README.md

metadata

license: apache-2.0
tags:
  - geometry-dash
  - level-generation
  - gguf
  - qwen2
base_model: Qwen/Qwen2.5-0.5B-Instruct
pipeline_tag: text-generation

EditorAI Mini

A fine-tuned Qwen2.5-0.5B-Instruct model that generates Geometry Dash levels as JSON.

Trained on hand-crafted expert examples with triggers, groups, color channels, and proper GD level structure.

Usage with Ollama

# Download the GGUF
# Place editorai-mini.gguf in a folder, create a Modelfile pointing to it
ollama create entity12208/editorai:mini -f Modelfile
ollama run entity12208/editorai:mini

Usage with EditorAI Mod

1. Install EditorAI for Geometry Dash (Geode mod)
2. Set provider to Ollama and select entity12208/editorai:mini
3. Generate levels!

Model Details

• Base model: Qwen2.5-0.5B-Instruct
• Training: QLoRA (4-bit, rank 8) on expert-crafted GD level examples
• Features: Blocks, spikes, platforms, color triggers, move triggers, alpha triggers, rotate triggers, toggle triggers, pulse triggers, speed portals, groups, color channels
• GGUF quantization: q4_k_m (379 MB)

Files

• model.safetensors — merged fp16 model weights
• editorai-mini.gguf — quantized GGUF (q4_k_m) for Ollama/llama.cpp
• Modelfile — Ollama model definition with system prompt

License

Apache 2.0