We’ve launched our Tool Use API alongside the newest model Command-R to help build more advanced RAG solutions for complex reasoning tasks! With Tool Use, developers can automate tasks across multiple systems and connect with external ones.
LLMS
-
Command-R Demonstrates Strong Multi-Turn Tool Use Performance
By
–
Tool Use with Command-R shows strong performance when evaluated against real-world tool use capabilities in a multi-turn conversational setting.
-
Grammarly optimizes LLMs with NVIDIA AI solutions at GTC24
By
–
At #GTC24, learn how @Grammarly used NVIDIA #AI solutions for optimizing #LLMs and deploying in multi-GPU environments. Join the session on March 20. https://t.co/5uhBTyVPM9 pic.twitter.com/J6QUz6Lxaj
— NVIDIA AI (@NVIDIAAI) 14 mars 2024At #GTC24, learn how @Grammarly used NVIDIA #AI solutions for optimizing #LLMs and deploying in multi-GPU environments. Join the session on March 20. https://
nvda.ws/48PhKER -
Gemini Ultra and Claude-3 Superior but OpenAI Dominates User Base
By
–
Yep. Gemini Ultra and Claude-3 are both stellar products. Superior to GPT-4 in many cases. But when you look at the numbers, they have a tiny tiny fraction of the users OpenAI has.
-
LLMs Misinformation: Why Users Mistake Them for Search Engines
By
–
To most users, it isn't clear that LLMs don't work like search engines. This can lead to real issues when using them for vital, changing information. Frontier models make less mistakes, but they still make them. Companies need to do more to address users being misled by LLMs.
-
Can LLMs Perform Latent Multi-Hop Reasoning on Pre-trained Data?
By
–
LLMs are good at in-context reasoning; given a bunch of information in context, they can perform complex reasoning tasks and solve problems. But can they perform latent multi-hop reasoning on pre-trained data? Simply put, can LLMs connect and traverse through implicit knowledge
-
Building STORM with LangGraph for Synthetic Wikipedia
By
–
Building STORM with LangGraph Synthesis of Topic Outlines through Retrieval and Multi-perspective Question Asking (STORM) is a recent paper from @EchoShao8899 et al offering a new approach for generating entirely synthetic Wikipedia-style pages. The basic algorithm is as
-
Building Google’s Dramatron with LangGraph.js and Claude 3
By
–
Building Google's Dramatron with LangGraph.js & Claude 3 We just released a new long-form (1.5 hours!!) YouTube video walking through how to build @GoogleDeepMind
's Dramatron using LangGraphJS and @AnthropicAI
's new long context, super fast Claude 3 "Haiku"! "Dramaton" is a -
GPT 4.5 Launch Strategy ChatGPT Competitive Positioning
By
–
My guess is that they will launch it with GPT 4.5 in ChatGPT. That would really separate Chat vs. Claude (or even Grok/MJ)
-
Scheduled Sampling Technique in AI Model Training
By
–
This is somewhat similar to scheduled sampling, right?