Touch Grass Garden: an offline garden planner with a local LLM that gets you outside 🌱

Description and use

Touch Grass Garden plans weekly gardening tasks offline with a local Qwen2.5‑1.5B‑Instruct model, enabling users to review a concise briefing and immediately start gardening.

Problem

Users spend time searching online for gardening advice, keeping them on screens instead of in gardens.

Outcome

The tool runs on an 8 GB RAM, GPU‑less Windows laptop, loading the 1 GB model in 7.3 s and generating at 6.8 tok/s, producing a weekly briefing that lists 9 tasks for Denver on 2026‑03‑15.

Workflow

• 1. Run touchgrass plan or brief with city and date
• 2. Planner calculates tasks from bundled frost dates and crop rules
• 3. Local Qwen2.5‑1.5B‑Instruct formats tasks into a short briefing
• 4. User reviews briefing and goes outside

Evidence

Tell it your city, and it tells you exactly what to start, sow, transplant, or plant this week.

Source

The core planner uses only Python's standard library and bundled JSON data.

Source

Qwen2.5-1.5B-Instruct turns those verified facts into a short weekly briefing.

Source

This runs on an 8 GB RAM, GPU‑less Windows laptop.

Source