TextAI
Q&A and Extract Data from Text
Last updated
Q&A and Extract Data from Text
Last updated
This tool can be used to ask question and or extract data from a piece of text. The operations will be executed with LLM models..
The query can also be the output of a previous cell or any column in the previous cell.
To apply the query to a column, use @column name. The operation will be then executed on each row of this column and a new column will be added to the output.
There are 2 run types for this tool: Per Row and Full Table.
Per Row: For each of the rows of the referenced column (if any), a prompt will be sent and the answers will be returned per row.
Full Table: All of the values in the referenced column (if any) will be concatenated altogether, and only one prompt will be sent and there will only be one answer.
If your reference is a table, you can manipulate each row individually by selecting Per Row, and if you want to process all rows at once, or if the reference data is text or something else, select Full Table.
In this example, the answer column contains text with the answer to a requirement:
Explain the process of shape optimization in local optimization using the gradient method in this paper. Please answer this question, including the highlighted formula in LaTeX code.
This cell also utilizes the information in the text column of previous cell (by typing From @2.text).
TextAI can also be applied directly to the text output from the previous cell. In this case, you can simply refer to the text above in the query like the following. Full Table is chosen for this example. If it's a column reference and Full Table is chosen, the app will concatenate the values of each row together and use the concatenated result to run the query.
