST]  SELECT col1, TRUNC(col2/20) AS c2, CURDATE() AS c3 FROM ttt WHERE col3=4;</s>