Imakandi-Labs/yarngpt-finetune-trainer

YarnGPT Fine-tuning Trainer

Fine-tune YarnGPT2 on Nigerian speech data using LoRA for improved clarity.

Features

• —LoRA fine-tuning (efficient, ~8GB VRAM)
• —T4 GPU accelerated
• —Automatic push to HuggingFace Hub

Usage

1. 1.Click "Start Training"
2. 2.Wait for training to complete (~1-2 hours)
3. 3.Your fine-tuned model will be saved to your HuggingFace repo

Dataset

Uses Imakandi-Labs/aasd-nigerian-tts-processed - 7+ hours of Nigerian English speech.