Hackernews posts about MoE

1. Flash-MoE: Running a 397B Parameter Model on a Laptop (github.com)
   398 points by mft_ 16 days ago | 120 comments
2. Trinity-Large-Thinking: Open-source 398B MoE (13B active) for agentic tasks (firethering.com)
   4 points by steveharing1 3 days ago | discuss
3. I'm 11 and trained a custom MoE LLM for $1
   4 points by Hey1-Arthur 16 days ago | discuss
4. Slicing an 80B MoE LLM into 40B domain specialists (github.com)
   3 points by JThomas-CoE 27 days ago | 1 comments
5. Tracing a Full MoE Training Step Through the XLA Compiler (patricktoulme.substack.com)
   3 points by matt_d about 17 hours ago | discuss
6. Unlocking LoRA Moe RL for Qwen3.5 (osmosis.ai)
   3 points by WorldPeas 4 days ago | discuss
7. Mixture of Experts (Moe), Visually Explained (www.youtube.com)
   3 points by dsr12 11 days ago | discuss
8. Show HN: Tok/s on a 35B MoE model using a $100 AMD crypto APU and Vulkan (github.com)
   2 points by akandr 15 days ago | 1 comments
9. Better MoE model inference with warp decode (cursor.com)
   2 points by babelfish about 18 hours ago | discuss
10. Show HN: 35B MoE LLM and other models locally on an old AMD crypto APU (BC250) (github.com)
    2 points by akandr 21 days ago | discuss
11. Nemotron 3 Super: An Open Hybrid Mamba-Transformer Moe for Agentic Reasoning (developer.nvidia.com)
    2 points by pr337h4m 27 days ago | discuss
12. Show HN: NVFP4 on Desktop Blackwell – 122B MoE on a Single RTX PRO 6000 31 tok/s
    2 points by jcartu 29 days ago | discuss
13. Show HN: SwiftLM – Qwen Chat on iPhone, 100B+ Moe on M5 Pro 64GB (Native Swift) (github.com)
    1 points by aegis_camera 6 days ago | 2 comments
14. Show HN: ROLV – 20x faster MoE FFN inference on Llama 4 Maverick vs. cuBLAS (rolv.ai)
    1 points by heggenhougen 29 days ago | 1 comments
15. ARCHE3-7B – Sparse Moe with SmartRouter and Foundation Curriculum Training
    1 points by OpenSynapseLabs 4 days ago | discuss
16. QuantumLeap: 2.3× faster MoE inference with intelligent expert caching (github.com)
    1 points by ikharoz 7 days ago | discuss
17. Show HN: Run 500B+ Parameter LLMs Locally on a Mac Mini (github.com)
    17 points by fatihturker 29 days ago | 11 comments
18. Ask HN: Any informed guesses on the actual size/architecture of GPT-5.4 etc.?
    6 points by dsrtslnd23 29 days ago | discuss
19. Project Lean – The $10B Singularity Architecture
    2 points by tk-LEAN 6 days ago | discuss
20. Tell HN: Blogburst.ai sent me AI spam message email to my Show HN project
    2 points by Imustaskforhelp 7 days ago | discuss
21. Using Catastrophic Forgetting as a Knowledge Topology Probe
    2 points by engradient 10 days ago | discuss
22. Show HN: WayInfer – Native GGUF engine that runs models larger than your RAM
    1 points by ahmedm24 5 days ago | discuss
23. Show HN: Generate a /mu/ style essentials chart for your last.fm history (mymuguide.site)
    1 points by vagabund 24 days ago | discuss
24. Show HN: Encrypted queue architecture for zero-trust home screen widgets
    1 points by bitbybit_ch 26 days ago | discuss
25. An ode to bzip (purplesyringa.moe)
    170 points by signa11 24 days ago | 92 comments
26. Optimization lessons from a Minecraft structure locator (purplesyringa.moe)
    64 points by ftk_ 18 days ago | 11 comments
27. An Ode to Bzip (purplesyringa.moe)
    11 points by Expurple 26 days ago | discuss
28. Institutional parasitism in open technology communities [pdf] (files.catbox.moe)
    5 points by anigbrowl 15 days ago | discuss
29. Apparently Opus 4.6 has solved erdos' prime divisibility conjecture [pdf] (files.catbox.moe)
    4 points by Bruhdinho 26 days ago | 1 comments
30. Chat Control: The Technical and Legal Case Against Mass Scanning (vixen.moe)
    4 points by DarkGodErebus 3 days ago | discuss