S3RL4 commited on
Commit
4a3b9c0
·
verified ·
1 Parent(s): 6712df3

Update DATASHEET.md: difficulty wording and caveats

Browse files
Files changed (1) hide show
  1. DATASHEET.md +3 -5
DATASHEET.md CHANGED
@@ -15,7 +15,7 @@
15
  | library_ident — SBOM-style: name+version of a production library from artifacts | easy/med | exact substring (name+version) | 14 |
16
  | arch_semantics_qa / isa_semantics_qa — instruction-sequence semantics | medium | rubric judge | 14 |
17
  | pe_import_reasoning — infer program behavior from imports+strings | medium | rubric judge | 4 |
18
- | pe_structure_qa — PE header/section facts | easy | exact field / rubric | 3 |
19
 
20
  Architectures: x86-64 (Linux ELF, Windows PE), i686, AArch64, ARM32, RISC-V, MIPS(o32/n64), PPC64LE.
21
  Sources: Ubuntu production binaries (649 distinct third-party file names referenced in shipped provenance;
@@ -38,7 +38,7 @@ expat 2.9.0 — platform-tagged in provenance)**, curated PE samples, cross-comp
38
  - Scrubbing: every symbol equal to the answer or a callee replaced with `<...>`; automated answer-leak scan through our in-house analysis pipeline: **0/10,251 hits**.
39
  - Ground truth derived from artifact itself (symbol tables, header fields, decompiler output).
40
  - Auto-grader included (`grading.py`): deterministic graders for exact/MC/crossarch/flag/library families (exact-match is alias-tolerant); rubric judge for open-ended. Test harness `test_grading.py`: 10,172 deterministic items scored 1.0 on gold answers, 0.0 on abstention; 79 routed to judge.
41
- - **Full agentic benchmark measured** (`EVAL_RESULTS-passk.md`, 2026-10-29): Claude Opus 5.5 solved **29/29 envs, 145/145 attempts** with tools (median 44 s) — every answer byte-exact vs our gold: oracle confirmation that the shipped binary accepts those answers and verifies our grading chain; not a proof that each answer is unique. The public tier is frontier-warm-up level; the paid product is the private hard tier (not yet per-task calibrated; calibration on your model is part of a pilot) + per-model pass-rate calibration service. Single-shot no-tools-requested ID reference stays 7/12 small-n, and a fixed 20-item agentic hard spot check found the sampled command-output/static-analysis families practice-level for Opus with tools (`EVAL_RESULTS-frontier.md` §C).
42
  - Agentic challenges: round 1 (8) + round 2 (6, stripped) + **round 3 (6: self-decrypting .text via .init_array, ptrace/TracerPid/timing anti-instrumentation, chained-seed stages, masked-blob hash + explicit_bzero, control-flow flattening with decoy states, encrypted-bytecode VM with permuted opcodes)** + **round 4 (5: two-stage ARX+FNV memory-only chain, MIPS-BE branch-delay-slot trap, Windows PE TLS-callback packer, AArch64 computed-`blr` chain with PAC noise, RISC-V PC-relative jump-table `jalr`)** + **round 5 (4: x86-64 static-pie relocation-table/GOT-indirect hunt, ARM32 Thumb/ARM interworking trap, PPC64LE condition-register juggling gate, PE PEB-walk API hiding with export-name hashing)** — the set covers multiple distinct RE techniques; every binary verified OK-on-flag / NO-on-mutation (24 single-char mutations each, zero false OKs); exit status = first failing position enables black-box probing; `strings` shows no flag fragments.
43
  - Provenance per item: binary name and sha256_16 on every item; function address/size and architecture where applicable.
44
 
@@ -61,9 +61,7 @@ https://huggingface.co/datasets/S3RL4/binary-reasoning-v1/resolve/main/harbor_pi
61
  | AssemblyBench (arXiv:2604.03750 / HF) | **robotics** mechanical assembly — name collision only | unrelated |
62
  | atul10 RE code datasets, ollvm-decompilation (HF dl<400) | source-level or decompiler-output pairs | no executable oracles, no verification bar, single arch |
63
 
64
- Conclusion: no shipped, verified, production-binary RE environment pack of this scale exists on
65
- the Hub (365k tasks = SWE/terminal/search only; figure from 2026-10-28 survey — re-verify before quoting as fresh)
66
- or on HF. First-mover claim holds with evidence (re-checked live 2026-10-29: "flag hunt" n=0, category neighbors ≤400 dl).
67
  Extension capability demonstrated across two rounds: 9 extension envs built+container-verified spanning x86-64
68
  (chain+FNV; static-pie relocation-table hunt), MIPS BE (delay slots), AArch64 (PAC + computed blr), RISC-V
69
  (PC-relative jump table), ARM32 (Thumb/ARM interworking), PPC64LE (condition-register gate) and Windows PE
 
15
  | library_ident — SBOM-style: name+version of a production library from artifacts | easy/med | exact substring (name+version) | 14 |
16
  | arch_semantics_qa / isa_semantics_qa — instruction-sequence semantics | medium | rubric judge | 14 |
17
  | pe_import_reasoning — infer program behavior from imports+strings | medium | rubric judge | 4 |
18
+ | pe_structure_qa — PE header/section facts | easy | rubric judge | 3 |
19
 
20
  Architectures: x86-64 (Linux ELF, Windows PE), i686, AArch64, ARM32, RISC-V, MIPS(o32/n64), PPC64LE.
21
  Sources: Ubuntu production binaries (649 distinct third-party file names referenced in shipped provenance;
 
38
  - Scrubbing: every symbol equal to the answer or a callee replaced with `<...>`; automated answer-leak scan through our in-house analysis pipeline: **0/10,251 hits**.
39
  - Ground truth derived from artifact itself (symbol tables, header fields, decompiler output).
40
  - Auto-grader included (`grading.py`): deterministic graders for exact/MC/crossarch/flag/library families (exact-match is alias-tolerant); rubric judge for open-ended. Test harness `test_grading.py`: 10,172 deterministic items scored 1.0 on gold answers, 0.0 on abstention; 79 routed to judge.
41
+ - **Full agentic benchmark measured** (`EVAL_RESULTS-passk.md`, 2026-10-29): Claude Opus 5.5 solved **29/29 envs, 145/145 attempts** with tools (median 44 s) — every answer byte-exact vs our gold: oracle confirmation that the shipped binary accepts those answers and verifies our grading chain; not a proof that each answer is unique. The public tier is frontier-warm-up level; the paid product is the private hard tier (not yet per-task calibrated; calibration on your model is part of a pilot) + per-model pass-rate calibration service. Single-shot no-tools-requested ID reference stays 7/12 small-n, and a fixed 20-item agentic hard spot check found the sampled command-output/static-analysis families practice-level for Opus with tools (40/40 attempts; non-isolated host run, upper bound; `EVAL_RESULTS-frontier.md` §C).
42
  - Agentic challenges: round 1 (8) + round 2 (6, stripped) + **round 3 (6: self-decrypting .text via .init_array, ptrace/TracerPid/timing anti-instrumentation, chained-seed stages, masked-blob hash + explicit_bzero, control-flow flattening with decoy states, encrypted-bytecode VM with permuted opcodes)** + **round 4 (5: two-stage ARX+FNV memory-only chain, MIPS-BE branch-delay-slot trap, Windows PE TLS-callback packer, AArch64 computed-`blr` chain with PAC noise, RISC-V PC-relative jump-table `jalr`)** + **round 5 (4: x86-64 static-pie relocation-table/GOT-indirect hunt, ARM32 Thumb/ARM interworking trap, PPC64LE condition-register juggling gate, PE PEB-walk API hiding with export-name hashing)** — the set covers multiple distinct RE techniques; every binary verified OK-on-flag / NO-on-mutation (24 single-char mutations each, zero false OKs); exit status = first failing position enables black-box probing; `strings` shows no flag fragments.
43
  - Provenance per item: binary name and sha256_16 on every item; function address/size and architecture where applicable.
44
 
 
61
  | AssemblyBench (arXiv:2604.03750 / HF) | **robotics** mechanical assembly — name collision only | unrelated |
62
  | atul10 RE code datasets, ollvm-decompilation (HF dl<400) | source-level or decompiler-output pairs | no executable oracles, no verification bar, single arch |
63
 
64
+ Conclusion: our niche is verified, scrubbed, oracle-graded production-binary RE plus the agentic env set, not "no such data exists anywhere." Academic function-name-prediction datasets exist at scale (e.g. AGNOMIN arXiv:2509.25514: 396k functions / 3 architectures); our differentiator is verification, provenance, deterministic grading, and the agentic environments.
 
 
65
  Extension capability demonstrated across two rounds: 9 extension envs built+container-verified spanning x86-64
66
  (chain+FNV; static-pie relocation-table hunt), MIPS BE (delay slots), AArch64 (PAC + computed blr), RISC-V
67
  (PC-relative jump table), ARM32 (Thumb/ARM interworking), PPC64LE (condition-register gate) and Windows PE