BendMlLlama.group_att: def of bend-ml-llama@0.2.1.0
Import: import bend-ml-llama@0.2.1.0/main.bend as BendMlLlama
Lines 788-796 of bend-ml-llama@0.2.1.0/main.bend: https://jend.leeeet.dev/read.txt?path=bend-ml-llama@0.2.1.0/main.bend&start_line=788&end_line=796

import Base

# the query heads of one group over its key/value head, in order; the arrays are threaded through
def group_att(hs: List<&2, Head>, +hd: Nat, +t: Nat, +scale: F32, k: Array<F32>, v: Array<F32>, +acc: List<&2, F32>) -> HO:
  match hs:
    case Nil{}:
      HO{k, v, acc}
    case Con{h, rest}:
      HO.then(HO, head_att(hd, t, scale, k, v, join(h)), k2 => v2 => +o =>
        # bend-lint: disable-next bend-lint-rules/performance/append_in_loop -- the loop runs over one group of query heads (at most 15 of 64 numbers) not-expected-because: measured 2026-10-17 with a 1,190-token prompt: no change (NOTES)
        group_att(rest, hd, t, scale, k2, v2, List.append(&2, F32, acc, o)))

Uses (10):
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Array
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=F32
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=List
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=List.append
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Nat
  https://jend.leeeet.dev/definition.txt?path=bend-ml-llama@0.2.1.0/main.bend&name=HO
  https://jend.leeeet.dev/definition.txt?path=bend-ml-llama@0.2.1.0/main.bend&name=HO.then
  https://jend.leeeet.dev/definition.txt?path=bend-ml-llama@0.2.1.0/main.bend&name=Head
  https://jend.leeeet.dev/definition.txt?path=bend-ml-llama@0.2.1.0/main.bend&name=head_att
  https://jend.leeeet.dev/definition.txt?path=bend-ml-llama@0.2.1.0/main.bend&name=join