LLM and Knowledge Graphs; GPT-4 with Wolfram; CHITA by Google AI; Train ChatGPT on Your Documents via APIs; Why Kindness At Work Pays Off; and More

Editor's Paper Recommendations

Large Language Models and Knowledge Graphs: Opportunities and Challenges: Large Language Models (LLMs) have taken Knowledge Representation -- and the world -- by storm. This inflection point marks a shift from explicit knowledge representation to a renewed focus on the hybrid representation of explicit and parametric knowledge. In this position paper, we will discuss some of the common debate points within the community on LLMs (parametric knowledge) and Knowledge Graphs (explicit knowledge) and speculate on opportunities and visions that the renewed focus brings, as well as related research topics and challenges.

Language models as master equation solvers: Master equations play a pivotal role in depicting stochastic dynamic systems. Nevertheless, tackling these equations proves demanding due to the exponential surge in potential states or trajectories linked to the state space's dimensions. This investigation proposes an innovative application of language models for addressing master equations using machine learning. A prompt-based neural network is crafted to directly correlate rate parameters, initial conditions, and temporal values to states' precise joint probability distribution, aligning perfectly with the provided inputs. Through this method, a close approximation to the comprehensive solution of the master equation is achieved. The network is honed via the policy gradient technique within a reinforcement learning framework, with feedback rewards gleaned from a set of variational autoregressive models. Applying this methodology to representative instances reveals remarkable accuracy, even in intricate and multi-dimensional systems. The trained network showcases the ability to extrapolate, thus extending its predictive capacity to unfamiliar data points. This exploration establishes a significant link between language models and master equations, underscoring the prospect of employing a solitary pre-trained expansive model to resolve any master equation effectively.

Testing GPT-4 with Wolfram Alpha and Code Interpreter plug-ins on math?and science problems: This report describes a test of the large language model GPT-4 with the Wolfram Alpha and the Code Interpreter plug-ins on 105 original problems in science and math at the high school and college levels, carried out in June-August 2023. Our tests suggest that the plug-ins significantly enhance GPT's ability to solve these problems. There are still often "interface" failures; GPT often has trouble formulating problems in a way that elicits useful answers from the plug-ins. Fixing these interface failures seems to be a central challenge in making GPT a reliable tool for college-level calculation problems.

Challenges and Opportunities of Using Transformer-Based Multi-Task Learning in NLP Through ML Lifecycle: A Survey: The increasing adoption of natural language processing (NLP) models across industries has led to practitioners' need for machine learning systems to handle these models efficiently, from training to serving them in production. However, training, deploying, and updating multiple models can be complex, costly, and time-consuming, mainly when using transformer-based pre-trained language models. Multi-task learning (MTL) has emerged as a promising approach to improve efficiency and performance through joint training rather than training separate models. Motivated by this, we first provide an overview of transformer-based MTL approaches in NLP. Then, we discuss the challenges and opportunities of using MTL approaches throughout typical ML lifecycle phases, specifically focusing on the challenges related to data engineering, model development, deployment, and monitoring phases. This survey focuses on transformer-based MTL architectures and, to our knowledge, is novel in that it systematically analyses how transformer-based MTL in NLP fits into ML lifecycle phases. Furthermore, we motivate research on the connection between MTL and continual learning (CL), as this area still needs to be explored. It would be practical to have a model that can handle both MTL and CL, as this would make it easier to periodically re-train the model, update it due to distribution shifts, and add new capabilities to meet real-world requirements.

ChatGPT for PLG: Talk with Your Salesforce or Segment Data

In today’s competitive landscape, Product Led Growth (PLG) is emerging as a crucial strategy for scaling your business. Central to PLG is deeply understanding your data and user activity, and knowing what campaigns are working and why is essential for high-growth and successful products.

What You'll Learn:

• Hands-on Demo: Building an app using Langchain and a vector database
• Data Privacy & Security: Best practices for keeping your internal data private and secure
• Multi-Model Data Handling: Strategies for integrating and querying data in different formats from various sources

Registration Link

Industry Insights

• Google AI Introduces CHITA: An Optimization-based Approach for Pruning Pre-Trained Neural Networks at Scale
• Meta releases Code Llama, a code-generating AI model
• You can now train ChatGPT on your own documents via API
• Platypus 2 70B AI Open-Source LLM Tops the Leaderboard

--

Are you looking to advertise a product, job opening, or event to an audience of over 35,000 AI researchers and engineers? Get in touch with us at?[email protected]?to explore your options.

Enjoy the newsletter? Help us make it bigger and better by sharing it with colleagues and friends.

--

Growth Zone

Why Kindness at Work Pays Off