Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: information science – Page 112

Google DeepMind Researchers Introduce Promptbreeder: A Self-Referential and Self-Improving AI System that can Automatically Evolve Effective Domain-Specific Prompts in a Given Domain

Large Language Models (LLMs) have gained a lot of attention for their human-imitating properties. These models are capable of answering questions, generating content, summarizing long textual paragraphs, and whatnot. Prompts are essential for improving the performance of LLMs like GPT-3.5 and GPT-4. The way that prompts are created can have a big impact on an LLM’s abilities in a variety of areas, including reasoning, multimodal processing, tool use, and more. These techniques, which researchers designed, have shown promise in tasks like model distillation and agent behavior simulation.

The manual engineering of prompt approaches raises the question of whether this procedure can be automated. By producing a set of prompts based on input-output instances from a dataset, Automatic Prompt Engineer (APE) made an attempt to address this, but APE had diminishing returns in terms of prompt quality. Researchers have suggested a method based on a diversity-maintaining evolutionary algorithm for self-referential self-improvement of prompts for LLMs to overcome decreasing returns in prompt creation.

LLMs can alter their prompts to improve their capabilities, just as a neural network can change its weight matrix to improve performance. According to this comparison, LLMs may be created to enhance both their own capabilities and the processes by which they enhance them, thereby enabling Artificial Intelligence to continue improving indefinitely. In response to these ideas, a team of researchers from Google DeepMind has introduced PromptBreeder (PB) in recent research, which is a technique for LLMs to better themselves in a self-referential manner.

Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution (Paper Explained)

#evolution.

Promptbreeder is a self-improving self-referential system for automated prompt engineering. Give it a task description and a dataset, and it will automatically come up with appropriate prompts for the task. This is achieved by an evolutionary algorithm where not only the prompts, but also the mutation-prompts are improved over time in a population-based, diversity-focused approach.

OUTLINE:
0:00 — Introduction.
2:10 — From manual to automated prompt engineering.
10:40 — How does Promptbreeder work?
21:30 — Mutation operators.
36:00 — Experimental Results.
38:05 — A walk through the appendix.

Paper: https://arxiv.org/abs/2309.

Abstract:
Popular prompt strategies like Chain-of-Thought Prompting can dramatically improve the reasoning abilities of Large Language Models (LLMs) in various domains. However, such hand-crafted prompt-strategies are often sub-optimal. In this paper, we present Promptbreeder, a general-purpose self-referential self-improvement mechanism that evolves and adapts prompts for a given domain. Driven by an LLM, Promptbreeder mutates a population of task-prompts, and subsequently evaluates them for fitness on a training set. Crucially, the mutation of these task-prompts is governed by mutation-prompts that the LLM generates and improves throughout evolution in a self-referential way. That is, Promptbreeder is not just improving task-prompts, but it is also improving the mutationprompts that improve these task-prompts. Promptbreeder outperforms state-of-the-art prompt strategies such as Chain-of-Thought and Plan-and-Solve Prompting on commonly used arithmetic and commonsense reasoning benchmarks. Furthermore, Promptbreeder is able to evolve intricate task-prompts for the challenging problem of hate speech classification.

Authors: Chrisantha Fernando, Dylan Banarse, Henryk Michalewski, Simon Osindero, Tim Rocktäschel.

Continue reading “Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution (Paper Explained)” | >

Meta-Learning Machines in a Single Lifelong Trial

The most widely used machine learning algorithms were designed by humans and thus are hindered by our cognitive biases and limitations. Can we also construct meta-learning algorithms that can learn better learning algorithms so that our self-improving AIs have no limits other than those inherited from computability and physics? This question has been a main driver of my research since I wrote a thesis on it in 1987. In the past decade, it has become a driver of many other people’s research as well. Here I summarize our work starting in 1994 on meta-reinforcement learning with self-modifying policies in a single lifelong trial, and — since 2003 — mathematically optimal meta-learning through the self-referential Gödel Machine. This talk was previously presented at meta-learning workshops at ICML 2020 and NeurIPS 2021. Many additional publications on meta-learning can be found at https://people.idsia.ch/~juergen/metalearning.html.

Jürgen Schmidhuber.
Director, AI Initiative, KAUST
Scientific Director of the Swiss AI Lab IDSIA
Co-Founder & Chief Scientist, NNAISENSE
http://www.idsia.ch/~juergen/blog.html.

AI predicts 70% of earthquakes a week before they occur

The system only flagged eight false warnings and missed one earthquake.

High precision and accuracy in earthquake prediction continues to be a key scientific challenge, and artificial intelligence (AI) has been investigated as a technique to enhance our capabilities in this crucial area.

This is because AI can analyze large datasets of seismic activity and identify patterns or anomalies that human analysts might miss. Machine learning algorithms can thus help researchers understand earthquake patterns better.

Stanford introduces autonomous robot dogs with AI brains

There’s a new kind of robot dog in town and it gets its prowess from an artificial intelligence (AI) algorithm.

An AI algorithm for a brain

The new vision-based algorithm, according to AI researchers at Stanford University and Shanghai Qi Zhi Institute who lead these efforts, enables the robodogs to scale tall objects, jump across gaps, crawl under low-hanging structures, and squeeze between cracks. This is because the robodog’s algorithm serves as its brain.

Researchers create a neural network for genomics that explains how it achieves accurate predictions

A team of New York University computer scientists has created a neural network that can explain how it reaches its predictions. The work reveals what accounts for the functionality of neural networks—the engines that drive artificial intelligence and machine learning—thereby illuminating a process that has largely been concealed from users.

The breakthrough centers on a specific usage of that has become popular in recent years—tackling challenging biological questions. Among these are examinations of the intricacies of RNA splicing—the focal point of the study—which plays a role in transferring information from DNA to functional RNA and protein products.

“Many neural networks are —these algorithms cannot explain how they work, raising concerns about their trustworthiness and stifling progress into understanding the underlying biological processes of genome encoding,” says Oded Regev, a computer science professor at NYU’s Courant Institute of Mathematical Sciences and the senior author of the paper, which was published in the Proceedings of the National Academy of Sciences.

New technique based on 18th-century mathematics shows simpler AI models don’t need deep learning

Researchers from the University of Jyväskylä were able to simplify the most popular technique of artificial intelligence, deep learning, using 18th-century mathematics. They also found that classical training algorithms that date back 50 years work better than the more recently popular techniques. Their simpler approach advances green IT and is easier to use and understand.

The recent success of artificial intelligence is significantly based on the use of one core technique: . Deep learning refers to techniques where networks with a large number of data processing layers are trained using massive datasets and a substantial amount of computational resources.

Deep learning enables computers to perform such as analyzing and generating images and music, playing digitized games and, most recently in connection with ChatGPT and other generative AI techniques, acting as a conversational agent that provides high-quality summaries of existing knowledge.

/* */