Good afternoon,
I’ve headaches everytime I upload an Excel to ChatGPT, and ask it to do some actions about it.
As an example:
- XLSX with like 1000 rows. First column is group ID, and data comes ordered by this very column.
- columns basically are text descriptions. Short titles, Long titles, short descriptions and long ones. Multi-languages: english, spanish, french…, so alot of columns.
So, you upload file while prompting “him” to homogenize texts on specific columns. The idea? given a row group, standarize X column descriptions deleting anything related with colours, translating to column language if needed and as a result, all group rows on this X column, have to have same texts on cells.
Same behaviour by all specified columns.
If you test it on “sceen”, directly to the chat through an small test group, works like expected. For every column, gets the texts, homogenize them, deletes color references and rewrites descriptions or titles to be the same on every row and to be “perfect” as you would expect.
BUT, if it enters Analyzer mode, then you’re dead. Because it tries via Python to do what you will aproximetly do in code too: regex, replacing, trims, blabla… so “no intelligence” at all. Just blablah code. Wich gives a result, it’s true. But far from what it can do “on LLM mode on screen”.
Any idea on how to make it behaviour like on screen while on Analyzer mode?
Thanks.