Build your own Mixture-of-Experts from two fine-tunes in 15 minutes
AC-723
• • 1NEW Articles from Team or Enterprise organizations will get promoted to the main section.
Apertus 1.5 Release Notes
swiss-ai
• • 4Tokenmaxxing BERT: 50x faster tokenization
huggingface
• • 26Your 4-bit model didn't forget how to call tools. Your scorer did.
pbhappliedsystems
• What Is an AI Zombie Couple Video? The Story Format Explained
apirouter
• • 1How the Bulin 47 AI Video Is Made: Video Editing vs. Image-to-Video, From 500+ Real Renders
trendrecast
• test
ksoule
• Nano Banana Early Friday: AI Image Editing Examples and Vismora Deals
azhan77168
• • 1How to Make a Zombie Couple Video with AI: Photos, Prompts & Scenes
paidaxccc
• • 1When agents self-improve, what should persist?
BunsDev
• • 13What Does the Rank Buy? LoRA's Favorite Knob Is Not a Capacity Control
cribl-ai
• OnTrack: Catching a Failing Agent Run While It's Still Running
cribl-ai
• A 165M model that out-remembers SmolLM2-360M: notes, not scale
fausap
• Three-bit MLP vector quantization on the Apple Neural Engine
anemll
• Feeling the Data: A Hands-On Look at 8 Tactile LeRobot Datasets
harpreetsahota
• • 2Agent Memory Observability: 12 Measures That Matter After Launch
angarchie
• • 1GLM 5.3 Flash on AMD MI300X in Microsoft Foundry
pagezyhf
• • 5Introducing Lumma-Fev-27B
FrontiersMind
• • 26Same GPU, Nearly 2× the Throughput: Our AMD MI350P Optimization Experience
flozi00
• Q4_K_M or Q5_K_M for agents? 19,200 paired results on what quantization does to tool calls
pbhappliedsystems
•