Upload README.md with huggingface_hub

README.md

@@ -1,53 +1,53 @@
----
-license: "other"
-tags:
-- reinforcement-learning
-- evaluation
-- reverse-engineering
-- binaries
-- elf
-- pe
-- cross-architecture
-- taskset
-- harbor-framework
-- rl-environments
-size_categories:
-- 10K<n<100K
-pretty_name: binary-reasoning-v1
----
-
-# binary-reasoning-v1
-
-**Verified reverse-engineering tasks over real production binaries** — an RL / eval environment
-pack where every answer is derived from the artifact itself and machine-checked.
-
-- **10,242 items · 10 families · 7 ISAs** (x86-64, i686, AArch64, ARM32, RISC-V, MIPS o32/n64, PPC64LE)
-- Sources: ~1,200 real Ubuntu ELF files (system libs incl. libcrypto/zstd/sqlite3/libxml2, Mesa
-  DRI drivers, QEMU device models, systemd internals), Debian production cross-toolchain glibc,
-  Android/Termux production aarch64 libs, curated Windows PEs, labeled cross-compiles
-- Provenance per item: binary name, sha256_16, address, arch
-- **Verification:** independent REA audit (answers re-verified against the binaries); 10,163
-  deterministic items score 1.0 on gold / 0.0 on abstention; 0/10,242 answer leaks; all tasks load
-  against the real `verifiers` v1 framework; 20 agentic challenges ship Harbor-format task packages,
-  built + scored in real Docker containers with reward chain tested 20/20 (gold→1, wrong→0)
-- **Baseline:** a frontier local reasoning model scores 22% on our deterministic subset — headroom
-  is the feature
-
-## Contents of this repo (free sample)
-| file | what |
-|---|---|
-| `DATASHEET.md` | full composition table, guarantees, known limits, positioning vs existing work |
-| `sample_free.jsonl` | 9 items — one per family |
-| `EVAL_RESULTS.md` | baseline eval methodology + results |
-| `harbor_pilot_20.zip` | all 20 agentic Harbor task packages (task.toml + instruction + Dockerfile + verifier), ready for `harbor run` or a free HUD import ([direct download](https://huggingface.co/datasets/S3RL4/binary-reasoning-v1/resolve/main/harbor_pilot_20.zip)) |
-
-Also live on the Prime Intellect Environments Hub: `s3rl4/binary-reasoning-v1`
-(`prime env install s3rl4/binary-reasoning-v1`).
-
-The **full pack** (10,242 items + runnable verifiers v1 environment + Harbor packages + agentic
-binaries & sources) lives in the gated companion repo:
-[S3RL4/binary-reasoning-v1-full](https://huggingface.co/datasets/S3RL4/binary-reasoning-v1-full) —
-request access, approved same day. Commercial licensing and custom taskset builds: samueljanttila@gmail.com.
-
-## License
-All rights reserved pending commercial license; research evaluation use granted for the free sample.
+---
+license: "other"
+tags:
+- reinforcement-learning
+- evaluation
+- reverse-engineering
+- binaries
+- elf
+- pe
+- cross-architecture
+- taskset
+- harbor-framework
+- rl-environments
+size_categories:
+- 10K<n<100K
+pretty_name: binary-reasoning-v1
+---
+
+# binary-reasoning-v1
+
+**Verified reverse-engineering tasks over real production binaries** — an RL / eval environment
+pack where every answer is derived from the artifact itself and machine-checked.
+
+- **10,247 items · 10 families · 7 ISAs** (x86-64, i686, AArch64, ARM32, RISC-V, MIPS o32/n64, PPC64LE)
+- Sources: ~1,200 real Ubuntu ELF files (system libs incl. libcrypto/zstd/sqlite3/libxml2, Mesa
+  DRI drivers, QEMU device models, systemd internals), Debian production cross-toolchain glibc,
+  Android/Termux production aarch64 libs, curated Windows PEs, labeled cross-compiles
+- Provenance per item: binary name, sha256_16, address, arch
+- **Verification:** independent REA audit (answers re-verified against the binaries); 10,163
+  deterministic items score 1.0 on gold / 0.0 on abstention; 0/10,247 answer leaks; all tasks load
+  against the real `verifiers` v1 framework; 25 agentic challenges ship Harbor-format task packages,
+  built + scored in real Docker containers with reward chain tested 20/20 (gold→1, wrong→0)
+- **Baseline:** a frontier local reasoning model scores 22% on our deterministic subset — headroom
+  is the feature
+
+## Contents of this repo (free sample)
+| file | what |
+|---|---|
+| `DATASHEET.md` | full composition table, guarantees, known limits, positioning vs existing work |
+| `sample_free.jsonl` | 9 items — one per family |
+| `EVAL_RESULTS.md` | baseline eval methodology + results |
+| `harbor_pilot_25.zip` | all 25 agentic Harbor task packages (task.toml + instruction + Dockerfile + verifier), ready for `harbor run` or a free HUD import ([direct download](https://huggingface.co/datasets/S3RL4/binary-reasoning-v1/resolve/main/harbor_pilot_25.zip)) |
+
+Also live on the Prime Intellect Environments Hub: `s3rl4/binary-reasoning-v1`
+(`prime env install s3rl4/binary-reasoning-v1`).
+
+The **full pack** (10,247 items + runnable verifiers v1 environment + Harbor packages + agentic
+binaries & sources) lives in the gated companion repo:
+[S3RL4/binary-reasoning-v1-full](https://huggingface.co/datasets/S3RL4/binary-reasoning-v1-full) —
+request access, approved same day. Commercial licensing and custom taskset builds: samueljanttila@gmail.com.
+
+## License
+All rights reserved pending commercial license; research evaluation use granted for the free sample.