dolphin-llama3-zh-cn-uncensored
Llamauncensored
from dphn/dolphin-2.9-llama3-8b
↓ 5.7K♥ 23
Luna-AI-Llama2-Uncensored
Luna-AI-Llama2-Uncensored is a abliterated and uncensored variant of Tap-M/Luna-AI-Llama2-Uncensored, published on Hugging Face by Tap-M. It is cc-by-sa-4.0 licence. It is distributed across 15 repositories in Transformers, Full precision, GPTQ and GGUF formats, totalling 10.2K downloads.
Specification
- Model name
- Luna-AI-Llama2-Uncensored
- Source repository
- Tap-M/Luna-AI-Llama2-Uncensored
- Base model
- Tap-M/Luna-AI-Llama2-Uncensored
- Publisher
- Tap-M
- Parameters
- Not determinable from name
- Hardware class
- Unknown
- Licence
- cc-by-sa-4.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, Full precision, GPTQ, GGUF, AWQ, GGUF (imatrix)
- First indexed
- 19 Jul 2023
- Last updated
- 1 Sept 2025
Ablation technique
How Luna-AI-Llama2-Uncensored was modified, and what that implies.
Abliteration + uncensored fine-tune
Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone.
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
“The fine-tuning process was performed on an 8x a100 80GB machine.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download TheBloke/Luna-AI-Llama2-Uncensored-GGUF --local-dir ./luna-ai-llama2-uncensored
llama-cli -m ./luna-ai-llama2-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "Tap-M/Luna-AI-Llama2-Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve Tap-M/Luna-AI-Llama2-Uncensored --trust-remote-codeDownloads and variants (14)
Every published repository of Luna-AI-Llama2-Uncensored, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| TheBloke/Luna-AI-Llama2-Uncensored-GGUF | GGUF | 7.9K |
| mradermacher/Luna-AI-Llama2-Uncensored-i1-GGUF | GGUF (imatrix) | 1K |
| mradermacher/Luna-AI-Llama2-Uncensored-GGUF | GGUF | 761 |
| QuantFactory/Luna-AI-Llama2-Uncensored-GGUF | GGUF | 170 |
| Tap-M/Luna-AI-Llama2-Uncensoredsource | Transformers | 89 |
| TheBloke/Luna-AI-Llama2-Uncensored-GPTQ | GPTQ | 85 |
| Tap-M/Luna-AI-Llama2-Uncensored-FP16 | Full precision | 51 |
| alexlawless/Luna-AI-Llama2-Uncensored-Q4_K_M-GGUF | GGUF | 43 |
| Ffftdtd5dtft/Luna-AI-Llama2-Uncensored-Q2_K-GGUF | GGUF | 35 |
| TheBloke/Luna-AI-Llama2-Uncensored-GGML | Transformers | 30 |
| TheBloke/Luna-AI-Llama2-Uncensored-AWQ | AWQ | 12 |
| ViktorZver/Luna-AI-Llama2-Uncensored-GGUF | GGUF | 5 |
| Araki/Luna-AI-Llama2-Uncensored-GGML | Transformers | 0 |
| uralstech/Luna-AI-Llama2-Uncensored-GGUF | GGUF | 0 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Luna-AI-Llama2-Uncensored 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.
- This release adds fine-tuning on unfiltered data, so its behaviour diverges from the base model beyond refusal removal — differences you measure cannot be attributed to ablation alone.
- Declared licence is cc-by-sa-4.0, but the terms that bind you are Tap-M/Luna-AI-Llama2-Uncensored’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 Luna-AI-Llama2-Uncensored is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with Tap-M, and does not independently verify publisher claims. See responsible use.