Databricks' dolly-v2-12b vs GPT-4

Databricks introduces dolly-v2-12b, an inventive language model providing high-quality, instruction-following capabilities. This 12-billion-parameter model is built on Pythia-12b, yielding exceptional performance, not inherent in its foundational model. Dolly-v2-12b is optimized for commercial use, having been fine-tuned on a diverse mix of instructions covering areas like brainstorming, classification, and summarization. While it may not be at the forefront of AI models, its proficiency in instruction adherence is remarkable. Available in varying sizes, dolly-v2-12b is complemented by the smaller dolly-v2-7b and dolly-v2-3b models. Users can leverage this model easily with Transformers and PyTorch using the Hugging Face platform, tapping into the practical insights and extensive dataset, databricks-dolly-15k, to power their AI applications.

GPT-4 is the latest milestone in OpenAI’s effort in scaling up deep learning.

GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks. For example, it passes a simulated bar exam with a score around the top 10% of test takers; in contrast, GPT-3.5’s score was around the bottom 10%. We’ve spent 6 months iteratively aligning GPT-4 using lessons from our adversarial testing program as well as ChatGPT, resulting in our best-ever results (though far from perfect) on factuality, steerability, and refusing to go outside of guardrails.

GPT-4 is more creative and collaborative than ever before. It can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style.