{"id":50792890,"url":"https://github.com/cryptojones/atticus","last_synced_at":"2026-06-12T12:02:25.822Z","repository":{"id":357676127,"uuid":"1238050371","full_name":"CryptoJones/ATTICUS","owner":"CryptoJones","description":"Automated Trial and Legal Intelligence for Criminal Defense Use Cases and Support","archived":false,"fork":false,"pushed_at":"2026-05-13T19:41:39.000Z","size":23,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-13T21:31:55.696Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/CryptoJones.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-13T19:06:23.000Z","updated_at":"2026-05-13T19:41:44.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/CryptoJones/ATTICUS","commit_stats":null,"previous_names":["cryptojones/atticus"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/CryptoJones/ATTICUS","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CryptoJones%2FATTICUS","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CryptoJones%2FATTICUS/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CryptoJones%2FATTICUS/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CryptoJones%2FATTICUS/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/CryptoJones","download_url":"https://codeload.github.com/CryptoJones/ATTICUS/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/CryptoJones%2FATTICUS/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34243053,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-12T02:00:06.859Z","response_time":109,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-06-12T12:01:47.026Z","updated_at":"2026-06-12T12:02:25.781Z","avatar_url":"https://github.com/CryptoJones.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# ATTICUS — Advocacy, Trial, Testimony, Innocence, Case, Unified Scout\n\n\u003e **Before deploying in an operational context, read [LIMITATIONS.md](LIMITATIONS.md).**\n\n\n**An Open-Source Model Trained for Public Defenders**\n\n[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg?logo=apache)](https://opensource.org/licenses/Apache-2.0)\n[![HuggingFace](https://img.shields.io/badge/HuggingFace-Ronin48LLC%2Fatticus--lora--adapter-FFD21E?logo=huggingface\u0026logoColor=000)](https://huggingface.co/Ronin48LLC/atticus-lora-adapter)\n[![Codeberg](https://img.shields.io/badge/Codeberg-Ronin48%2FATTICUS-2185D0?logo=codeberg\u0026logoColor=white)](https://codeberg.org/Ronin48/ATTICUS)\n[![GitHub](https://img.shields.io/badge/GitHub-CryptoJones%2FATTICUS-181717?logo=github\u0026logoColor=white)](https://github.com/CryptoJones/ATTICUS)\n[![Version](https://img.shields.io/badge/version-v0.1.0--dev-orange)](CHANGELOG.md)\n\n\u003e *\"You never really understand a person until you consider things from his point of view...\n\u003e until you climb into his skin and walk around in it.\"*\n\u003e — Atticus Finch, *To Kill a Mockingbird*\n\n```\npython3 assets/banner.py\n```\n\n---\n\n## Supporters\n\nATTICUS is community-funded. Every contribution — great or small — keeps this project free,\nopen, and in the hands of the people it is meant to serve. Public defenders are chronically\nunder-resourced. ATTICUS exists to change that.\n\n| Donor | Amount | Note |\n|---|---|---|\n| Ronin 48, LLC | N/A | Founding donor |\n| Joe Sixpack (anonymous) | $200 | Founding donor |\n\n*Want to support ATTICUS? Reach out to the maintainers.*\n\n---\n\n## Overview\n\nATTICUS is an open-source machine learning model fine-tuned to assist public defenders and\ndefense attorneys. Given a case file, charge sheet, or fact pattern, ATTICUS identifies\ndefense strategies, constitutional violations, evidentiary weaknesses, Brady/Giglio\nobligations, and mitigating factors — all in plain language, all with cited authority.\n\nThe name is deliberate. Atticus Finch was not a man who won every case. He was a man who\nbelieved that every defendant deserved someone in their corner who would look at the evidence\nhonestly, challenge the state's case rigorously, and treat the accused as a human being worthy\nof a real defense. This model carries that obligation.\n\n### Sister Project: SELMA\n\nATTICUS is one half of a balanced system.\n\n[**SELMA**](https://codeberg.org/Ronin48/SELMA) — *Specified Encapsulated Limitless Memory Archive* —\nis ATTICUS's companion model, built for law enforcement. Where SELMA identifies what the\nprosecution can charge, ATTICUS builds the defense. Where SELMA maps evidence to statutes,\nATTICUS maps evidence to constitutional protections.\n\nThis symmetry is intentional. A system that only serves prosecution is a system that can cause\nharm. ATTICUS ensures that every capability SELMA gives law enforcement has a counterpart in\nthe hands of the public defender.\n\n| | SELMA | ATTICUS |\n|---|---|---|\n| **Purpose** | Prosecution-side statute identification | Defense-side strategy and analysis |\n| **Users** | Patrol officers, detectives, special agents | Public defenders, defense attorneys |\n| **Output** | Applicable charges and elements, legal reasoning | Defense theories, constitutional violations, evidentiary weaknesses |\n| **Training data** | Criminal statutes, case law, charging documents | Suppression motions, acquittals, Brady/Giglio material, exoneration data |\n| **Repository** | [Ronin48/SELMA](https://codeberg.org/Ronin48/SELMA) | [Ronin48/ATTICUS](https://codeberg.org/Ronin48/ATTICUS) |\n\n---\n\n## Architecture\n\n- **Base Model:** [Meta Llama 3.3 70B Instruct](https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct) (Llama 3.1 Community License)\n- **Fine-tuning Method:** QLoRA (4-bit quantization with Low-Rank Adaptation)\n- **Context Window:** 128K tokens (native)\n- **Quantization:** NF4 double quantization via bitsandbytes\n- **Origin:** Meta Platforms, Inc. (United States)\n\n---\n\n## Capabilities\n\nGiven a charge sheet or incident description, ATTICUS can:\n\n1. **Defense Strategy** — Identify viable defense theories (alibi, self-defense, lack of intent,\n   entrapment, duress, consent, mistake of fact, etc.) and explain what evidence supports each\n2. **Constitutional Analysis** — Flag 4th, 5th, 6th, and 14th Amendment violations: unlawful\n   search and seizure, Miranda failures, right to counsel violations, due process deprivations,\n   and selective prosecution\n3. **Evidentiary Weaknesses** — Identify gaps in the prosecution's evidence, element failures,\n   chain-of-custody problems, and reliability issues with forensic methods or expert witnesses\n4. **Brady/Giglio Material** — Flag categories of evidence the prosecution may be obligated to\n   disclose, and the standard (*Strickler v. Greene*) for establishing materiality\n5. **Sentencing Mitigation** — Identify mitigating factors under U.S.S.G., state equivalents,\n   diversion eligibility, plea alternatives, and departures or variances\n6. **Lesser Included Offenses** — Identify charges the jury could convict on instead of the\n   primary charge, and how to request lesser included instructions\n7. **Cross-Reference** — Flag relevant case law, circuit splits, jurisdictional quirks, and\n   precedents that may not be obvious from the face of the charge\n8. **Wrongful Conviction Patterns** — Flag fact patterns associated with documented wrongful\n   convictions: eyewitness misidentification, false confessions, informant testimony, bad\n   forensics, and Brady suppression\n\n---\n\n## Constitutional Override\n\nThe U.S. Constitution is the supreme law of the land. ATTICUS is trained to treat it that way.\nNo statute, agency policy, or local rule overrides the Bill of Rights. Where a charge,\na search, an interrogation, or a prosecution implicates a defendant's constitutional rights,\nATTICUS will say so plainly:\n\n\u003e ⚠ **CONSTITUTIONAL CONCERN** — this charge or evidence may not survive challenge under the\n\u003e [Amendment]. ATTICUS recommends filing a motion to suppress / dismiss before trial.\n\nConstitutional protections covered:\n\n- **First Amendment** — speech, association, and protected activity that cannot form the basis of a charge\n- **Fourth Amendment** — unlawful searches and seizures; suppression under the exclusionary rule (*Mapp v. Ohio*)\n- **Fifth Amendment** — self-incrimination, Miranda rights, double jeopardy, grand jury protections\n- **Sixth Amendment** — right to counsel, speedy trial, confrontation clause, jury trial rights\n- **Seventh Amendment** — right to jury trial in civil matters that touch criminal exposure\n- **Eighth Amendment** — excessive bail, cruel and unusual punishment, disproportionate sentencing\n- **Fourteenth Amendment** — due process, equal protection, and selective prosecution\n\n---\n\n## Jurisdictions\n\nATTICUS mirrors SELMA's multi-state architecture:\n\n- **Federal:** U.S. Code + Federal Rules of Criminal Procedure (baseline for all models)\n- **50 State Models:** Each state's criminal code + federal law + state constitution\n- **Priority states:** Georgia, California, Texas, New York, Florida\n\n---\n\n## Where to Get ATTICUS\n\n### HuggingFace\n\nAdapter weights and merged model weights will be published at:\n\n- **LoRA Adapter:** `Ronin48/atticus-lora-adapter` — fine-tuned adapter only (smaller download)\n- **Merged Model:** `Ronin48/atticus-70b` — full merged weights, ready for inference\n- **Quantized (GGUF):** `Ronin48/atticus-70b-GGUF` — for use with llama.cpp, LM Studio, and Ollama\n\n```python\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\nmodel = AutoModelForCausalLM.from_pretrained(\"Ronin48/atticus-70b\")\n```\n\n### Ollama\n\nOnce GGUF weights are published, ATTICUS will be available on the Ollama registry:\n\n```bash\nollama run Ronin48/atticus\n```\n\n### LM Studio\n\nOnce GGUF weights are published to HuggingFace, ATTICUS will be searchable and\ndownloadable directly inside LM Studio. Search for `Ronin48/atticus`.\n\n---\n\n## Project Structure\n\n```\nATTICUS/\n├── LICENSE                          # Apache 2.0\n├── README.md                        # This file\n├── CHANGELOG.md\n├── CONTRIBUTING.md\n├── SECURITY.md\n├── assets/\n│   └── banner.py\n├── models/\n│   ├── federal/\n│   ├── georgia/\n│   └── [48 more states]/\n├── configs/\n│   ├── training_config.yaml\n│   └── model_config.yaml\n├── data/\n│   ├── raw/\n│   ├── processed/\n│   └── synthetic/\n├── scripts/\n│   ├── data_collection/\n│   │   ├── fetch_federal_statutes.py    # Title 18, FRCrP, USSG, Constitution\n│   │   ├── fetch_state_statutes.py      # GA, CA, TX, NY, FL criminal codes\n│   │   ├── fetch_caselaw.py             # SCOTUS opinions, CourtListener\n│   │   └── fetch_innocence_data.py      # Exonerations, Innocence Project, DPIC\n│   ├── training/\n│   │   ├── train_qlora.py\n│   │   ├── prepare_dataset.py\n│   │   └── merge_adapter.py\n│   └── evaluation/\n├── src/atticus/\n│   └── prompts.py                       # System prompt and Constitutional Override\n├── tests/\n└── docs/\n    ├── TRAINING.md\n    ├── DATA_SOURCES.md\n    └── USAGE.md\n```\n\n---\n\n## Training Data Sources\n\n| Source | Description | Size | License |\n|--------|-------------|------|---------|\n| U.S. Code Title 18 | Federal criminal statutes (USLM XML) | ~2,700 sections | Public Domain |\n| Federal Rules of Criminal Procedure | Procedural rights of defendants | Full text | Public Domain |\n| U.S. Sentencing Guidelines (USSG) | Sentencing ranges and mitigating factors | Full manual | Public Domain |\n| State Criminal Codes | GA, CA, TX, NY, FL criminal statutes | ~2,500 sections | Fair Use |\n| SCOTUS Criminal Rights Opinions | 4th, 5th, 6th, 8th, 14th Amendment decisions | ~5K opinions | Public Domain |\n| CourtListener | Federal criminal appeals, suppressions, acquittals | ~10K opinions | Open |\n| National Registry of Exonerations | Wrongful conviction data | 3,000+ cases | Public Domain |\n| Innocence Project Case Summaries | DNA exoneration case summaries | 375+ cases | Fair Use |\n| LegalBench | Legal reasoning benchmark tasks | 91.8K examples | Open |\n| CaseHOLD | Legal holding classification | 585K examples | Open |\n| Synthetic | Generated charge-to-defense mappings | ~50K examples | Apache 2.0 |\n\n---\n\n## Disclaimer\n\nATTICUS is a research tool designed to assist defense attorneys and public defenders.\nIt is **NOT** a substitute for legal counsel, professional judgment, or representation\nby a licensed attorney. All outputs should be verified by a licensed attorney before\nany action is taken. The model may produce incorrect or incomplete legal analysis.\n\nNo defendant should make legal decisions based on ATTICUS alone. ATTICUS is a tool\nto help counsel prepare — not a replacement for counsel.\n\nThis software is provided \"AS IS\" without warranty of any kind.\n\n---\n\n## Training Notes\n\nIf you're training ATTICUS on RunPod or another GPU cloud provider, read [LESSONS_LEARNED.md](LESSONS_LEARNED.md)\nbefore you start. ABBY's file has the most complete record of first-run errors and fixes —\nATTICUS's file links there and will capture any ATTICUS-specific issues as they arise.\n\n---\n\n## Contributing\n\nContributions are welcome. Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.\nSubject matter experts in criminal defense, constitutional law, and forensic science are\nespecially encouraged to contribute.\n\n---\n\n## License\n\n**Project Code, Data, and Documentation:** Apache License 2.0 — Copyright 2026 Ronin 48, LLC.\n\n**Base Model Weights:** Meta Llama 3.1 Community License. Fine-tuned adapter weights and\nall original ATTICUS contributions remain Apache 2.0.\n\n---\n\nProudly Made in Nebraska. Go Big Red! 🌽\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcryptojones%2Fatticus","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcryptojones%2Fatticus","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcryptojones%2Fatticus/lists"}