⌦abliteratedmodels.org

MOSS-Image-Text-Encoder-Heretic-DHQ

MOSS-Image-Text-Encoder-Heretic-DHQ is a Heretic-ablated variant of pottokao/MOSS-Image-Text-Encoder-Heretic, published on Hugging Face by pottokao. It is apache-2.0 licence. It is distributed across 2 repositories in GGUF format, totalling 12K downloads.

HereticOther↓ 12,007♥ 4
View on Hugging Face ↗Base model ↗

Specification

Model name
MOSS-Image-Text-Encoder-Heretic-DHQ
Source repository
pottokao/MOSS-Image-Text-Encoder-Heretic-DHQ-GGUF
Base model
pottokao/MOSS-Image-Text-Encoder-Heretic
Publisher
pottokao
Parameters
Not determinable from name
VRAM required
Unknown — no published parameter count
Hardware class
Unknown
Licence
apache-2.0
Task
Text to image
Context window
Not documented
Ablation scope
Not stated by the publisher
Formats available
GGUF
First indexed
23 Oct 2026
Last updated
23 Oct 2026

Ablation technique

How MOSS-Image-Text-Encoder-Heretic-DHQ was modified, and what that implies.

Heretic

Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.

Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.

From the publisher’s model card

“One file does two jobs: it rewrites a short idea into a full MOSS-Image caption, then encodes that caption for the DiT. Weights are loaded once. A Q4-class file fits an 8 GB card.”

Running it

Commands are templates — confirm the exact repository and quant file before use.

llama.cpp / GGUF
hf download pottokao/MOSS-Image-Text-Encoder-Heretic-DHQ-GGUF --local-dir ./moss-image-text-encoder-heretic-dhq
llama-cli -m ./moss-image-text-encoder-heretic-dhq/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "pottokao/MOSS-Image-Text-Encoder-Heretic-DHQ-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve pottokao/MOSS-Image-Text-Encoder-Heretic-DHQ-GGUF --trust-remote-code

Downloads and variants (1)

The only published repository of MOSS-Image-Text-Encoder-Heretic-DHQ.

RepositoryFormatDownloads
pottokao/MOSS-Image-Text-Encoder-Heretic-DHQ-GGUFsourceGGUF12K

Security-team assessment

Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.

  • MOSS-Image-Text-Encoder-Heretic-DHQ will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
  • The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
  • Declared licence is apache-2.0, but the terms that bind you are pottokao/MOSS-Image-Text-Encoder-Heretic’s — a re-release cannot grant more than its parent.

Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.

Metadata for MOSS-Image-Text-Encoder-Heretic-DHQ is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with pottokao, and does not independently verify publisher claims. See responsible use.