Tpot 2 — Fla

Set max_eval_time_mins lower than you think. Because FLA evaluates pipelines faster, you want to kill poorly performing pipelines quickly to allow genetic diversity.

| Metric | Standard TPOT 2 | TPOT 2 FLA | | :--- | :--- | :--- | | | 3,420 seconds | 1,134 seconds | | Peak RAM Usage | 11.2 GB | 4.1 GB | | Final Pipeline CV Score | 0.823 | 0.835 | | Disk I/O Overhead | High | Minimal (memory-mapped) | Tpot 2 Fla

If you have been frustrated by TPOT taking days to run on a standard dataset, offers a lifeline. Here are the defining features: Set max_eval_time_mins lower than you think