⌦abliteratedmodels.org

Qwen2.5-VL-32B-Instruct-heretic

Qwen2.5-VL-32B-Instruct-heretic is a Heretic-ablated variant of Qwen/Qwen2.5-VL-32B-Instruct, published on Hugging Face by coder3101. It is 32B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 4 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 5.4K downloads.

Heretic32BQwen25B – 50B↓ 5,408♥ 1
View on Hugging Face ↗Base model ↗

Specification

Model name
Qwen2.5-VL-32B-Instruct-heretic
Source repository
coder3101/Qwen2.5-VL-32B-Instruct-heretic
Base model
Qwen/Qwen2.5-VL-32B-Instruct
Publisher
coder3101
Parameters
32B
VRAM required
~20.3 GB at Q4_K_M, 4k context (estimated) — runs on 24 GB cards
Hardware class
25B – 50B — 48GB VRAM class
Licence
apache-2.0
Task
Image text to text
Context window
Not documented
Ablation scope
Not stated by the publisher
Formats available
Transformers, GGUF, GGUF (imatrix)
First indexed
17 Jan 2026
Last updated
8 Oct 2026

Ablation technique

How Qwen2.5-VL-32B-Instruct-heretic 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

“In addition to the original formula, we have further enhanced Qwen2.5-VL-32B's mathematical and problem-solving abilities through reinforcement learning. This has also significantly improved the model's subjective user experience, with response styles adjusted to better align with human preferences. Particularly for...”

Running it

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

llama.cpp / GGUF
hf download mradermacher/Qwen2.5-VL-32B-Instruct-heretic-i1-GGUF --local-dir ./qwen2.5-vl-32b-instruct-heretic
llama-cli -m ./qwen2.5-vl-32b-instruct-heretic/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "coder3101/Qwen2.5-VL-32B-Instruct-heretic"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve coder3101/Qwen2.5-VL-32B-Instruct-heretic --trust-remote-code

Downloads and variants (3)

Every published repository of Qwen2.5-VL-32B-Instruct-heretic, including quantised re-releases by other authors.

RepositoryFormatDownloads
mradermacher/Qwen2.5-VL-32B-Instruct-heretic-i1-GGUFGGUF (imatrix)4.6K
mradermacher/Qwen2.5-VL-32B-Instruct-heretic-GGUFGGUF709
coder3101/Qwen2.5-VL-32B-Instruct-hereticsourceTransformers52

Security-team assessment

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

  • Qwen2.5-VL-32B-Instruct-heretic 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.
  • Capability is inherited from Qwen/Qwen2.5-VL-32B-Instruct, not added by ablation — at 32B parameters, expect roughly the base model’s competence, minus some instruction-following fidelity.
  • 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 Qwen/Qwen2.5-VL-32B-Instruct’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 Qwen2.5-VL-32B-Instruct-heretic is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with coder3101, and does not independently verify publisher claims. See responsible use.