"Smooth Operators team on AMD AI League Hackathon"
Team Idea
ForgeML is a web-based platform designed to simplify the lifecycle of fine-tuning and evaluating large language models (LLMs). It provides a unified dashboard where users can register models and datasets, configure training jobs, monitor training progress, evaluate model performance, and manage trained model artifacts. The platform uses a FastAPI backend to orchestrate ML workflows, a Next.js frontend for interactive monitoring and management, and a Python-based training pipeline integrating Hugging Face Transformers and PEFT for LoRA fine-tuning. A persistent database stores model, dataset, training-job, artifact, and evaluation records. ForgeML aims to make LLM experimentation more accessible, organized, and reproducible by bringing model management, training orchestration, and evaluation into one place. The planned roadmap includes GPU-accelerated training, scalable job execution, and model inference and deployment capabilities.
Team Members
Shivansh Gupta
shivansh_gupta657
Submission
No submission yet