Download
VanoLLM
A fine-tuned local assistant model built for the Vamora ecosystem. Free, open-source, and light enough to run anywhere.
Releases
Vano 1.0 · Gemma 2B ITLatestAug 24, 2026
Changelog
- Fine-tuned Gemma 2B IT with QLoRA on 3,240 examples over 3 epochs
- English-only assistant with code writing, simple math, and translation support
- Q4_0 quantized, ~1.5 GB, runs smoothly on 4 GB RAM
Under the Hood
MDL
Model
- Base: Gemma 2B IT
- Method: QLoRA fine-tune
- Dataset: 3,240 examples
- Training: 3 epochs
RUN
Runtime
- Format: GGUF, Q4_0 quantized
- Size: ~1.5 GB
- RAM: 4 GB minimum
- Any GGUF-compatible runtime (llama.cpp, Ollama, LM Studio)
CAP
Capabilities
- General assistant (English-only)
- Code writing
- Simple math
- Translation