Open nniiicc opened 10 months ago
In the context of studying versions of a package we discussed the idea between versions what might change (besides code) are people and dependencies. Empirical software engineering has typically treated the stability of package dependencies as a proxy measurement for the hardening of a code base...We are interested in how this hardening happens in research software... In exploring this might think about, for example. a study that looks at Bits and Bytes config - peft
peft
for fine-tunning , and that included how to quantize model code...peft
and look at what it replaced over the last year (essentially just a diff on code state before and after the release of peft)
Keep open so that we can thread discussion about different ideas...