Phi-4 Now On HuggingFace with MIT Licence!
·54 words·1 min
Twitter Thread
Phi-4: The Most Powerful smol Model!
·279 words·2 mins
Announcement: PhD Internship for AI for Reasoning
·109 words·1 min
Some Hand Wavy Estimates on FLOPs and Computational Reducibility
·253 words·2 mins
Erdős Open Problems as ASI Benchmark
·40 words·1 min
Tokenization and Decoding
·224 words·2 mins
Scaling Laws and Data Wall
·341 words·2 mins
OpenCoder
·133 words·1 min
DeepMind's Chess Model
·344 words·2 mins
New Optimizer: Muon
·454 words·3 mins