CodeGen2 vs GPT-4

Salesforce's GitHub repository "CodeGen2" presents their latest program synthesis models, detailed in their official research release. Geared towards helping developers with code generation, CodeGen2 offers models of various capacities, starting from 1B up to 16B parameters. Hosted on GitHub, users are encouraged to create an account to contribute to the ongoing development of these models. Additionally, the repository is integrated with Hugging Face, a widely recognized hub for open-source model sharing and collaboration, providing accessible checkpoints for each of the models. The models, specifically featured at the International Conference on Learning Representations (ICLR) 2023, are designed for both causal and infill sampling, offering a glimpse into the future of programming and the functionalities of large language models (LLMs) in the natural language processing domain.

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.