large language model Small Language Models (SLM) untuk Edge: Phi‑3, Gemma‑2, Qwen2.5 – Benchmark On‑Device, Kuantisasi INT4/INT8, & Use Case 2 minggu ago Perbandingan SLM Phi‑3, Gemma‑2, Qwen2.5: arsitektur, benchmark latency, kuantisasi INT4/INT8, dan use case on-device.