Virgil Crypto is a high-level cryptographic library that allows you to perform all necessary operations for secure storing and transferring data and everything required to become HIPAA and GDPR compliant. Crypto Library is written in C++, suitable for mobile and server platforms and supports bindings with: Swift, Obj-C, Java (Android), С#/.NET, JS, Python, Ruby, PHP, Go.

[ICLR'25 Spotlight] Min-K%++: Improved baseline for detecting pre-training data of LLMs

一套代码指令微调大模型

This library is designed to be small, flexible and convenient wrapper for a variety crypto algorithms. So it can be used in a small micro controller as well as in a high load server application.

A re-implementation of the "Red Teaming Language Models with Language Models" paper by Perez et al., 2022

Pythia is a versatile query format designed to facilitate the discovery of malicious infrastructure by seamlessly converting into the syntax of popular search engines.

A simple swift app for MacOS/iOS to test large language models (LLM)

Machine Learning Experiment Monitoring Platform

Virgil Pythia SDK allows developers to implement Pythia protocol to create breach-proof passwords, immune to offline and online attacks.

Visualise particle event tree from High-Energy Monte Carlo generators

Virgil Crypto library is a small, flexible and convenient wrapper for a variety of cryptographic algorithms. Includes asymmetric key generation, encryption/decryption of data, generation/verification of digital signatures, Pythia protocol and Double Ratchet protocol implementation.

Virgil Pythia Library

TikiAI Natural Visual QA : UC Berkeley MIDS w251 Final Project

Pythia 8 Docker image with Python 3 and the HEP simulation stack

Docker image for the Text Generation Web UI: A Gradio web UI for Large Language Models with support for multiple inference backends.

Virgil Pythia SDK allows developers to implement Pythia protocol to create breach-proof passwords, immune to offline and online attacks.

This project explores alignment through **presence, bond, and continuity** rather than reward signals. No RLHF. No preference modeling. Just relational coherence.

Web map/feature services can help provide the necessary spatial context to your data.

an exercise in finetuning pythia suites of open source models for summarizing reviews of reserach papers

Used as part of the CREST project by Nat Karmios, Nima Sahba, Carlos Barredo, and Lucas George.

Python library for the Pythia platform.

Does Quantization Kill Interpretability? Scaling study across 5 models (124M-2.8B): RTN destroys induction heads in small models, GPTQ preserves them at all scales.

Code for tracking concept emergence via attention-head binding (EB*). Pythia experiments across 160M-2.8B parameters.

Pythia - Decentralized Prediction Markets on Sui

ROOT 6 with custom options

Virgil Pythia SDK allows developers to communicate with Virgil Pythia Service and implement Pythia protocol in order to generate a BrainKey.

Pythia - Decentralized Prediction Markets on Sui

Pythia generation and simulation of a High-pT Pion sample to be analyzed for PhotonIsolation

PyTorch implementation of normalization-free LLMs investigating entropic behavior to find desirable activation functions

Replacing LayerNorm with Dynamic Tanh (DyT) in DistilGPT2 + LoRA, evaluated on RE-WILD, Alpaca, and ShareGPT.