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

😎 News From The Web

• Anthropic’s $5B, 4-year plan to take on OpenAI. Anthropic aims to raise $5B in 2 yrs to compete with OpenAI using its new AI model, Claude-Next. It has various industries as potential customers.

• Samsung workers made a major error by using ChatGPT. Samsung workers leaked proprietary data through ChatGPT. OpenAI may now possess trade secrets. The incident raised alarms about data privacy and GDPR rule.

• Meta introduced Segment Anything: Working toward the first foundation model for image segmentation. Introducing Segment Anything: democratizing image segmentation with SAM - a versatile, promptable model trained on a versatile dataset under Apache 2.0.

• LangChain announced a $10M seed round led by Benchmark. Python and TypeScript package LangChain raised $10M to empower developers building language model-powered apps. Offers modular components to ease prototyping.

• $335,000 pay offered for ‘AI whisperer’ jobs in red-hot market. Skills in prompt engineering, training on AI tools, paying up to $335K p.a., are in demand. Red-hot market attracting various sectors, may be overestimating technology.

• Amazon tells employees it isn’t falling behind on AI. Amazon execs reassure cloud staff that despite no generative AI wars involvement, company not behind AI game. Partners with Stability AI, Hugging Face.

• Defamed by ChatGPT: My Own Bizarre Experience with Artificiality of “Artificial Intelligence”. AI program ChatGPT created false claim of sexual harassment w/o factual basis, warns of biases & dangers of AI in tackling disinformation.

📚 Guides From The Web

• StackLLaMA: A hands-on guide to train LLaMA with RLHF. Learn to train a LLaMA model using Reinforcement Learning from Human Feedback, Supervised Fine-tuning, and Reward Modeling.

• AI might not steal your job, but it could change it. AI can impact 300 million jobs globally. Legal industry shows promise in automating language-based jobs but human guidance is still vital for AI.

🔬 Interesting Papers and Repositories

• Torantulino/Auto-GPT: An experimental open-source attempt to make GPT-4 fully autonomous. Auto-GPT is an open-source app exemplifying GPT-4 capabilities, autonomously managing businesses to increase net worth.

• mayooear/gpt4-pdf-chatbot-langchain: GPT4 & LangChain Chatbot for large PDF docs. Make a question-answering chatbot with GPT-4 & LangChain over several large PDF files.

• ChatDoctor: A Medical Chat Model Fine-tuned. ChatDoctor uses 205k medical conversations to fine-tune a language model that can understand patients’ needs and give informed advice. The framework and dataset are publicly available for further dialogue model development.

• IDEA-Research/Grounded-Segment-Anything: Marrying Grounding DINO with Segment Anything & Stable Diffusion & BLIP. Combine Grounding DINO, Segment Anything, and Stable Diffusion models for a powerful pipeline solving complex problems. Demos, tips, and future possibilities included.

• HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in HuggingFace. HuggingGPT uses language models like ChatGPT to connect AI models in the Hugging Face community, simplifying complex AI tasks across modalities and domains.

• Summary of ChatGPT/GPT-4 Research. Survey of ChatGPT & GPT-4 models, including potential applications in education, medicine, physics, etc. Ethical concerns and future advancements discussed.

• ChatGPT Survey: Performance on NLP datasets. Survey of 141 arXiv papers comparing ChatGPT to other NLP models found its performance to be lower than expected. Reasons include suboptimal utilization, biased results, and incomplete multilingual evaluation.

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!