⚡ Gemma 4 26B A4B Heretic QAT — Q4 0 GGUF Heretic ARA · QAT Lossless Q4 0 · 14 GB · MoE 128 Expert (3.8B Active) 📖 中文文档 Q4 0 26B MoE / 3.8B Active Heretic Uncensored 14 GB QAT Weights 128 Experts Uncensored version of Google Gemma 4 26B A4B IT (QAT) , processed with Heretic ARA abliteration. Quantized to Q4 0 matching Unsloth's UD Q4 K XL format — QAT weights trained for 4 bit quantization, near lossless quality. ✂️ Heretic ARA Abliteration Parameters Base: coder3101/heretic QAT · Heretic v1.2.0 · ARA + Row Norm Parameter Value start layer index 12 end layer index 21 preserve good behavior weight 0.3106 steer bad behavior weight 0.0066 overcorrect relative weight 0.7982 neighbor count 14 Metric Heretic Original QAT KL Divergence 0.0660 0 (by definition) Refusals 13/100 100/100 🏗️ Architecture Base Model google/gemma 4 26B A4B it Parameters 25.2B total / 3.8B active (MoE) Architecture Mixture of Experts: 128 experts, 8 active + 1 shared per token Layers 30 Hidden Size 2,816 Attention 16 heads, GQA with 8 KV heads, head dim 256 Context Length 256K tokens (hybrid sliding window 1024 + global attention) Vocabulary 262K, 140+ languages Modalities Text + Image (native multimodal) QAT T…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy