nx_research_judgment_bench.nx
buildroot/runtime/nx_research_judgment_bench.nx
about
nx_research_judgment_bench.nx -- the first AGGREGATE honest measurement of the unified researcher's JUDGMENT.
On a LABELED case set, does nxr_assess's verdict correctly separate PUBLISH-worthy reports (grounded + cited +
numbers confirmed) from NOT-publish reports (fabricated number / uncited / dangling cite / MIS-CITED)? Reports
discrimination accuracy + the two error kinds, with the SAFETY-CRITICAL direction FALSE-PUBLISH (shipping a
defective report as PUBLISH) called out. It measures TWO policies side by side: CURRENT (nxr_assess's verdict)
and a PROPOSED STRICT policy (any unconfirmed numeric claim is a gap) computed from the SAME out[] slots -- so
the value of the fix is shown WITHOUT editing the concurrently-edited nx_research_unified (read-only import).
GREEN iff the PROPOSED policy has zero false-publish AND zero false-hold (fix is sound + strictly additive).
$0, deterministic, model-free. license_tier: ORIGINAL
dependencies 2 imports · 0 importers
imports: nx_syscalls.nxnx_research_unified.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
| 13 | func jb_puts(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 } |
| 14 | func jb_putn(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 as u8;k=1} while m>0{t[k]=(48+(m%10)) as u8;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 } |
| 15 | func jb_len(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} return n } called by 1: jb_case |
| 16 | func jb_verdict(v: i64) -> i64 |
| 24 | func jb_case(ans: *u8, srcv: *i64, nsrc: i64, expected_pub: i64, label: *u8, st: *i64) -> i64 |
| 49 | func main() -> i64 |