Squad Models
Collection
Models trained on squad data • 3 items • Updated
Model from: https://huggingface.co/TheBloke/wizardLM-7B-HF/tree/main
Trained on: https://huggingface.co/datasets/squad
For about 4500 steps (1 epoch) with a batch size of 8, 2 accumulation steps, and using LoRA adapters on all layers.