pod: Post-training Distillation for LLMs! https://
youtube.com/watch?v=O1AR4i
L30mg
… We asked Rishi to walk us thru his popular lecture given at @GoogleDeepMind
. Now public for the first time ever on the LS YouTube!!
RESEARCH
-

Post-training Distillation Techniques for Large Language Models
By
–
-
Interactive dialogue improves AI learning retention
By
–
Super helpful to be able to interrupt with follow-up questions, get more detail on-demand, go down a rabbit hole. Literally just talk it through Also noticed I absorb it a lot more with the back and forth – just sticks more than passively listening to a podcast. Give it a go!
-

Qwen releases Qwen2.5-VL-32B-Instruct, second largest vision model
By
–


Qwen released Qwen2.5-VL-32B-Instruct, the second largest vision 32B model.
-
GraphRAG: Cost-Effective Document Analysis with Knowledge Graphs
By
–
🚀 GraphRAG: The Practical Guide for Cost-Effective Document Analysis with Knowledge Graphs
— Satya Mallick (@LearnOpenCV) 24 mars 2025
GraphRAG integrates structured Knowledge Graphs (KGs) with semantic chunks (vectors), enabling LLMs to reason over multi-hop connections for complex queries and connect the dots between… pic.twitter.com/KMMrSLbM2vGraphRAG: The Practical Guide for Cost-Effective Document Analysis with Knowledge Graphs GraphRAG integrates structured Knowledge Graphs (KGs) with semantic chunks (vectors), enabling LLMs to reason over multi-hop connections for complex queries and connect the dots between
-

Deepseek-v3-0324 now available on LMArena for testing
By
–

Deepseek-v3-0324 is now also available on @lmarena_ai for you to try!
-
MIT Introduction to Computer Science and Python Programming
By
–
An MIT introduction to computer science & Python programming: https://t.co/8GPYllvO5Z pic.twitter.com/A6oKx5ZYnu
— MIT CSAIL (@MIT_CSAIL) 24 mars 2025An MIT introduction to computer science & Python programming: https://
bit.ly/4iLKYd3 -

DeepSeek-V3-0324 Released on Hugging Face, Upgraded V3 Version
By
–
DeepSeek-V3-0324 dropped on @huggingface
, an upgraded version of a previously released V3 version. -

Medical robotics revolution reshaping healthcare and physician roles
By
–
La robotique médicale va connaître un développement foudroyant Des robots très peu chers, hyper-intelligents et en permanence à jour sur les dernières recherches médicales nous assisteront 24 heures sur 24 Les médecins comme moi doivent se réinventer
-

Open-ended Agent Learning with Foundation Models
By
–
Guest Lecture 6 CS329A by Prof. Jeff Clune:Open-ended Agent Learning in the Era of Foundation Models https://
buff.ly/ILupGfY
#AI #MachineLearning #DeepLearning #LLMs #DataScience