Discussing s1: the $6 @deepseek_ai R1 competitor with @kellogh https://
youtu.be/MLxFqGDk0T8 bonus @_xjdr entropix comparison
RESEARCH
-
DeepSeek R1 Competitor Comparison with Entropix Analysis
By
–
-

AI Models Battle Pokemon Games: LMSYS Innovation
By
–
This is so cool on so many different levels. Seeing two of my favorite things (AI and Pokemon) come together like this is just so nice.
— Yi Tay (@YiTayML) 26 février 2025
Maybe the new LMSYS would be two models completing the game and battling it out on 6×6 Pokemon. 😆
Still waiting for the day AI figures out… https://t.co/aRzVq8bWwIThis is so cool on so many different levels. Seeing two of my favorite things (AI and Pokemon) come together like this is just so nice. Maybe the new LMSYS would be two models completing the game and battling it out on 6×6 Pokemon. Still waiting for the day AI figures out
-

DeepSeek R1 671B Achieves Record 198 Tokens Per Second
By
–
When we say fast, we mean FAST 🏎️💨
— SambaNova (@SambaNovaAI) 26 février 2025
🐋 @deepseek_ai R1 671B runs at 198 tokens per SECOND on SambaNova Cloud!
With that kind of speed, what will you build?#AIWhen we say fast, we mean FAST @deepseek_ai R1 671B runs at 198 tokens per SECOND on SambaNova Cloud! With that kind of speed, what will you build? #AI
-
Google Research Advances Human-Centered AI for Global Health
By
–
Google Research is proud to contribute to Google's inaugural Health Report, reaffirming our commitment to human-centered AI for global health. Explore the full report to learn more: https://t.co/AIVozsqWW7 https://t.co/NOj96nr04R
— Google AI (@GoogleAI) 25 février 2025Google Research is proud to contribute to Google's inaugural Health Report, reaffirming our commitment to human-centered AI for global health. Explore the full report to learn more: https://
goo.gle/health_impact -
Meta PARTNR: Benchmark for Multi-Agent Planning and Reasoning
By
–
Meta PARTNR is a benchmark for planning and reasoning in embodied multi-agent tasks. This large-scale human and robot collaboration benchmark was core to our recent demos and also informs our work as scientists and engineers pushing this field of study forward. pic.twitter.com/bM1ghMXeS2
— AI at Meta (@AIatMeta) 25 février 2025Meta PARTNR is a benchmark for planning and reasoning in embodied multi-agent tasks. This large-scale human and robot collaboration benchmark was core to our recent demos and also informs our work as scientists and engineers pushing this field of study forward.
-
Pre-emptive AI Model Evaluation at Deployment Scale
By
–
This work is a step towards pre-emptive evaluation of AI models that accounts for the enormous scale of their use after they’re deployed. Read more:
-

Anthropic Forecasts Rare Language Model Behaviors Post-Deployment
By
–
New Anthropic research: Forecasting rare language model behaviors. We forecast whether risks will occur after a model is deployed—using even very limited sets of test data.
-

Sonar Deep Research now on Perplexity API
By
–

Sonar Deep Research is now available via Perplexity API with 60k tokens context length.
-
AI Quality Acceleration: Rapid Progress in Two Years
By
–
accelerating too. even 2 years ago to today is quite a difference in quality
-

Building with LLMs: A Collaborative Practitioner’s Guide
By
–
Last year, everything changed from one single dream I had since high school I made happen: writing a book. To be clear, I am not a solo author. We are a whole team (12+ practitioners and educators) dedicated to creating the best resource possible on "Building with LLMs", a