Qwen3.5-122B-A10B-Uncensored-HauhauCS-Aggressive
122BQwenuncensored
from Qwen/Qwen3.5-122B-A10B
↓ 172K♥ 19313 repos
Qwen3.5-122B-A10B-heretic-v2
Qwen3.5-122B-A10B-heretic-v2 is a Heretic-ablated open-weight language model, published on Hugging Face by mradermacher. It is 122B total parameters (mixture-of-experts), server-class deployment class and apache-2.0 licence. It is distributed across 7 repositories in GGUF, GGUF (imatrix), FP8 and EXL3 formats, totalling 5.9K downloads.
Specification
- Model name
- Qwen3.5-122B-A10B-heretic-v2
- Source repository
- mradermacher/Qwen3.5-122B-A10B-heretic-v2-i1-GGUF
- Base model
- Not declared
- Publisher
- mradermacher
- Parameters
- 122B (mixture-of-experts)
- VRAM required
- ~74.7 GB at Q4_K_M, 4k context (estimated) — runs on 80 GB cards
- Hardware class
- 100B+ — server-class deployment
- Licence
- apache-2.0
- Task
- Unspecified
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, GGUF (imatrix), FP8, EXL3, NVFP4
- First indexed
- 11 Mar 2026
- Last updated
- 31 Oct 2026
Ablation technique
How Qwen3.5-122B-A10B-heretic-v2 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
“For a convenient overview and download list, visit our model page for this model.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Qwen3.5-122B-A10B-heretic-v2-i1-GGUF --local-dir ./qwen3.5-122b-a10b-heretic-v2
llama-cli -m ./qwen3.5-122b-a10b-heretic-v2/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "mradermacher/Qwen3.5-122B-A10B-heretic-v2-i1-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve mradermacher/Qwen3.5-122B-A10B-heretic-v2-i1-GGUF --trust-remote-codeDownloads and variants (6)
Every published repository of Qwen3.5-122B-A10B-heretic-v2, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Qwen3.5-122B-A10B-heretic-v2-i1-GGUFsource | GGUF (imatrix) | 4.1K |
| mradermacher/Qwen3.5-122B-A10B-heretic-v2-GGUF | GGUF | 1.5K |
| nivvis/Qwen3.5-122B-A10B-heretic-v2-FP8 | FP8 | 188 |
| UnstableLlama/Qwen3.5-122B-A10B-heretic-v2-exl3-4.06bpw | EXL3 | 19 |
| catplusplus/Qwen3.5-122B-A10B-heretic-v2-NVFP4 | NVFP4 | 11 |
| UnstableLlama/Qwen3.5-122B-A10B-heretic-v2-exl3-2.08bpw | EXL3 | 8 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Qwen3.5-122B-A10B-heretic-v2 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 its base model, not added by ablation — at 122B 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.
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 Qwen3.5-122B-A10B-heretic-v2 is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with mradermacher, and does not independently verify publisher claims. See responsible use.