SparkID: Fast, sortable, compact unique IDsyoussefm.com3 points·by pinkbeanz·vor 4 Monaten·1 comments
Evaluating and Optimizing LLM Applications with DSPypedramnavid.com2 points·by pinkbeanz·vor 10 Monaten·0 comments
I tried DSPy and now I get why everyone won't shut up about itpedramnavid.com3 points·by pinkbeanz·vor 11 Monaten·1 comments
Migrate connectors from MIT to ELv2 – Pull Request #63723 – airbytehq/airbytegithub.com2 points·by pinkbeanz·vor 11 Monaten·0 comments
Dagster saved $40k on ELT spend with embedded-eltdagster.io2 points·by pinkbeanz·vor 2 Jahren·0 comments
Breaking Packages: finding the sharp edges of Python's package management systemdagster.io1 points·by pinkbeanz·vor 2 Jahren·0 comments
Scaling Dagster's DAG visualization to handle assetsdagster.io2 points·by pinkbeanz·vor 3 Jahren·0 comments
Ripgrep now twice as fast on Apple Silicon with new aarch64 SIMD implementationsgithub.com6 points·by pinkbeanz·vor 3 Jahren·1 comments
pinkbeanz·vor 2 Jahren·discussThis is neat -- how would you think about evaluating the quality loss as you change to more efficient models? I saw you did an analysis on the number of messages, but wondering if there's more robust methods?
pinkbeanz·vor 2 Jahren·discussthis is neat, where can i find more details on how the automatic routing logic works?