model-alignment

Educational analysis of LLM alignment, safety behavior, and framing-sensitive response patterns.

Targeted Data Generation with Large Language Models

Test and evaluate Large Language Models against prompt injections, jailbreaks, and adversarial attacks with a web-based interactive lab.

A comprehensive blueprint for Large Language Models(LLMs). A rigorous curriculum covering transformer architectures, pre-training dynamics, parameter-efficient fine-tuning (PEFT), RAG orchestration, model quantization, evaluation, and alignment.

SimOutUtils - Utilities for analyzing time series simulation output

Value-Spectrum: Quantifying Preferences of Vision-Language Models via Value Decomposition in Social Media Contexts

Constrained adaptation of language models on a manifold of valid adapters, studied as a safety mechanism: poisoning resistance, out-of-distribution detection, and few-shot regularization, by construction rather than by heuristics.

French course summary on LLM post-training, covering SFT, DPO, reinforcement learning, preference optimization, and model alignment.

Top Model Alignment Optimization Software (Opensource) Star if you like it!

Aligning an instruction-tuned Llama 3.2 model to a professional brand voice using Direct Preference Optimization (DPO) with TRL — no separate reward model required.

This repository introduces the AraIslaMorals framework, utilized to investigate the alignment of Large Language Models (LLMs) with Arabic-Islamic moral values through three complementary methodologies.