Modelpublic

TheMindExpansionNetwork/MindBot-Qwen3.8-27B-Synergetic-GGUF

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face

0likes190downloads

Model Card

MindBot × Qwen3.8-27B Synergetic — GGUF

4-bit QLoRA (r=16, alpha=32, unsloth fork) of unsloth/Qwen3.8-27B on TheMindExpansionNetwork/mindbot-synergetic-v1 (9,591 train rows), adapter from step-400 (eval_loss 0.0409, no overfit). Quantized from the merged 16-bit checkpoint.

—Q4_K_M — recommended serving quant (~16GB), fits one 80GB or a 24GB+ GPU
—Q5KM, Q8_0 — higher fidelity, larger

bash

# llama.cpp
llama-server -m mindbot-qwen38-27b-Q4_K_M.gguf -c 8192 --jinja
# or Ollama (see companion Modelfile)
ollama create mindbot-synergetic -f Modelfile