update README

README.md

@@ -74,9 +74,10 @@ The training split of `nitibench-ccl` can be found in the [WangchanX-Legal-ThaiC
 
 Each data point contains four columns:
 - `question: str` — A question relevant to the `relevant_laws`.
+- `answer_full: str` — The original answer generated by an LLM, which has been revised and edited by legal experts to include both the reasoning steps and the final answer.
+- `relevant_laws: List[Dict[str, str]]` — A list of relevant law name, section, and contents.
 - `answer: str` — The answer to the question based on the `relevant_laws`, provided without the reasoning steps.
-- `relevant_laws: List[Dict[str, str]]` — A list of relevant laws that should be used as context when answering the question. **The value is formatted as string since Huggingface's dataset won't allow parsing nested dictionary. Make sure you parse this to dictionary for later use**.
-- `reference_answer: str` — The original answer generated by an LLM, which has been revised and edited by legal experts to include both the reasoning steps and the final answer.
+- `reference_laws: List[Dict[str, str]]` - A list of referenced law mentioned in `relevant_laws` column.
 
 Formally, given the data triple \((q, T=\{p_1, p_2, \dots, p_K\}, y)\), \(q\) represents the `question`, \(T\) represents `relevant_laws`, and \(y\) represents the `answer`.