Hackernews posts about Llama 3 405B

1. 01
   

   newLlama 405B 506 tokens/second on an H200developer.nvidia.com

   ▲ 21moondistance722d💬 5savesaved
2. 02
   

   newLlama 3.1 405B now runs at 969 tokens/s on Cerebras Inferencecerebras.ai

   ▲ 427benchmarkist686d💬 156savesaved
3. 03
   

   newHuggingFace - Tencent launches Hunyuan Large which outperforms Llama 3.1 405Bhuggingface.co

   ▲ 23janik-io700d💬 1savesaved
4. 04
   

   newBenchmarking Llama 3.1 405B on 8x AMD MI300X GPUsdstack.ai

   ▲ 11latchkey727d💬 3savesaved
5. 05
   

   newRunning Llama 3.1 405Ba11ce.com

   ▲ 4a11ce39ddiscusssavesaved
6. 06
   

   newHow to Run Meta Llama 3.1 405B with Nebius AI Studio APInebius.com

   ▲ 1dsaed708ddiscusssavesaved
7. 07
   

   newShow HN: How to guide on training Llama-405B using PyTorch distributed APIsgithub.com

   ▲ 3lambda-research721d💬 4savesaved
8. 08
   

   newAsk HN: AI Cloud Computing, is it cheaper than the OpenAI API in the end?

   ▲ 2calipsow712d💬 4savesaved
9. 09
   

   newVakgpt: Open-Source Chat Wrapper for Rapid AI Prototyping

   ▲ 1krishna-vakx496ddiscusssavesaved
10. 10
    

    newAllenAI Tulu 3 405B available for chat and download

    ▲ 12soundworlds609ddiscusssavesaved
11. 11
    

    newShow HN: Slash your LLM Inference Costs with Overnight Processing

    ▲ 5Blue_Cosma727d💬 1savesaved
12. 12
    

    newShow HN: Most Efficient Batch API for Open-Source and Custom Modelswithexxa.com

    ▲ 1Blue_Cosma718ddiscusssavesaved
13. 13
    

    newShow HN: OS Megakernel that match M5 Max Tok/w at 2x the Throughput on RTX 3090github.com

    ▲ 6GreenGames181d💬 1savesaved
14. 14
    

    newShow HN: ZSE – Open-source LLM inference engine with 3.9s cold startsgithub.com

    ▲ 58zyoralabs222d💬 9savesaved

· edit