A multimodal AI agent is characterized by its ability to process and learn from various types of data, including natural language, visual, and audio inputs, to inform its actions. Despite advancements in large language models that incorporate visual data, such as GPT-4V, effectively translating image-based data into actionable outcomes for AI agents continues to be challenging. In this paper, we introduce a multimodal model that incorporates the concept of functional token specifically designed for AI agent applications. To ensure compatibility with edge devices, our model is optimized to a compact size of less than 1B parameters. Like GPT-4, our model can process both English and Chinese. We demonstrate that this model is capable of operating efficiently on a wide range of edge devices, including as constrained as a Raspberry Pi.

The current state-of-the-art (SOTA) for automated text-to-SQL still falls well short of expert human performance as measured by execution accuracy (EX) on the BIRD-SQL benchmark. The most accurate methods are also slow and expensive. To advance the SOTA for text-to-SQL while reducing cost and improving speed, we explore the combination of low-cost fine tuning, novel methods for diverse retrieval-augmented generation (RAG) and new input and output formats that help large language models (LLMs) achieve higher EX. We introduce two new methods, Dubo-SQL v1 and v2. Dubo-SQL v1 sets a new record for EX on the holdout test set of BIRD-SQL. Dubo-SQL v2 achieves even higher performance on the BIRD-SQL dev set. Dubo-SQL v1 relies on LLMs from OpenAI, but uses the low-cost GPT-3.5 Turbo while exceeding the performance of the next-best model using OpenAI, which instead uses the more expensive GPT-4. Dubo-SQL v1 exceeds the performance of the next-best model using GPT-3.5 by over 20%. Dubo-SQL v2 uses GPT-4 Turbo and RAG in place of fine tuning to push EX higher.

Reinforcement Learning From Human Feedback (RLHF) has been a critical to the success of the latest generation of generative AI models. In response to the complex nature of the classical RLHF pipeline, direct alignment algorithms such as Direct Preference Optimization (DPO) have emerged as an alternative approach. Although DPO solves the same objective as the standard RLHF setup, there is a mismatch between the two approaches. Standard RLHF deploys reinforcement learning in a specific token-level MDP, while DPO is derived as a bandit problem in which the whole response of the model is treated as a single arm. In this work we rectify this difference, first we theoretically show that we can derive DPO in the token-level MDP as a general inverse Q-learning algorithm, which satisfies the Bellman equation. Using our theoretical results, we provide three concrete empirical insights. First, we show that because of its token level interpretation, DPO is able to perform some type of credit assignment. Next, we prove that under the token level formulation, classical search-based algorithms, such as MCTS, which have recently been applied to the language generation space, are equivalent to likelihood-based search on a DPO policy. Empirically we show that a simple beam search yields meaningful improvement over the base DPO policy. Finally, we show how the choice of reference policy causes implicit rewards to decline during training. We conclude by discussing applications of our work, including information elicitation in multi-tun dialogue, reasoning, agentic applications and end-to-end training of multi-model systems.

We introduce a novel software abstraction termed "model caller," acting as an intermediary for AI and ML model calling, advocating its transformative utility beyond existing model-serving frameworks. This abstraction offers multiple advantages: enhanced accuracy and reduced latency in model predictions, superior monitoring and observability of models, more streamlined AI system architectures, simplified AI development and management processes, and improved collaboration and accountability across AI/ML/Data Science, software, data, and operations teams. Model callers are valuable for both creators and users of models within both predictive and generative AI applications. Additionally, we have developed and released a prototype Python library for model callers, accessible for installation via pip or for download from GitHub.

Incorporating Generative AI (GenAI) and Large Language Models (LLMs) in education can enhance teaching efficiency and enrich student learning. Current LLM usage involves conversational user interfaces (CUIs) for tasks like generating materials or providing feedback. However, this presents challenges including the need for educator expertise in AI and CUIs, ethical concerns with high-stakes decisions, and privacy risks. CUIs also struggle with complex tasks. To address these, we propose transitioning from CUIs to user-friendly applications leveraging LLMs via API calls. We present a framework for ethically incorporating GenAI into educational tools and demonstrate its application in our tool, Feedback Copilot, which provides personalized feedback on student assignments. Our evaluation shows the effectiveness of this approach, with implications for GenAI researchers, educators, and technologists. This work charts a course for the future of GenAI in education.

Note: This is not an official technical report from OpenAI. Sora is a text-to-video generative AI model, released by OpenAI in February 2024. The model is trained to generate videos of realistic or imaginative scenes from text instructions and show potential in simulating the physical world. Based on public technical reports and reverse engineering, this paper presents a comprehensive review of the model's background, related technologies, applications, remaining challenges, and future directions of text-to-video AI models. We first trace Sora's development and investigate the underlying technologies used to build this "world simulator". Then, we describe in detail the applications and potential impact of Sora in multiple industries ranging from film-making and education to marketing. We discuss the main challenges and limitations that need to be addressed to widely deploy Sora, such as ensuring safe and unbiased video generation. Lastly, we discuss the future development of Sora and video generation models in general, and how advancements in the field could enable new ways of human-AI interaction, boosting productivity and creativity of video generation.

The rapid evolution of text-to-image diffusion models has opened the door of generative AI, enabling the translation of textual descriptions into visually compelling images with remarkable quality. However, a persistent challenge within this domain is the optimization of prompts to effectively convey abstract concepts into concrete objects. For example, text encoders can hardly express "peace", while can easily illustrate olive branches and white doves. This paper introduces a novel approach named Prompt Optimizer for Abstract Concepts (POAC) specifically designed to enhance the performance of text-to-image diffusion models in interpreting and generating images from abstract concepts. We propose a Prompt Language Model (PLM), which is initialized from a pre-trained language model, and then fine-tuned with a curated dataset of abstract concept prompts. The dataset is created with GPT-4 to extend the abstract concept to a scene and concrete objects. Our framework employs a Reinforcement Learning (RL)-based optimization strategy, focusing on the alignment between the generated images by a stable diffusion model and optimized prompts. Through extensive experiments, we demonstrate that our proposed POAC significantly improves the accuracy and aesthetic quality of generated images, particularly in the description of abstract concepts and alignment with optimized prompts. We also present a comprehensive analysis of our model's performance across diffusion models under different settings, showcasing its versatility and effectiveness in enhancing abstract concept representation.

In the realm of aerospace design, achieving smooth curves is paramount, particularly when crafting objects such as airfoils. Generative Adversarial Network (GAN), a widely employed generative AI technique, has proven instrumental in synthesizing airfoil designs. However, a common limitation of GAN is the inherent lack of smoothness in the generated airfoil surfaces. To address this issue, we present a GAN model featuring a customized loss function built to produce seamlessly contoured airfoil designs. Additionally, our model demonstrates a substantial increase in design diversity compared to a conventional GAN augmented with a post-processing smoothing filter.

Amazon Music is joining Spotify in starting to offer a generative AI-powered playlist feature. For now, Maestro is available in beta to a small number of Amazon Music users in the US on iOS and Android. Folks who are included in the beta will see Maestro on the home screen after they update to the latest version of the app. They can also access the tool by tapping the plus button to create a new playlist. The idea is to use natural language prompts to create any kind of playlist imaginable.

OpenAI named the former president of Amazon Web Services's Japan arm to spearhead its push to woo enterprise clients in the world's fourth-largest economy. The artificial intelligence darling is opening an office in Tokyo as it releases a custom GPT-4 model catering to Japanese language users. OpenAI said it has 2 million weekly active users in the country, while its enterprise clients include Daikin Industries, Rakuten Group and an affiliate of Toyota Motor. "We want to build a track record through repeated dialogue with companies in Japan," said Tadao Nagasaki, the newly named Japan president for OpenAI, during a news conference Monday. The Tokyo office -- OpenAI's third overseas outpost following offices in London and Dublin -- will grow to about 10 to 20 workers this year, he said.