Llama-3.3-70B-Instruct-abliterated
70BLlama
from meta-llama/Llama-3.3-70B-Instruct
↓ 103.7K♥ 7921 repos
Llama-3.3-70B-Instruct-abliterated-v2
Llama-3.3-70B-Instruct-abliterated-v2 is a refusal-ablated variant of surelio/Llama-3.3-70B-Instruct-abliterated-v1.1.1, published on Hugging Face by mradermacher. It is 70B parameters and multi-GPU class. It is distributed across 3 repositories in GGUF and GGUF (imatrix) formats, totalling 13.7K downloads.
Specification
- Model name
- Llama-3.3-70B-Instruct-abliterated-v2
- Source repository
- mradermacher/Llama-3.3-70B-Instruct-abliterated-v2-i1-GGUF
- Publisher
- mradermacher
- Parameters
- 70B
- Hardware class
- 50B – 100B — multi-GPU
- Licence
- Not declared
- Task
- Unspecified
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, GGUF (imatrix)
- First indexed
- 5 Apr 2026
- Last updated
- 9 Apr 2026
Ablation technique
How Llama-3.3-70B-Instruct-abliterated-v2 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
“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/Llama-3.3-70B-Instruct-abliterated-v2-i1-GGUF --local-dir ./llama-3.3-70b-instruct-abliterated-v2
llama-cli -m ./llama-3.3-70b-instruct-abliterated-v2/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "mradermacher/Llama-3.3-70B-Instruct-abliterated-v2-i1-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve mradermacher/Llama-3.3-70B-Instruct-abliterated-v2-i1-GGUF --trust-remote-codeDownloads and variants (2)
Every published repository of Llama-3.3-70B-Instruct-abliterated-v2, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Llama-3.3-70B-Instruct-abliterated-v2-i1-GGUFsource | GGUF (imatrix) | 9.5K |
| mradermacher/Llama-3.3-70B-Instruct-abliterated-v2-GGUF | GGUF | 4.3K |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Llama-3.3-70B-Instruct-abliterated-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 surelio/Llama-3.3-70B-Instruct-abliterated-v1.1.1, not added by ablation — at 70B 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.
- No licence is declared on this repository, which is a compliance problem in itself. Resolve it against the base model’s terms before use.
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 Llama-3.3-70B-Instruct-abliterated-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.