Finetuning Starcoder2-3B for Code Completion on a single A100 GPU
Fine-tune StarCoder2-3b for SQL tasks on limited resources with LORA. LORA reduces model size for faster training on smaller datasets. StarCoder2 is a family of code generation models (3B, 7B, and 15B), trained on 600+ programming languages from The Stack v2 and some natural language text such as Wikipedia, Arxiv, and GitHub issues.
🚀 AI development platform with MEP architecture - stop writing prompts, start building with 90% less typing
Replication package for the paper: "How Much Do Code Language Models Remember? An Investigation on Data Extraction Attacks before and after Fine-tuning"
A sketch-guided multi-objective alignment framework for improving the correctness, efficiency, and algorithmic reasoning of code language models.
This work was done as part of SRIP 2025 Internship, IIT Gandhinagar
Three-stage TypeScript instruction data pipeline using StarCoder2 and SelfOSSInstruct
Desktop SAST with local AI. Heuristic patterns for SQLi, XSS, weak crypto and hardcoded secrets, plus StarCoder2-3B for context-aware code review. Optional Trivy. GTK4, fully offline.
To associate your repository with the starcoder2 topic, visit your repo's landing page and select "manage topics."