Playground
Pick a model and a training step, then give it a prompt. Slide back to step 0 to see where it started. LoRA vs full fine-tuning (our GPT-2 124M): curves and timeline
LoRA vs full fine-tuning (our GPT-2 124M)
These are base models: they continue text, they don't answer questions. Chat comes with Phase 2's fine-tuning.