•1 min read•from Machine Learning
Was looking at a ICLR 2025 Oral paper and I am shocked it got oral [D]
After my last post about score analysis of ICLR, I am looking into the review itself now.
They evaled SQL code generation by LLM using nature language metric and not executation metric, and they tested it and found around 20% false positive rate. This is a major flaw how is it even getting oral?
[link] [comments]
Want to read more?
Check out the full article on the original site
Tagged with
#rows.com
#natural language processing for spreadsheets
#AI formula generation techniques
#generative AI for data analysis
#conversational data analysis
#Excel alternatives for data analysis
#no-code spreadsheet solutions
#natural language processing
#data analysis tools
#ICLR
#oral paper
#false positive rate
#SQL code generation
#score analysis
#LLM
#natural language metric
#execution metric
#evaluation
#flaw
#testing