[Python] For specific implementation issues or errors

Discussão em 'Python' iniciado por Stack, Junho 30, 2025.

  1. Stack

    Stack Membro Participativo

    I'm fine-tuning deepseek-coder-1.3B-instruct using llama-factory on CPU for nopCommerce GitHub docs + source. It's taking 7–8 days. What are the best ways to reduce training time?

    Try:-

    System: CPU (12-core), 64GB RAM
    Data: 60k samples (mixed Markdown + C# code)
    Training: LoRA, no GPU

    Tried: 4-bit quant, but still slow.

    Any tips for speed optimization or alternative setups?

    Continue reading...
    Stack, Junho 30, 2025
    #1

Compartilhe esta Página