jm

@lomz.me

34 Followers 239 Following

the wayne gretzky of refactoring code he/him

Python service writing eval which shows Haiku 4.5, Sonnet 4.5, and GLM-4.7 from Z.AI. While GLM-4.7 is at Sonnet 4.5 levels of performance on some measures, GLM-4.7 does not consistently follow the instructions the skill file.
Python service writing evaluation - Haiku averaged 88.9% across all checks - Sonnet achieved 100.0% across all checks
Stuffed pastry cut open stuffed with some kind of ube paste