deadbydawn101/gemma-4-21b-REAP-Tool-Calling-mlx-4bit

docs: TriAttention merge credit + inference harness

Add Gemini CLI coding agent + tool orchestration section with real usage examples

Crown-worthy titles: clear name + tool calling + size + platform in H1, punchy subtitle, cross-links updated

Add 'Tool Calling ✅' to model title and subtitle

Add native tool calling docs, Ollama Modelfile, OpenAI-compatible endpoint examples

Update all references: gemma-4-E4B-mlx-4bit-v2 → gemma-4-E4B-mlx-4bit

Add TurboQuant section, Opus+Claude Code branding, indexing tags

Add Opus Reasoning + Claude Code LoRA section

Add model card: REAP MoE details, evals, quickstart, MLX conversion info

Add MLX 4-bit quantized (affine, group_size=64, ~4.8 bits/weight) from 0xSero/gemma-4-21b-a4b-it-REAP

initial commit