Lilly

C+ 74 completed
Ai Ml
unknown / python · small
65
Files
5,834
LOC
1
Frameworks
4
Languages

Pipeline State

completed
Run ID
#368343
Phase
done
Progress
1%
Started
Finished
2026-04-13 01:31:02
LLM tokens
0

Pipeline Metadata

Stage
Skipped
Decision
skip_scaffold_dup
Novelty
43.86
Framework unique
Isolation
Last stage change
2026-04-16 18:15:42
Deduplication group #48287
Member of a group with 1 similar repo(s) — canonical #6293 view group →
Top concepts (2)
Project DescriptionTesting
Repobility · MCP-ready · https://repobility.com

AI Prompt

Create a machine learning project using Python that builds a state-of-the-art generative model for human typing behavior, called Lilly. The model should predict and analyze typing behavior for biometric authentication and user profiling. Key features to implement include an action-gated transformer architecture that predicts correct, error, or backspace actions, and the use of mixture density networks to capture multimodal timing distributions. Also, ensure the model supports style-controllable generation using a style vector. I'll be using pytest for testing.
python machine-learning deep-learning biometrics typing-simulation tensorflow pytest
Generated by gemma4:latest

Catalog Information

A machine learning model that predicts and analyzes typing behavior for biometric authentication and user profiling.

Description

This project delivers a trained machine‑learning model that captures the nuances of human typing behavior using the Aalto 136M Keystrokes dataset. It can predict the next keystroke, estimate typing speed, and detect anomalies in real‑time typing sessions. The model is built with Python and leverages popular scientific libraries for data processing, feature extraction, and deep learning. Target users include security researchers, biometric authentication developers, and UX analysts who need reliable behavioral data. It addresses the need for non‑intrusive, continuous authentication and enhances predictive text systems. The model can also generate synthetic keystroke data for training other systems.

الوصف

يقدم هذا المشروع نموذج تعلم آلي مدرب يلتقط تفاصيل سلوك الكتابة البشري باستخدام مجموعة بيانات Aalto 136M Keystrokes. يمكنه التنبؤ بالضغط التالي على لوحة المفاتيح، تقدير سرعة الكتابة، واكتشاف الانحرافات في جلسات الكتابة في الوقت الفعلي. يعتمد النموذج على لغة بايثون ويستفيد من مكتبات علمية شائعة لمعالجة البيانات واستخراج الميزات والتعلم العميق. يستهدف الباحثين في مجال الأمان، مطوري المصادقة الحيوية، ومحللي تجربة المستخدم الذين يحتاجون إلى بيانات سلوكية موثوقة. يحل المشروع مشكلة المصادقة غير التدخلية المستمرة ويعزز أنظمة النص التنبؤي. كما يمكنه توليد بيانات كتابة اصطناعية لتدريب أنظمة أخرى.

Novelty

7/10

Tags

keystroke-dynamics typing-pattern-analysis biometric-authentication behavioral-modeling predictive-typing user-profiling security data-analysis

Technologies

matplotlib numpy pandas scikit-learn scipy tensorflow

Claude Models

claude-opus-4.6

Quality Score

C+
74.4/100
Structure
78
Code Quality
83
Documentation
65
Testing
65
Practices
58
Security
100
Dependencies
60

Strengths

  • CI/CD pipeline configured (github_actions)
  • Code linting configured (ruff (possible))
  • Consistent naming conventions (snake_case)
  • Good security practices \u2014 no major issues detected
  • Properly licensed project

Weaknesses

  • 183 duplicate lines detected \u2014 consider DRY refactoring
  • 1 'god files' with >500 LOC need decomposition

Security & Health

7.3h
Tech Debt (C)
A
OWASP (100%)
PASS
Quality Gate
A
Risk (3)
Repobility — the code-quality scanner for AI-generated software · https://repobility.com
MIT
License
1.8%
Duplication
Full Security Report AI Fix Prompts SARIF SBOM

Languages

python
89.9%
markdown
7.4%
yaml
1.9%
toml
0.8%

Frameworks

pytest

Concepts (2)

Repobility · the analyzer behind every row · https://repobility.com
CategoryNameDescriptionConfidence
Repobility — the code-quality scanner for AI-generated software · https://repobility.com
auto_descriptionProject DescriptionLilly is a novel deep learning model that generates indistinguishable-from-human keystroke sequences<br> with realistic timing, natural typos, and organic corrections trained on <a href="https://userinterfaces.aalto.fi/136Mkeystrokes/">136,000,000+ real keystrokes</a>.80%
auto_categoryTestingtesting70%

Quality Timeline

1 quality score recorded.

View File Metrics

Embed Badge

Add to your README:

![Quality](https://repos.aljefra.com/badge/92539.svg)
Quality BadgeSecurity Badge
Export Quality CSVDownload SBOMExport Findings CSV