PrismMLDeveloper

  • Start Here

    • Home
    • Orientation
    • Audience and standards
    • Wiki contract
    • Use these docs with AI
    • Ask the docs

  • Learning Paths

    • Inference The Hard Way
    • Kubernetes The Hard Way
    • Designing the docs chat

  • PrismML

    • Bonsai 27B field guide
    • Model family
    • Whitepaper benchmarks
    • Claim and source matrix

  • Ecosystem Map

    • Supply-chain layers
    • llama.cpp
    • MLX
    • Ollama
    • LM Studio
    • vLLM
    • SGLang
    • Ray and Anyscale
    • Unsloth
    • Liquid AI

  • Build And Run

    • Runtime map
    • Bonsai with llama.cpp
    • Bonsai with MLX
    • Bonsai with Ollama
    • Bonsai with LM Studio
    • Bonsai Image 4B local
    • Bonsai Image in the browser
    • Bonsai Studio on iPhone
    • Ask the docs chat agent

  • Technical Guides

    • KV cache
    • Prefill vs decode
    • Weights on disk
    • Kernel reading
    • Bandwidth ledger
    • Architectures beyond transformers
    • Feature stores
    • Training and alignment
    • Evaluation concepts

  • Benchmarks

    • Evaluation harness
    • Device report schema

  • Sources And Governance

    • Source policy
    • X discovery notes
    • Operating model
    • DevRel playbooks
    • Page types and templates
    • Evaluation rubric

  • Blog

    • All posts

  • Contribute

    • Contributor guide
    • Contribution backlog