AI This Week: Claude’s New Feature Revolutionizes AI Interactions

Language Models: Anthropic’s Claude Now Lets You Work Smarter

Anthropic has launched a groundbreaking feature for its AI assistant, Claude, known as “Tool Use” or “function calling.” This innovation allows Claude to interact with external APIs and tools, enhancing its ability to perform complex tasks. By providing Claude with a set of tools and user prompts in natural language, the system can determine and execute the necessary actions to handle the requests. You can get started with tool use today on the Anthropic Messages API, Amazon Bedrock, and Google Cloud’s Vertex AI.

Claude can perform tasks such as data extraction, API calls, customer support, automation, and scheduling. Claude selects tools from predefined sets based on natural language requests. You have the option to enforce specific tool choices using the “tool_choice” parameter. Claude’s ability to process data in real-time improves the speed and quality of user interactions, notably in customer support applications. Claude cannot perform parallel tool calls but can handle both text and image data. Claude integrates with popular programming languages: Python, JavaScript, Java.

Trending Signals

• LlamaIndex launches a framework for building knowledge graphs with LLMs.
• Perplexity announces “Pages”, a new feature that generates a customizable webpage based on prompts.
• Scale AI Releases their own LLM leaderboard, the first expert-driven, trustworthy LLM contest open to all.
• You can now start building with Gemini 1.5 Flash and Pro models via API, 1000 req/min.
• ElevenLabs latest model generates sound effects, short instrumental tracks, soundscapes, and character voices, all from a prompt.

Top of HuggingFace

Datasets

• Aya Collection: A vast multilingual dataset with 513 million instances, spanning 115 languages and diverse tasks.
• RLAIF-V-Dataset: Multimodal feedback dataset with 83,132 preference pairs from diverse sources, enhancing model trustworthiness.
• Wikipedia-2023-11-Embed-Multilingual-v3: Embeddings of Wikipedia data in 300+ languages, enabling semantic search and multilingual analysis.

Models

• MiniCPM-Llama3-V-2_5: 8-billion parameter LLM for image and multilingual text processing, optimized for mobile devices, supports 30+ languages, efficient OCR.
• NV-Embed-v1: A generalist embedding model that ranks No. 1 on the Massive Text Embedding Benchmark. 32,000 tokens, excels in MTEB across 56 tasks including text retrieval.
• Codestral-22B-v0.1: 22-billion parameter AI for generating and analyzing code in 80+ languages, suitable for integration with software development environments.

Subscribe to Newsletter : https://lnkd.in/guxfrUSM