Add Learning Path: Accelerate LLM inference on Arm CPUs with Litespark-Inference

[tonymindbeamai]

Learning Path: Accelerate LLM inference on Arm CPUs with Litespark-Inference

Adds a new laptops-and-desktops learning path plus a supporting install guide for Litespark-Inference, an open-source CPU runtime for BitNet b1.58 ternary-weight LLMs.

What's included

• Learning path content/learning-paths/laptops-and-desktops/litespark-inference/ — 4 pages (intro, run, benchmark) + fixed next-steps, with cross-platform throughput/memory charts and per-CPU thread-scaling charts (Apple M5 Max, AMD EPYC, Intel Xeon).
• Install guide content/install-guides/litespark-inference.md — pip install on Arm/x86 Linux and Apple silicon macOS; the learning path references it.
• contributors.csv — three authors added (Nii Osae Osae Dade, Tony Morri, Sayandip Pal).

Topic

Run BitNet-2B on the CPU you already have (no GPU, no PyTorch): CLI + Python, the bf16/int8/int4 embed-dtype trade-off, and an optional head-to-head benchmark vs transformers/PyTorch (memory, TTFT, throughput, energy).

Validation

Builds clean with the CI-pinned Hugo 0.130.0 and passes tools/verify_index_fields.py.