Best AI tools for< Improving The Safety Of Llms >

Prompt Engineering is a discipline focused on developing and optimizing prompts to efficiently utilize language models (LMs) for various applications and research topics. It involves skills to understand the capabilities and limitations of large language models, improving their performance on tasks like question answering and arithmetic reasoning. Prompt engineering is essential for designing robust prompting techniques that interact with LLMs and other tools, enhancing safety and building new capabilities by augmenting LLMs with domain knowledge and external tools.

Covera Health is a clinical intelligence platform that supports the end-to-end delivery of clinical-grade, AI-powered quality insights for providers and insurers. The platform is seamlessly integrated across the healthcare ecosystem to elevate everything from diagnosis and care coordination to prior authorization and claims administration. Covera Health is certified by AHRQ as a Patient Safety Organization to safeguard access to provider and patient data.

VirtuSense Technologies is a leading provider of fall prevention and remote patient monitoring (RPM) solutions powered by artificial intelligence (AI). Their AI-driven solutions, VSTAlert and VSTBalance, are designed to help healthcare providers reduce falls, improve patient safety, and enhance care delivery. VSTAlert is a fall prevention system that uses AI to detect falls before they happen, reducing the risk of injury and improving patient outcomes. VSTBalance is a balance assessment tool that helps clinicians identify patients at risk of falling and provides personalized exercises to improve their balance and mobility. VirtuSense's solutions integrate with various healthcare systems and are used by hospitals, post-acute care facilities, and ambulatory care centers to improve patient care and reduce costs.

Limbic is a clinical AI application designed for mental healthcare providers to save time, improve outcomes, and maximize impact. It offers a suite of tools developed by a team of therapists, physicians, and PhDs in computational psychiatry. Limbic is known for its evidence-based approach, safety focus, and commitment to patient care. The application leverages AI technology to enhance various aspects of the mental health pathway, from assessments to therapeutic content delivery. With a strong emphasis on patient safety and clinical accuracy, Limbic aims to support clinicians in meeting the rising demand for mental health services while improving patient outcomes and preventing burnout.

DrugCard is an AI-enabled Data Intelligence platform designed to streamline drug safety routines for pharmacovigilance processes. It offers solutions for local literature screening, catering to CROs, MAHs, and freelancers in the pharmaceutical industry. With support for multiple languages and regions, DrugCard ensures continuous, transparent, and scalable drug safety processes, saving time and improving efficiency. The platform leverages AI technology to enhance pharmacovigilance practices, providing accurate and holistic screening of medical journals to meet regulatory requirements.

FYLD is an award-winning digital platform that utilizes machine learning to automatically transform video and audio footage into real-time workflows, video risk assessments, and analytics dashboards. It aims to eliminate paperwork, save time, and create safer work sites for various sectors such as Highways, Energy, Water, and Wastewater. FYLD helps managers prioritize high-risk sites, reduces paperwork, and enhances efficiency by providing remote visibility of site conditions. The platform empowers fieldworkers, contractors, and civil engineers by streamlining job processes, improving safety measures, and minimizing environmental impact.

Carnegie Mellon University's School of Computer Science (SCS) is a world-renowned institution dedicated to advancing the field of computer science and training the next generation of innovators. With a rich history of groundbreaking research and a commitment to excellence in education, SCS offers a comprehensive range of programs, from undergraduate to doctoral levels, covering various specializations within computer science. The school's faculty are leading experts in their respective fields, actively engaged in cutting-edge research and collaborating with industry partners to solve real-world problems. SCS graduates are highly sought after by top companies and organizations worldwide, recognized for their exceptional skills and ability to drive innovation.

Inductor is a developer tool for evaluating, ensuring, and improving the quality of your LLM applications – both during development and in production. It provides a fantastic workflow for continuous testing and evaluation as you develop, so that you always know your LLM app’s quality. Systematically improve quality and cost-effectiveness by actionably understanding your LLM app’s behavior and quickly testing different app variants. Rigorously assess your LLM app’s behavior before you deploy, in order to ensure quality and cost-effectiveness when you’re live. Easily monitor your live traffic: detect and resolve issues, analyze usage in order to improve, and seamlessly feed back into your development process. Inductor makes it easy for engineering and other roles to collaborate: get critical human feedback from non-engineering stakeholders (e.g., PM, UX, or subject matter experts) to ensure that your LLM app is user-ready.

Article Fiesta is an AI-powered content creation platform that helps users generate high-quality, SEO-optimized articles in just a few clicks. With Article Fiesta, you can create unique, plagiarism-free content that is ready to be published on your blog or website. Article Fiesta's AI uses advanced techniques to ensure that your content not only reads well but also ranks well in search engines.

Happysales.ai is an AI-powered sales automation tool designed to revolutionize sales outreach and efficiency. It combines internal data with internet-scale intelligence to provide strategic insights, personalized emails, and role-specific messaging for each prospect. The tool offers features such as AI-driven prospect intelligence, personalized recommendations, conversation starters, practice pitch simulations, hyper-personalized outreach, multilingual support, and LinkedIn integration. Happysales AI aims to empower sales teams with scalable AI solutions for prospecting, engagement, and training, ultimately improving response rates and sales outcomes.

GitPoet is an AI-powered tool that generates meaningful and accurate git commit messages based on your git diff. It utilizes advanced AI technology, specifically ChatGPT-3.5 and ChatGPT-4 pro, to streamline your workflow and save you valuable time. With GitPoet, users can easily create automated commit messages by simply pasting their git diff output.

Adobe Firefly is a cloud-based AI platform that helps businesses automate and accelerate their creative processes. It provides a suite of tools for image editing, video editing, and audio editing, all powered by AI. With Firefly, businesses can save time and money on their creative projects, while also improving the quality of their work.

Gerwin AI is a neural network content generator and writer assistant that helps users create unique text and images using artificial intelligence. It offers a range of features for marketers, entrepreneurs, copywriters, and agencies, including automated text writing, social media post creation, article and long-form content generation, and image generation. Gerwin AI is designed to save users time and money while improving the quality and consistency of their content.

Applitools is an AI-powered test automation platform that helps businesses improve the quality of their digital experiences. It uses visual AI to validate user interfaces across any type of screen or device, and it can be deployed on-prem, in the cloud, or as a SaaS solution. Applitools integrates with all of the major development tools and workflows, and it offers a wide range of features and advantages that can help businesses save time and money while improving the quality of their software.

Zipscore.ai is an AI-powered platform that helps businesses automate their recruiting processes. It uses machine learning to screen and rank candidates, schedule interviews, and make hiring decisions. Zipscore.ai is designed to help businesses save time and money while improving the quality of their hires.

New Relic is an AI monitoring platform that offers an all-in-one observability solution for monitoring, debugging, and improving the entire technology stack. With over 30 capabilities and 750+ integrations, New Relic provides the power of AI to help users gain insights and optimize performance across various aspects of their infrastructure, applications, and digital experiences.

EasyDaddy is an AI-powered tool that simplifies and automates the process of filling out online forms. With EasyDaddy, users can create a profile once and use it to fill out forms on any website, saving time and hassle. EasyDaddy's AI learns from each form submission, improving the accuracy and relevance of future responses. The tool also allows users to attach files, making it even easier to complete forms.

WowTo is an all-in-one support video platform that helps businesses create how-to videos, host video knowledge bases, and provide in-app video help. With WowTo's AI-powered video creator, businesses can easily create step-by-step how-to videos without any prior design expertise. WowTo also offers a variety of pre-made video knowledge base layouts to choose from, making it easy to create a professional-looking video knowledge base that matches your brand. In addition, WowTo's in-app video widget allows businesses to provide contextual video help to their visitors, improving the customer support experience.

Walles.AI is a cloud-based AI-powered writing assistant that helps businesses create high-quality content, including articles, blog posts, social media posts, and more. It uses natural language processing (NLP) and machine learning (ML) to analyze data, generate text, and provide feedback on writing style and tone. Walles.AI is designed to help businesses save time and money on content creation while also improving the quality of their writing.

Cincinnati AI Catalyst is a platform dedicated to improving the lives of people in the Cincinnati Region by providing an inclusive, coordinated, collective Artificial Intelligence capability. The platform is committed to Responsible AI, enabling new products and services, attracting capital, creating/preserving jobs, developing/improving skills, and providing a trusted source of AI-related communication and education.

This repository contains code and results for jailbreaking leading safety-aligned LLMs with simple adaptive attacks. We show that even the most recent safety-aligned LLMs are not robust to simple adaptive jailbreaking attacks. We demonstrate how to successfully leverage access to logprobs for jailbreaking: we initially design an adversarial prompt template (sometimes adapted to the target LLM), and then we apply random search on a suffix to maximize the target logprob (e.g., of the token ``Sure''), potentially with multiple restarts. In this way, we achieve nearly 100% attack success rate---according to GPT-4 as a judge---on GPT-3.5/4, Llama-2-Chat-7B/13B/70B, Gemma-7B, and R2D2 from HarmBench that was adversarially trained against the GCG attack. We also show how to jailbreak all Claude models---that do not expose logprobs---via either a transfer or prefilling attack with 100% success rate. In addition, we show how to use random search on a restricted set of tokens for finding trojan strings in poisoned models---a task that shares many similarities with jailbreaking---which is the algorithm that brought us the first place in the SaTML'24 Trojan Detection Competition. The common theme behind these attacks is that adaptivity is crucial: different models are vulnerable to different prompting templates (e.g., R2D2 is very sensitive to in-context learning prompts), some models have unique vulnerabilities based on their APIs (e.g., prefilling for Claude), and in some settings it is crucial to restrict the token search space based on prior knowledge (e.g., for trojan detection).

OpenRedTeaming is a repository focused on red teaming for generative models, specifically large language models (LLMs). The repository provides a comprehensive survey on potential attacks on GenAI and robust safeguards. It covers attack strategies, evaluation metrics, benchmarks, and defensive approaches. The repository also implements over 30 auto red teaming methods. It includes surveys, taxonomies, attack strategies, and risks related to LLMs. The goal is to understand vulnerabilities and develop defenses against adversarial attacks on large language models.

Reinforcement Learning from AI Feedback (RLAIF) is a concept that describes a type of machine learning approach where **an AI agent learns by receiving feedback or guidance from another AI system**. This concept is closely related to the field of Reinforcement Learning (RL), which is a type of machine learning where an agent learns to make a sequence of decisions in an environment to maximize a cumulative reward. In traditional RL, an agent interacts with an environment and receives feedback in the form of rewards or penalties based on the actions it takes. It learns to improve its decision-making over time to achieve its goals. In the context of Reinforcement Learning from AI Feedback, the AI agent still aims to learn optimal behavior through interactions, but **the feedback comes from another AI system rather than from the environment or human evaluators**. This can be **particularly useful in situations where it may be challenging to define clear reward functions or when it is more efficient to use another AI system to provide guidance**. The feedback from the AI system can take various forms, such as: - **Demonstrations** : The AI system provides demonstrations of desired behavior, and the learning agent tries to imitate these demonstrations. - **Comparison Data** : The AI system ranks or compares different actions taken by the learning agent, helping it to understand which actions are better or worse. - **Reward Shaping** : The AI system provides additional reward signals to guide the learning agent's behavior, supplementing the rewards from the environment. This approach is often used in scenarios where the RL agent needs to learn from **limited human or expert feedback or when the reward signal from the environment is sparse or unclear**. It can also be used to **accelerate the learning process and make RL more sample-efficient**. Reinforcement Learning from AI Feedback is an area of ongoing research and has applications in various domains, including robotics, autonomous vehicles, and game playing, among others.

Awesome-Jailbreak-on-LLMs is a collection of state-of-the-art, novel, and exciting jailbreak methods on Large Language Models (LLMs). The repository contains papers, codes, datasets, evaluations, and analyses related to jailbreak attacks on LLMs. It serves as a comprehensive resource for researchers and practitioners interested in exploring various jailbreak techniques and defenses in the context of LLMs. Contributions such as additional jailbreak-related content, pull requests, and issue reports are welcome, and contributors are acknowledged. For any inquiries or issues, contact [email protected]. If you find this repository useful for your research or work, consider starring it to show appreciation.

This repository contains a curated list of **papers using Large Language/Multi-Modal Models for Robotics/RL**. Template from awesome-Implicit-NeRF-Robotics Please feel free to send me pull requests or email to add papers! If you find this repository useful, please consider citing and STARing this list. Feel free to share this list with others! ## Overview * Surveys * Reasoning * Planning * Manipulation * Instructions and Navigation * Simulation Frameworks * Citation

This repository is a curated list of papers focused on the detection of Large Language Models (LLMs)-generated content. It includes the latest research papers covering detection methods, datasets, attacks, and more. The repository is regularly updated to include the most recent papers in the field.

LLM-PLSE-paper is a repository focused on the applications of Large Language Models (LLMs) in Programming Language and Software Engineering (PL/SE) domains. It covers a wide range of topics including bug detection, specification inference and verification, code generation, fuzzing and testing, code model and reasoning, code understanding, IDE technologies, prompting for reasoning tasks, and agent/tool usage and planning. The repository provides a comprehensive collection of research papers, benchmarks, empirical studies, and frameworks related to the capabilities of LLMs in various PL/SE tasks.

The 'llm-misinformation-survey' repository is dedicated to the survey on combating misinformation in the age of Large Language Models (LLMs). It explores the opportunities and challenges of utilizing LLMs to combat misinformation, providing insights into the history of combating misinformation, current efforts, and future outlook. The repository serves as a resource hub for the initiative 'LLMs Meet Misinformation' and welcomes contributions of relevant research papers and resources. The goal is to facilitate interdisciplinary efforts in combating LLM-generated misinformation and promoting the responsible use of LLMs in fighting misinformation.

The LLM course is divided into three parts: 1. 🧩 **LLM Fundamentals** covers essential knowledge about mathematics, Python, and neural networks. 2. 🧑‍🔬 **The LLM Scientist** focuses on building the best possible LLMs using the latest techniques. 3. 👷 **The LLM Engineer** focuses on creating LLM-based applications and deploying them. For an interactive version of this course, I created two **LLM assistants** that will answer questions and test your knowledge in a personalized way: * 🤗 **HuggingChat Assistant**: Free version using Mixtral-8x7B. * 🤖 **ChatGPT Assistant**: Requires a premium account. ## 📝 Notebooks A list of notebooks and articles related to large language models. ### Tools | Notebook | Description | Notebook | |----------|-------------|----------| | 🧐 LLM AutoEval | Automatically evaluate your LLMs using RunPod |  | | 🥱 LazyMergekit | Easily merge models using MergeKit in one click. |  | | 🦎 LazyAxolotl | Fine-tune models in the cloud using Axolotl in one click. |  | | ⚡ AutoQuant | Quantize LLMs in GGUF, GPTQ, EXL2, AWQ, and HQQ formats in one click. |  | | 🌳 Model Family Tree | Visualize the family tree of merged models. |  | | 🚀 ZeroSpace | Automatically create a Gradio chat interface using a free ZeroGPU. |  |

A comprehensive repository focusing on 'Model Merging in LLMs, MLLMs, and Beyond', providing an exhaustive overview of model merging methods, theories, applications, and future research directions. The repository covers various advanced methods, applications in foundation models, different machine learning subfields, and tasks like pre-merging methods, architecture transformation, weight alignment, basic merging methods, and more.

LangTest is a comprehensive evaluation library for custom LLM and NLP models. It aims to deliver safe and effective language models by providing tools to test model quality, augment training data, and support popular NLP frameworks. LangTest comes with benchmark datasets to challenge and enhance language models, ensuring peak performance in various linguistic tasks. The tool offers more than 60 distinct types of tests with just one line of code, covering aspects like robustness, bias, representation, fairness, and accuracy. It supports testing LLMS for question answering, toxicity, clinical tests, legal support, factuality, sycophancy, and summarization.

Awesome MLSecOps is a curated list of open-source tools, resources, and tutorials for MLSecOps (Machine Learning Security Operations). It includes a wide range of security tools and libraries for protecting machine learning models against adversarial attacks, as well as resources for AI security, data anonymization, model security, and more. The repository aims to provide a comprehensive collection of tools and information to help users secure their machine learning systems and infrastructure.

LLaMA Factory is a unified framework for fine-tuning 100+ large language models (LLMs) with various methods, including pre-training, supervised fine-tuning, reward modeling, PPO, DPO and ORPO. It features integrated algorithms like GaLore, BAdam, DoRA, LongLoRA, LLaMA Pro, LoRA+, LoftQ and Agent tuning, as well as practical tricks like FlashAttention-2, Unsloth, RoPE scaling, NEFTune and rsLoRA. LLaMA Factory provides experiment monitors like LlamaBoard, TensorBoard, Wandb, MLflow, etc., and supports faster inference with OpenAI-style API, Gradio UI and CLI with vLLM worker. Compared to ChatGLM's P-Tuning, LLaMA Factory's LoRA tuning offers up to 3.7 times faster training speed with a better Rouge score on the advertising text generation task. By leveraging 4-bit quantization technique, LLaMA Factory's QLoRA further improves the efficiency regarding the GPU memory.

This repository serves as a comprehensive hub for updates on generative AI research, interview materials, notebooks, and more. It includes monthly best GenAI papers list, interview resources, free courses, and code repositories/notebooks for developing generative AI applications. The repository is regularly updated with the latest additions to keep users informed and engaged in the field of generative AI.

"Goods Guru" represents a fusion of AI technology and in text and visual content creation, aimed at boosting online sales and improving the digital footprint of e-commerce businesses.

GPT created by Max Del Rosso for SEO optimization, specialized in identifying internal linking opportunities. Through the review of existing content, it suggests targeted changes to integrate effective anchor texts, contributing to improving SERP rankings and user experience.

Sarcastic Silicon Valley investor AI, roasting and improving startup pitches with puns.

Improved ChatGPT with Response Review and Persistent Contexts

A virtual doctor for cognitive assessment.

Subtle, adaptable interactive murder mystery host.

I only respond in jazz scats and skibidibops, mimicking scat singing.

Evaluates faces and rates them out of 10 ⭐ Provides valuable feedback to improving your attractiveness!

Guides in creating or improving BDD scenarios

A knowledgeable guide for improving your Minecraft Java skills.

Specialist in interpreting and improving pet behavior

A supportive guide for improving posture with personalized tips and exercises.

Offering guidance on improving social interactions with empathetic, respectful, and effective communication strategies.

Tech expert on improving phone speed and performance.

An assistant for annotating and improving fiction writing, chapter by chapter.

A critical thinker for analyzing questions and improving answers

I edit articles, improving clarity and respectfulness, maintaining your style.

Analyzes behavioral data to understand user interactions and preferences, improving product designs.

Hyper-intelligent quill for summarizing papers, writing LaTeX, and improving scientific writing.

Your AI research assistant, for turning a problem into a research, developing research questions, generating plans, analyzing data and improving research workflows for project success