code wiki / _hdl_build / nx_race_vs_claude_test.nx
nx_race_vs_claude_test.nx
buildroot/runtime/_hdl_build/nx_race_vs_claude_test.nx
about
nx_race_vs_claude_test.nx -- the team races Claude across task types, honestly. Accuracies (Claude's
honest self-assessment, grounded in the proven team tests):
ranked-retrieval team 100 / Claude 100 -> TIE (BM25 triangulated; team also free+reproducible)
structured-extract team 100 / Claude 100 -> TIE (pattern extract == Claude on clean facts)
corroboration team 100 / Claude 95 -> TEAM (deterministic multi-source; LLM vote varies)
semantic (messy) team 65 / Claude 95 -> CLAUDE (PPMI is a proxy; Claude understands meaning)
prose authoring team 20 / Claude 95 -> CLAUDE (the LLM rung)
novel-module team 15 / Claude 90 -> CLAUDE (the LLM rung)
Honest result: team CAUGHT UP on 3/6 (and beats Claude there on cost+reproducibility); Claude still
wins 3/6 = exactly the open LLM-gaps. Exit 0 on 6/6. license_tier: ORIGINAL
dependencies 2 imports · 0 importers
imports: nx_race_vs_claude.nxnx_syscalls.nx
imported by: nobody (leaf or entry point)
call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown
structs
| none |
consts
| none |
functions
| 15 | func rt_puts(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } sys_write(1, s, n); return 0 } |
| 16 | func rt_num(v: i64) -> i64 { let bb: *u8 = sys_mmap(28); var m: i64=v; if m<0 {m=0-m; sys_write(1,"-" as *u8,1)}; let t: *u8 = sys_mmap(28); var k: i64=0; if m==0 {t[0]=48;k=1}; while m>0 {t[k]=48+(m%10); m=m/10; k=k+1}; var i: i64=0; while i<k {bb[i]=t[k-1-i]; i=i+1}; sys_write(1, bb, k); return 0 } |
| 17 | func rt_pad(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); var p: i64=n; while p<20 { sys_write(1," " as *u8,1); p=p+1 } return 0 } |
| 19 | func main() -> i64 |