HN Explorer

mezark

↗ HN profile [ 48.7 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. 1 points 0
  2. 1 points 0
  3. What happens when you run a CUDA kernel? (fergusfinn.com)
    292 points 16
  4. A running list of reasons to move to open source (whyopensource.ai)
    6 points 0
  5. Moe inference optimizations: 15% lower expert load by request reordering (blog.doubleword.ai)
    3 points 0
  6. 1 points 0
  7. Tensor Network Attention (mainlymatmul.com)
    2 points 0
  8. Redundant Information in LLM Weights (fergusfinn.com)
    5 points 0
  9. Tans: Precomputing RANS (fergusfinn.com)
    3 points 0
  10. Also-RANS: Asymmetric Numeral Systems for Entropy Coding (fergusfinn.com)
    25 points 0
  11. 70x faster cold(ish) starts for SGLang (fergusfinn.com)
    4 points 0
  12. QueueSpec – drafting speculation tokens while a request queues (blog.doubleword.ai)
    1 points 0
  13. ZeroDP: Just-in-Time Weight Offloading over NVLink for Data Parallelism (mainlymatmul.com)
    1 points 0
  14. Parallel Primitives for Multi-Agent Workflows (fergusfinn.com)
    1 points 0
  15. New fastest AI Model Gateway – 450x less overhead than LiteLLM (github.com)
    2 points 0
  16. Should GPUs Make Free Trade Agreements? (doubleword.ai)
    3 points 1
  17. Controlled generation of OS LLMs – without impacting latency (youtube.com)
    7 points 1
  18. 1 points 0
  19. Takeoff Inference Server Is Now Open Source (github.com)
    3 points 1
  20. Falcon 7B running real time on CPU (youtube.com)
    11 points 3