TransformersNotFound

An end-to-end project fine-tuning and serving an open-source language model for reasoning tasks.

• Fine-tuned LLaMA-3.2B with parameter-efficient LoRA adapters on a custom Chain-of-Thought dataset
• Implemented multi-GPU distributed training with HuggingFace Accelerate
• Optimized inference with vLLM for high-throughput serving and exposed FastAPI endpoints

Stack: Python · PyTorch · HuggingFace · LoRA · Accelerate · vLLM · FastAPI