FinLab Desktop 0.1.18
Fixes Python memory accumulating across repeated AI strategy optimizations.
- Each completed Python program now releases its kernel and cached data. Concurrent Python requests run in sequence to reduce overlapping memory use.
- App shutdown now waits for agent and Python cleanup.
- In a controlled replay of 15 real AI-generated programs, peak desktop memory fell from 5.72 GiB to 3.88 GiB, and memory after the final run fell from 5.49 GiB to 0.74 GiB. Actual usage still depends on the strategy and data.
Each Python execution is self-contained: rerun the full program when it needs earlier variables. Notebook files remain available.