Weekly AI and NLP News — March 20th 2023

Here are your weekly articles, guides, and news about NLP and AI chosen for you by NLPlanet!

😎 News From The Web

• GPT-4 is out. GPT-4 is multimodal and accepts also images as inputs. It can handle over 25,000 words of text. GPT-4 can be tested with ChatGPT Plus. Otherwise, there's an API waitlist.

• Microsoft announced Microsoft 365 Copilot, which combines the power of large language models (LLMs) with your data in the Microsoft Graph—your calendar, emails, chats, documents, meetings, and more—and the Microsoft 365 apps.

• Microsoft spent hundreds of millions of dollars on a ChatGPT supercomputer. Microsoft says it connected tens of thousands of Nvidia A100 chips and reworked server racks to build the hardware behind ChatGPT and its own Bing AI bot.

• PyTorch 2.0 is out.

• Microsoft just laid off one of its responsible AI teams.

• New generative AI features in Google Workspace and Cloud.

• Adept raised $350M in new funding as part of its Series B.

📚 Guides From The Web

• 5 ways GPT-4 outsmarts ChatGPT. GPT-4 can see and understand images, is harder to trick, and has a longer memory.

• What we know so far on GPT4.

• Prompt Engineering Is Probably More Important Than You Think.

🔬 Interesting Papers and Repositories

• Stanford published the Alpaca project, which aims to build and share an instruction-following LLaMA model.

• The GPT-4 technical report.

• Self-Instruct: Aligning Language Model with Self Generated Instructions. Self-Instruct is a framework for improving the instruction-following capabilities of pretrained language models by bootstrapping off its own generations

• High-throughput Generative Inference of Large Language Models with a Single GPU. The paper studies how to do high-throughput LLM inference using limited resources.

• Scaling Vision-Language Models with Sparse Mixture of Experts.

• ChatGPT Participates in a Computer Science Exam. The model slightly passes the "Algorithms and Data Structures" exam.

• Language Is Not All You Need: Aligning Perception with Language Models. The authors introduce Kosmos-1, a Multimodal Large Language Model (MLLM) that can perceive general modalities, learn in context (i.e., few-shot), and follow instructions (i.e., zero-shot).

Thank you for reading! If you want to learn more about NLP, remember to follow NLPlanet. You can find us on LinkedIn, Twitter, Medium, and our Discord server!