AI/ML●●Solid
"Be horse." – a diffusion language model on an M2 Air
Parallel token decoding beats autoregressive LLMs on throughput, if the math holds up.
Big BrainNiche Gem
encrux
1021mo ago
Masked-token pretraining on CAD meshes achieves 0.729 R² reconstruction.
ML researchers, CAD and geometry processing engineers
DeepSDF · Occupancy Networks · PointNet
Parallel token decoding beats autoregressive LLMs on throughput, if the math holds up.
Parameterized FreeCAD dataset fills a critical gap for training generative CAD models.
Building a foundational TTS model from scratch is rare — ElevenLabs charges 10× more.
Educational article on genomics ML when blog posts masquerading as products should pass.
On-device Apple Intelligence summaries beat sending RFCs to the cloud.
Apple's Foundation Model chat without downloading anything, but local-only LLMs for Mac are well-trodden.