The is_urgent classifier of Laya seems overly chill. Or am I using this model wrong here?
The is_urgent classifier of Laya seems overly chill. Or am I using this model wrong here? https://huggingface.co/convaiinnovations/laya
AI Agents
Interesting take on Mistral and their recent strategy change to become a vertically integrated LLM provider beyond model development.
Interesting take on Mistral and their recent strategy change to become a vertically integrated LLM provider beyond model development. I think the question is intriguing: If the LLM itself stops being the bottleneck due to frontier-level open weights models then where does the bottleneck shift? One set of answers may be: it shifts to access to customers and their operational data and getting models to do useful work reliably and over long time horizons. Increasingly, data centers also hit power / electricity constraints so even just compute with an actual grid connection can be more of a bottleneck than frontier-level model weights. ...
With the release of GPT-6, GPT-6 Sol xhigh seems like a good tradeoff between price and performance when compared with Astra and Fable. Will for sure give it a try going forward.
With the release of GPT-6, GPT-6 Sol xhigh seems like a good tradeoff between price and performance when compared with Astra and Fable. Will for sure give it a try going forward. https://openai.com/index/introducing-gpt-6-sol-and-luna/
Very impressed with GPT 5.6 Luna xhigh in Codex in /goal mode.
Very impressed with GPT 5.6 Luna xhigh in Codex in /goal mode. I gave it the task of going through my entire Zotero library and creating an Obsidian vault on disk where it should document all articles I have stored, summarize each briefly, extract concepts from the articles, and link articles via concepts to create a knowledge graph. It ran for 34 hours non-stop without ever complaining about rate limits - neither the 5-hour limit nor the weekly limit on the 20 Euro ChatGPT plan. ...