- NLPlanet Newsletter
- Posts
- ChatGPT vs Claude, Azure AI, and Getty Images suing Stable Diffusion
ChatGPT vs Claude, Azure AI, and Getty Images suing Stable Diffusion
ChatGPT vs Claude: Introducing Claude, the AI assistant from Anthropic.
![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/fd1cc30d-5d9d-4351-bbd2-d5f40be1ee32/Screenshot_2023-01-19_at_16.19.51.png?t=1674141627)
Here are your weekly articles, guides, and news about NLP and AI chosen for you by NLPlanet!
😎 News and Guides From The Web
ChatGPT vs Claude: Anthropic has quietly begun testing a new, ChatGPT-like AI assistant named Claude.
ChatGPT will soon be available on Azure AI infrastructure: Azure OpenAI Service is now generally available, enabling customers to tap into large-scale generative AI models.
2022: A Year Full of Amazing AI papers- A Review: 2022 has been an amazing year for AI and Data Science. Here's a curated list of the latest breakthroughs, complete with explanations, links, and code.
Chain of Thought Prompting: Chain of Thought (CoT) prompting is an effective way to encourage AI reasoning. It's been shown to improve accuracy on tasks like arithmetic, commonsense, and symbolic reasoning. Models with 100B+ parameters get the most benefit from CoT prompting.
Getty Images is suing the creators of AI art tool Stable Diffusion for scraping its content: Getty Images is suing Stability AI, creators of popular AI art tool Stable Diffusion, over an alleged copyright violation.
Reverse Prompt Engineering of Notion AI: The author used prompt injection to obtain the complete source prompts of every Notion AI feature, leveraging several techniques.
Transformer models: an introduction and catalog — 2023 Edition: A catalog of popular transformer models from the last years, including ChatGPT, GPT3.5, and other models from Eleuther, Anthropic, Deepmind, and Stability.
Language, Vision and Generative Models at Google in 2022: The article contains a summary of the AI work done by Google in 2022, talking about LaMDA, PaLM, and other NLP advancements.
Nvidia Debuts Generative AI Model for Medical Research: Nvidia and Evozyne have debuted ProT-VAE, a generative AI model capable of producing proteins to use in medicine and other industries. The model uses deep learning and generative AI to synthesize variants of proteins to create more effective enzymes.
Boston Dynamics' Atlas shows off its acrobatic 'gopher' skills: Boston Dynamics' Atlas robot shows off its amazing acrobatic skills in a new demo video. It grabs and places a plank, jumps and spins, and even hucks a tool bag with impressive dexterity.
Introducing Karlo, an Open Source DALL-E 2 (unCLIP): Karlo is a text-conditional image generation model based on OpenAI's unCLIP architecture with the improvement over the standard super-resolution model from 64px to 256px, recovering high-frequency details only in the small number of denoising steps.
🥳 NLPlanet News
New lessons "Project: Named Entity Recognition", "Knowledge Graphs", and "Project: Building a Knowledge Base from Texts" are out. New lessons will be published on January 30th.
Following feedbacks from the community, there's now a newsletter of NLPlanet.
🔬 Interesting Papers
A Survey on In-context Learning: In-context learning (ICL) is a new paradigm for natural language processing that uses large language models. This paper surveys and summarizes progress, challenges, and future work in ICL, including training and prompting strategies.
Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback: The authors apply preference modeling and reinforcement learning from human feedback (RLHF) to finetune language models to act as helpful and harmless assistants.
MemPrompt: Help LLMs correct errors without retraining: The authors explore how user interactions can help correct errors in large pre-trained LMs like GPT-3 without retraining. The approach pairs GPT-3 with a memory of recorded cases & user feedback for clarification.
A Survey on Transformers in Reinforcement Learning: The authors seek to systematically review motivations and progress on using Transformers in RL, provide a taxonomy on existing works, discuss each sub-field, and summarize future prospects.
💻 Interesting Repositories
CarperAI/OpenELM: The ELM approach shows that an LLM trained on code can suggest intelligent mutations for genetic programming (GP) algorithms.
microsoft/torchscale: TorchScale is a PyTorch library that allows researchers and developers to scale up Transformers efficiently and effectively.
Have a nice week!