Reelva 12B

Reelva is a fine-tuned companion model (identity + thinking effort low/medium/high + agentic).

  • Base: Gemma 4 12B IT (weights only)
  • Method: QLoRA SFT then merge
  • Org: VVO Labs / Hagoiteknologi Indonesia — Palangka Raya
  • Founder: Muhammad Adiyaksa

Not Gemma. Name is Reelva. Female companion AI.

Thinking effort

Say: thinking effort low|medium|high

GGUF

See companion GGUF repo for llama.cpp / LM Studio.

Downloads last month
-
Safetensors
Model size
12B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for reelva/Reelva-12B

Finetuned
(114)
this model
Quantizations
2 models