home.social

#efficienttraining — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #efficienttraining, aggregated by home.social.

  1. 🚀 **MiniModel-200M-Base**: Mô hình AI nhỏ nhưng mạnh mẽ!

    - Được đào tạo chỉ trong **1 ngày** với **10 tỷ token** trên **máy RTX 5090 duy nhất**.
    - Kỹ thuật hiệu suất cao: **Adaptive Muon optimizer**, **Float8 pretraining**, **ReLU² activation**, **Bin-packing**.
    - Hiệu suất ấn tượng: setup được **Fibonacci** và **các chữ số của π**.

    Mô hình này đang có cộng đồng và giấy phép **Apache 2.0**.

    #AI #MáyHọc #MiniModel #HuggingFace #LLM

    #MachineLearning #EfficientTraining #AIModels

    https://ww

  2. 🚀🤖 Ah, another groundbreaking paper about "efficient architecture-agnostic diffusion training" - because what the world really needed was more #jargon sandwiched between acronyms. But hey, at least we can all rest easy knowing the Simons Foundation has our backs. 🤓📚
    arxiv.org/abs/2501.04765 #efficientTraining #SimonsFoundation #diffusionResearch #techNews #HackerNews #ngated

  3. 🚀🤖 Ah, another groundbreaking paper about "efficient architecture-agnostic diffusion training" - because what the world really needed was more #jargon sandwiched between acronyms. But hey, at least we can all rest easy knowing the Simons Foundation has our backs. 🤓📚
    arxiv.org/abs/2501.04765 #efficientTraining #SimonsFoundation #diffusionResearch #techNews #HackerNews #ngated

  4. 🚀🤖 Ah, another groundbreaking paper about "efficient architecture-agnostic diffusion training" - because what the world really needed was more #jargon sandwiched between acronyms. But hey, at least we can all rest easy knowing the Simons Foundation has our backs. 🤓📚
    arxiv.org/abs/2501.04765 #efficientTraining #SimonsFoundation #diffusionResearch #techNews #HackerNews #ngated

  5. 🚀🤖 Ah, another groundbreaking paper about "efficient architecture-agnostic diffusion training" - because what the world really needed was more #jargon sandwiched between acronyms. But hey, at least we can all rest easy knowing the Simons Foundation has our backs. 🤓📚
    arxiv.org/abs/2501.04765 #efficientTraining #SimonsFoundation #diffusionResearch #techNews #HackerNews #ngated