Blog

Introduction I’ve written a lot in recent months about what I call Large Action Models, or LAMs—a more active, autonomous variation on LLMs that don’t merely generate content like text or images but accomplish entire tasks and even participate in workflows, whether alongside people or on their

TL;DR: With CodeChain, a pretrained large language model (LLM) can solve challenging coding problems by integrating modularity in generation samples and self-improve by employing a chain of self-revisions on representative sub-modules. CodeChain can achieve state-of-the-art results with both OpenAI GPT models and open-source LLMs on challenging coding benchmarks like

TL;DR: We adapted our protein language model ProGen to optimize antibodies that bind to a protein called “CD40L”, a critical target for autoimmune disorders. We tested our AI designed antibodies in the laboratory and found that they bound very tightly to CD40L, showcasing the potential of this approach for

Other authors include: Can Qin, Stefano Ermon, Yun Fu GlueGen was accepted by ICCV. In the rapidly advancing field of text-to-image synthesis, the remarkable progress in generating lifelike images from textual prompts has been evident. However, a significant challenge remains: how can we seamlessly integrate powerful pre-trained text encoders into

Introduction Open source has become one of the hottest topics in AI, and the fanfare is well-deserved. The open source community is keeping a nimble pace with the state of the art, delivering ever-growing and ever-more-capable models that often compete impressively with their commercial counterparts. It’s an exciting thing

Meet the Office of Ethical and Humane Use Salesforce's Office of Ethical and Humane Use provides navigational guidance for the tough questions that arise when human potential meets emergent technology. We work across the company to guide the design, development, and deployment of trusted products, with a strong focus on

TLDR Generative AI methods for image generation have a wide variety of potential applications in marketing, sales, and e-commerce. With these applications in mind, the Salesforce Research team has developed several techniques based on image-generative diffusion models, including methods for image editing, improved classifier guidance, and improved controlled generation methods.

TL;DR: PyRCA is an open-source machine learning library specifically designed for conducting Root Cause Analysis (RCA) in IT operations. It offers a comprehensive framework that allows users to easily identify the complicated metric causal dependencies and automatically locate the root causes of incidents. The library provides a unified interface

Equal contribution between Erik Nijkamp and Hiroaki Hayashi. Paper Code Tweet Abstract The family of Salesforce CodeGen models is growing with CodeGen2.5 – a small, but mighty model! While there has been a recent trend of large language models (LLM) of increasing size, we show that a small model can

TLDR We trained a series of 7B LLMs named XGen-7B with standard dense attention on up to 8K sequence length for up to 1.5T tokens. We also fine tune the models on public-domain instructional data. The main take-aways are: * On standard NLP benchmarks, XGen achieves comparable or better results

LAMs: From Large Language Models to Large Action Models There’s no question that we’re living in the era of generative AI, and its impact is only growing. More and more, AI is helping us write emails, create imagery, consume information, and even code. But as empowering as it

Authors: Vibashan Vishnukumar Sharmini, Ning Yu, Ran Xu Have you ever wondered how long it takes for a human annotator to annotate a dataset like COCO? MORE THAN A YEAR. Not to mention, even training a detection model on this dataset would only equip it to detect those specific 80

TL;DR: Imagine a world where machines comprehend 3D objects just as humans do. The ULIP (CVPR2023) and ULIP-2 projects, backed by Salesforce AI, are making this a reality by revolutionizing 3D understanding. ULIP uniquely pre-trains models with 3D point clouds, images, and texts, aligning them into a unified representation

TL;DR: CodeT5+ is a new family of open code large language models (LLMs) with improved model architectures and training techniques. CodeT5+ achieves the state-of-the-art performance among the open-source LLMs on many challenging code intelligence tasks, including zero-shot evaluation on the code generation benchmark HumanEval.   Background: Code LLMs Large language

TLDR We present ConRad, a novel approach for generating a 3D model from a single RGB image. We introduce a 3D representation that allows us to explicitly constrain the appearance of the object using the input image which leads to preservation of depicted details.   Background Over the last few years,

We have established a set of five guidelines that build upon our Trusted AI Principles to provide more detailed guidance for the responsible development and implementation of GenAI.

It is not enough to deliver only the technological capabilities of AI – we also have an important responsibility to ensure that AI is safe and inclusive for all. These are our commitments to building AI responsibly.

TL;DR LogAI is an open-source library designed for log analytics and intelligence. It can process raw logs generated by computer systems and support log analytics tasks such as log clustering and summarization, as well as log intelligence tasks such as log anomaly detection and root-cause analysis. LogAI is compatible

The Salesforce AI Team is mourning the loss of our beloved friend and mentor, Dragomir Radev. Our team was first introduced to Drago in November 2018 when he gave a talk at our Research Speaker Series. His passion for research beamed through his talk and our leadership team unanimously decided

BLIP-2: Scalable Pre-training of Multimodal Foundation Models for the World's First Open-source Multimodal Chatbot

Salesforce Data Cloud, the first real-time CRM, is turning your data into real-time customer magic. You might have witnessed how our AI research is powering intelligent experiences with Identity Resolution for Fuzzy Matching featured at Dreamforce. The goal of identity resolution is to identify the same individuals across datasets and

TLDR; We introduce the Salesforce CausalAI Library, an open source library for causal analysis of time series and tabular data. The Salesforce CausalAI Library aims to provide a one-stop solution to the various needs in causal analysis including handling different data types, data generation, multi-processing for speed-up, utilizing domain knowledge

TL;DR: Text-to-image diffusion models are very adept at generating novel images from a text prompt, but current adaptations of these methods to image editing suffer from a lack of consistency and faithfulness to the original image. Many of these discrepancies can be traced to the difficulty of inverting the

TL;DR: Slack is a powerful productivity tool that professionals and teams around the globe depend on to communicate and keep up with vital information at work, making async work efficient. However, with an increase in the number (and variety) of conversations on Slack that a typical user keeps track

TL;DR: We present BotSIM, a data-efficient end-to-end Bot SIMulation toolkit for evaluation, diagnosis, and improvement of commercial task-oriented dialogue (TOD) systems. BotSIM's “generation-simulation-remediation'' paradigm can accelerate the end-to-end bot evaluation and iteration process by: (1) reducing the effort needed to create test cases; (2) enabling a better understanding of

We introduce an automated method for evaluating the quality of AI-generated text by repurposing prior human evaluation data. The method is called Near-Negative Distinction.

Conference Overview Next week, the Thirty-sixth annual Conference on Neural Information Processing Systems (NeurIPS) will be held in New Orleans, Louisiana from Monday, November 28th, through Friday, December 9th. NeurIPS will include invited talks, demonstrations, oral and poster presentations of accepted papers. Along with the conference is a professional exposition

Background We live in a digital world where many of us depend on customer service to be at its best. Whether it's calling about an order issue or needing help to activate an account, we are always reaching out to customer service to save the day. We are no longer

TL;DR: Applications are now open for the next cohort of Salesforce’s AI Residency Program. We invite researchers and engineers to join our 12-month training program, designed to help launch their careers while serving as AI Residents at Salesforce AI Research. Candidates based in the U.S. can apply

We summarize insights from our FAccT 2022 CRAFT workshop that focused on Concepts of Fairness & Transparency, Applied RAI Practices, Organizational Approaches to RAI & Cultural Change, and Public Policy & Regulation.

TL;DR: Deep reinforcement learning (RL), a powerful learning framework to train AI agents, can be slow as it requires repeated interaction with a simulation of the environment. Our original WarpDrive accelerates multi-agent deep RL on NVIDIA GPUs, enabling 10-100x speedups compared to alternative CPU+GPU implementations of multi-agent simulations.

AUTHORS: Chenghao Liu, Quang Pham, Doyen Sahoo, Donald Rose TL;DR: Nonstationary data, which changes its statistical properties over time, can make time series forecasting difficult. Despite the recent success of deep learning techniques for time series forecasting tasks, these methods are not scalable for applications where data arrives sequentially

TL;DR: The performance of existing time-series forecasting methods can degrade due to non-stationarity, where the statistical distribution of time-series data changes over time. Our new DeepTime method overcomes non-stationarity issues by leveraging a “forecasting as meta-learning” framework on deep time-index models. DeepTime achieves competitive accuracy on the long-sequence time-series

AUTHORS: Zeyuan Chen, Ran Xu, Luyu Yang, Donald Rose TL;DR: Many methods designed to preserve online privacy propose complex security measures to protect sensitive data. We believe that not storing any sensitive data is the optimal way to preserve privacy, so we propose a “burn after reading” online framework:

Imagine you find yourself in the cockpit of a next-generation spacecraft—the kind that can get you from low Earth orbit to the Kuiper belt without breaking a sweat. How do you envision controlling it? Science fiction has conditioned us to equate futuristic technology with dazzling complexity, so you might

As we say a fond farewell to summer (bummer!), let's look back and review some of the stellar work reported on by Salesforce AI researchers during the past few months. (For more details, we encourage you to click the link for each project to read the full blog post.) --------------------------------------------------------------------------------

TL;DR: LAVIS (short for LAnguage-VISion) is an open-source deep learning library for language-vision research and applications, offering comprehensive support for a wide range of tasks, datasets, and state-of-the-art models. Featuring a unified interface and modular design, it’s easy to use off-the-shelf and to extend with new capabilities. With

TL;DR: We developed a new time-series forecasting model called ETSformer that leverages the power of two frameworks. By combining the classical intuition of seasonal-trend decomposition and exponential smoothing with modern transformers – as well as introducing novel exponential smoothing and frequency attention mechanisms – ETSformer achieves state-of-the-art performance. Background Before diving

AUTHORS: Chen Xing, Mingfei Gao, Donald Rose TL;DR: Most AI object detection methods work only on limited object categories, due to the human effort required for bounding-box annotations of training data. We developed a new method that automatically generates pseudo bounding-box annotations of diverse objects from large-scale image-caption pairs,

We brought together 24 practitioners from 17 organizations to talk about a success in their practice they believed others could emulate or a challenge they were facing & wanted to brainstorm with others. We summarize the insights from 2 breakout discussions: Governance/Accountability & Incentives.

TL;DR:  Salesforce Research and Mila announce AI for Global Climate Cooperation, a working group collaboration and competition to design negotiation protocols and climate agreements. We plan to coauthor a peer-reviewed scientific paper with top-performing teams; insights will be distilled into a policy brief shared with leading policymakers, informing future

TL;DR: CodeRL is a new framework for program synthesis through holistic integration of pretrained language models and deep reinforcement learning. By utilizing unit test feedback as part of model training and inference, and integrating with an improved CodeT5 model, CodeRL achieves state-of-the-art results on competition-level programming tasks. The following

Conference Overview This weekend will kick off the thirty-ninth International Conference on Machine Learning (ICML). This conference specifically aims to bring together professionals who are dedicated to the advancement of Machine Learning (ML) in Artificial Intelligence. Participants at ICML come from many different backgrounds, including academic and industrial researchers, entrepreneurs

Conference Overview This weekend marks the start of the Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL). NAACL provides a regional focus for members of the Association for Computational Linguistics (ACL) in North America. NAACL organizes annual conferences, promotes cooperation and information exchange among

Conference Overview The IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR) is the annual conference on Computer Vision. CVPR is composed of both the main conference, as well as workshops and other courses, to provide a unique learning experience and networking opportunities in the field of Computer Vision. CVPR

AUTHORS: Sharvin Shah, Jin Qu, Donald Rose TL;DR: TaiChi is an open source library for few-shot NLP, designed for data scientists and software engineers who want to get some quick results or build proof-of-concept products but don’t have much experience with few-shot learning (FSL). The library abstracts complex

TL;DR:OmniXAI (short for Omni eXplainable AI) is designed to address many of the pain points in explaining decisions made by AI models. This open-source library aims to provide data scientists, machine learning engineers, and researchers with a one-stop Explainable AI (XAI) solution to analyze, debug, and interpret their

AUTHORS: Huan Wang, Aadyot Bhatnagar, Doyen Sahoo, Wenzhuo Yang, Steven Hoi, Caiming Xiong, Donald Rose TL;DR: Time series data is a critical source of insights for many applications, including IT Operations, Quality Management, Financial Analytics, and Inventory & Sales Management. While a variety of dedicated packages and software exist, engineers

TL;DR: We propose ALPRO, a new video-and-language representation learning framework which achieves state-of-the-art performance on video-text retrieval and video question answering by learning fine-grained alignment between video regions and textual entities via entity prompts. For more background (a review of key concepts used in this post), please see the

Lead Author: Xi Ye TL;DR: We propose RnG-KBQA, a Rank-and-Generate Approach for Question Answering over Knowledge Bases, which enables answering natural language questions over large-scale knowledge bases. Our approach is capable of answering questions about topics never seen in the training data, which makes it generalizable to a broad