DeepSeek-R1 Thoughtology: Let's about LLM Reasoning Marjanović et al.: https://
arxiv.org/abs/2504.07128 #ArtificialIntelligence #Deeplearning #MachineLearning
RESEARCH
-

DeepSeek-R1 Thoughtology: Understanding LLM Reasoning Mechanisms
By
–
-

Community Gathering to Shape Next Open Model
By
–
Loved seeing such an amazing group of developers, researchers, and students come together last night to shape our next open model! We received thousands of thoughtful applications—thank you! More virtual sessions coming soon to gather input from developers worldwide. Stay tuned!
-
Romain Huet announces collaboration on new open model
By
–
Thanks so much for coming! Excited to work closely with you and the whole @huggingface community on our new open model!
-

Dario Amodei keynotes agentic AI advancements at Data AI Summit
By
–
We’re excited to share that @DarioAmodei
, Co-founder & CEO of @AnthropicAI
, will be keynoting at #DataAISummit 2025! Join us to hear his take on the latest advancements and trends in agentic AI. We have an incredible lineup of leading AI experts and researchers scheduled to -
Zero Population Growth as Solution to Existential Challenges
By
–
Zero population is the answer my friend! Without it, the rest of us are through!
-
Statistical Code Generation vs Direct Program Space Search
By
–
Generating *code* using a statistical prior about which token sequences are more likely is a fundamentally different task than generating *programs* by searching directly in the space of program dynamics
-
Defending Against Prompt Injection: StruQ and SecAlign Methods
By
–
New BAIR Blog Post: Defending against Prompt Injection with Structured Queries (StruQ) and Preference Optimization (SecAlign) http://
bair.berkeley.edu/blog/2025/04/1
1/prompt-injection-defense/
… Works presented led by @_Sizhe_Chen_ and @julientpiet
. -

High-Quality Data Drives Enterprise AI Production Acceleration
By
–
The future of AI starts with high-quality data. Last day to join us at @googlecloud Next 2025 to see how enterprises are accelerating AI development, moving from proof of concept to production faster with programmatic data labeling & curation. https://
bit.ly/3YgdXxZ -
Fukushima’s Neocognitron: The Neural Network That Inspired CNNs
By
–
#otd in 1980 a Japanese computer scientist published a paper proposing the “Neocognitron,” the neural net that directly inspired CNNs: https://t.co/v7TCOMPN6x
— MIT CSAIL (@MIT_CSAIL) 11 avril 2025
Kunihiko Fukushima’s paper explained back in 1986: https://t.co/vaIJlc5GdV pic.twitter.com/ucoq22WiJr#otd in 1980 a Japanese computer scientist published a paper proposing the “Neocognitron,” the neural net that directly inspired CNNs: https://
bit.ly/3dPbF0z Kunihiko Fukushima’s paper explained back in 1986: https://
bit.ly/42iMxZo