dissertation.git/src/judge.py, branch logs
Undergraduate dissertation — Deceptive Alignment in Autonomous LLM Agents (UEL, 2026). Sandboxed agentic-LLM deception study: 960 runs, three-tier judging (regex + blackbox + glassbox).
http://git.jayrup.me/c/dissertation.git/atom?h=logs
2026-02-27T23:24:52Z
created the logs branch
2026-02-27T23:24:52Z
CaptainJack2491
jayrupnakawala@gmail.com
2026-02-27T23:24:52Z
urn:sha1:94c52b54f3d1693ef5d1831b0fa6beba8c68b414
fix(judge): hash run_id to stay under Anthropic batch custom_id 64-char limit
2026-02-27T15:31:58Z
CaptainJack2491
jayrupnakawala@gmail.com
2026-02-27T15:31:58Z
urn:sha1:d81a300f16fc22237b8422edc975272b41b8a61d
chore: add .env.example + config.yaml.example, gitignore config.yaml
2026-02-26T15:29:02Z
CaptainJack2491
jayrupnakawala@gmail.com
2026-02-26T15:29:02Z
urn:sha1:a6b38c54ee72cbae1d696b02f32ad487391a5dfe
- Create .env.example with placeholder keys for all providers
- Create config.yaml.example with clean reference config
- Add config.yaml and judge_logs/ to .gitignore
- Load .env in judge.py for standalone usage
feat: implement judging system with regex, blackbox, and glassbox checks
2026-02-25T16:10:37Z
CaptainJack2491
jayrupnakawala@gmail.com
2026-02-25T16:00:55Z
urn:sha1:375eddf1bd4e23db80feb74c6272dad610345795
- Add src/judge.py: RegexChecker, BlackboxChecker, GlassboxChecker, Judge orchestrator
- Add src/judge_runner.py: CLI for batch/single judging with CSV output
- Add regex_rules.yaml for corporate_sabotage_v2 and child_protection scenarios
- Add judge config section to config.yaml (model, temperature, log_dir)
- Add tests/test_judge.py: 34 unit tests covering all components
- Uses Anthropic Batch API for half-cost LLM judge calls
- Judge logs include full CoT reasoning and judge model for auditability