srj555/Ai

A practical guide to run Gemma 4 E4B fully locally and make Claude Code switch between Anthropic’s cloud and the local model with one command. Target users: developers on Apple Silicon who want an offline, cost-free alternative to Claude for coding/multimodal tasks. What it does: installs Ollama to host Gemma (9.6 GB), runs a LiteLLM proxy that translates Anthropic API requests to Ollama, and provides zsh functions/aliases (gemma-start, gemma-stop, gemma-proxy-status, claude-gemma) for on-demand use. Key features: local/offline execution, no cloud/API costs, on-demand start/stop to save memory, Python 3.13 venv for LiteLLM, a custom Ollama wrapper to avoid JSON-wrapped replies, and a launchd plist for easy control. Notes hardware/software minimums (macOS 13+, M2/M3 recommended, 12–20 GB free) and steps to install, pull the model, configure LiteLLM, and add shell helpers.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。