Model from: https://huggingface.co/WizardLM/WizardLM-13B-V1.2
Trained on: https://huggingface.co/datasets/gmongaras/reddit_political_2019
For about 18,000 steps with a batch size of 8, 2 accumulation steps, and using LoRA adapters on all layers.
Model from: https://huggingface.co/WizardLM/WizardLM-13B-V1.2
Trained on: https://huggingface.co/datasets/gmongaras/reddit_political_2019
For about 18,000 steps with a batch size of 8, 2 accumulation steps, and using LoRA adapters on all layers.