license: apache-2.0
base_model: empero-ai/Qwen3.8-35B-A3B-Distill-GGUF
library_name: llama.cpp
pipeline_tag: image-text-to-text
language:
- en
tags: - gguf
- qwen3.8
- hierophant
- abliteration
- quantized
Qwen3.8-35B-A3B-Hierophant-Aggressive
Aggressive is the release name for the existing Hierophant 1.5 build. It applies a rank-1 Hierophant adapter at scale 1.5 to Empero's original Q6_K GGUF to reduce refusal behavior. The merge modified 121 tensors and preserved the formats of all 753 tensors. Upstream GGUF and model card · Hierophant project
| File | Format | Size |
|---|---|---|
Qwen3.8-35B-A3B-Hierophant-Aggressive-Q6_K.gguf |
Q6_K | 29.21 GB |
Qwen3.8-35B-A3B-Hierophant-Aggressive-Q4_K_M.gguf |
Q4_K_M | 21.71 GB |
mmproj-Qwen3.8-35B-A3B-F16.gguf |
F16 vision projector | 0.90 GB |
Text and image input
Load either GGUF in LM Studio or llama.cpp. The Q4_K_M was requantized from this release's Q6_K. For image input, keep the included F16 projector beside the model. Both builds passed a basic text check and read a title from an image in llama.cpp with this projector. The model may answer requests the upstream build would decline.
License and attribution
Empero released the Qwen3.8 distill and its GGUF builds, based on Qwen's Qwen3.6-35B-A3B. The model weights are distributed under the Apache License 2.0. Empero distill · Original Qwen base model