I've come across a few cases now where I want to run an eval across multiple solvers or scorers. We effectively parallelize within those calls, but users would just map() across them if there are multiple. Maybe the solution is to future_map, though maybe we could drop a nice UI around this.
Out of scope for the first release, but maybe worth revisiting later.
I've come across a few cases now where I want to run an eval across multiple solvers or scorers. We effectively parallelize within those calls, but users would just
map()across them if there are multiple. Maybe the solution is to future_map, though maybe we could drop a nice UI around this.Out of scope for the first release, but maybe worth revisiting later.