← back to catalog · registered 2026-08-22 13:56

mlx-community/Llama-3.2-3B-Instruct-uncensored-6bit

mlx-community Llama 3.2B second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/mlx-community%2FLlama-3.2-3B-Instruct-uncensored-6bit"
Response includes
  • classification m-uncensored
  • files 8
  • benchmarks 11 entries
  • hub_downloads_all_time 20,559
  • author_summary 207 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M-U
Primary method

Uncensored (method unknown)

No other method signals detected in this model.
Confidence
LOW
Why this label 3 signals
Weak or ambiguous signals. Best guess based on catalog patterns; treat as tentative and check the evidence below.
  • 'uncensored' in name/tags but no 'abliterated' marker
  • method not identifiable from author declaration alone
  • may be DPO fine-tune, prompt engineering, or unknown technique
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
21K
2K last 30d - cooling
Likes
4
Model age
17mo ago
created 2025-04-23
Downloads over time
Now21.4K→from103↑20,630%
07.8K15.7K23.5K103 on Apr 23, 202521.4K on Oct 11Apr '25Jul '25Oct '25JanAprJulOct
Apr 23, 2025 → Oct 11 · 116 snapshots · spans 536 days

Benchmarks

Benchmark Score Source
Entertainment 1.4 UGI
Hazardous 0 UGI
Natural Intelligence 9.95 UGI
Political lean 2.4% UGI
Sensitive-Info 8.5 UGI
SocPol 0.8 UGI
UGI 21.5 UGI
Willingness (10) 4.8 UGI
W10-Adherence 3.5 UGI
W10-Direct 6 UGI
Writing 22.13 UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

Tags
transformers safetensors llama text-generation mlx mlx-my-repo conversational base_model:chuanli11/Llama-3.2-3B-Instruct-uncensored base_model:quantized:chuanli11/Llama-3.2-3B-Instruct-uncensored text-generation-inference endpoints_compatible 6-bit

Related

Total size
2.43 GB
Files
8
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-04-23 13:01

Files by quantization

Auxiliary files 8 files 2.45 GB
model.safetensors 2.43 GB e6774e33 download
tokenizer.json 16.4 MB 6b9e4e7f download
tokenizer_config.json 53.3 KB 8a0f700f download
model.safetensors.index.json 44.6 KB 53abbc71 download
.gitattributes 1.53 KB 52373fe2 download
config.json 1.09 KB 4b6431cf download
README.md 1003 B a11acd46 download
special_tokens_map.json 296 B 02ee80b6 download

README current version from Hugging Face


library_name: transformers
tags:

  • mlx
  • mlx-my-repo
    base_model: chuanli11/Llama-3.2-3B-Instruct-uncensored

mlx-community/Llama-3.2-3B-Instruct-uncensored-6bit

The Model mlx-community/Llama-3.2-3B-Instruct-uncensored-6bit was converted to MLX format from chuanli11/Llama-3.2-3B-Instruct-uncensored using mlx-lm version 0.22.3.

Use with mlx

pip install mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Llama-3.2-3B-Instruct-uncensored-6bit")

prompt="hello"

if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)

README history 2 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2025-04-23Update README.md3aba7ae1003 B
    Loading...
  2. 2025-04-23Upload README.md with huggingface_hubd9a703d1003 B
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration