NemoMix-Unleashed-12B
12BOther
from MarinaraSpaghetti/NemoMix-Unleashed-12B
↓ 23.6K♥ 28026 repos
LFM2-24B-A2B-abliterated
LFM2-24B-A2B-abliterated is a refusal-ablated variant of LiquidAI/LFM2-24B-A2B, published on Hugging Face by huihui-ai. It is 24B total parameters (mixture-of-experts), 24GB VRAM class class and other licence. It is distributed across 9 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 8.1K downloads.
Specification
- Model name
- LFM2-24B-A2B-abliterated
- Source repository
- huihui-ai/Huihui-LFM2-24B-A2B-abliterated
- Base model
- LiquidAI/LFM2-24B-A2B
- Publisher
- huihui-ai
- Parameters
- 24B (mixture-of-experts)
- Hardware class
- 10B – 25B — 24GB VRAM class
- Licence
- other
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix)
- First indexed
- 28 Feb 2026
- Last updated
- 29 Aug 2026
Ablation technique
How LFM2-24B-A2B-abliterated was modified, and what that implies.
Abliteration (directional ablation)
The single residual-stream direction that mediates refusal is identified from harmful/harmless prompt pairs and projected out of the model weights.
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
“This is an uncensored version of LiquidAI/LFM2-24B-A2B created with abliteration (see remove-refusals-with-transformers to know more about it). This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/LFM2-24B-A2B-abliterated-GGUF --local-dir ./lfm2-24b-a2b-abliterated
llama-cli -m ./lfm2-24b-a2b-abliterated/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "huihui-ai/Huihui-LFM2-24B-A2B-abliterated"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve huihui-ai/Huihui-LFM2-24B-A2B-abliterated --trust-remote-codeDownloads and variants (8)
Every published repository of LFM2-24B-A2B-abliterated, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/LFM2-24B-A2B-abliterated-GGUF | GGUF | 4.9K |
| mradermacher/Huihui-LFM2-24B-A2B-abliterated-i1-GGUF | GGUF (imatrix) | 1.6K |
| mradermacher/Huihui-LFM2-24B-A2B-abliterated-GGUF | GGUF | 764 |
| mradermacher/LFM2-24B-A2B-abliterated-i1-GGUF | GGUF (imatrix) | 529 |
| paperscarecrow/LFM2-24B-A2B-Abliterated | GGUF | 138 |
| huihui-ai/Huihui-LFM2-24B-A2B-abliteratedsource | Transformers | 66 |
| wangzhang/LFM2-24B-A2B-abliterated | Transformers | 33 |
| Justbackup/LFM2-24B-A2B-abliterated | Transformers | 14 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- LFM2-24B-A2B-abliterated 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 LiquidAI/LFM2-24B-A2B, not added by ablation — at 24B 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 other, but the terms that bind you are LiquidAI/LFM2-24B-A2B’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 LFM2-24B-A2B-abliterated is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with huihui-ai, and does not independently verify publisher claims. See responsible use.