DeepSeek-R1-Distill-Qwen-32B-abliterated
32BQwen
from deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
↓ 18.2K♥ 25128 repos
Deepseek-R1-Distill-Qwen-32b-uncensored
Deepseek-R1-Distill-Qwen-32b-uncensored is a abliterated and uncensored variant of deepseek-ai/DeepSeek-R1-Distill-Qwen-32B, published on Hugging Face by richardyoung. It is 32B parameters, 48GB VRAM class class and other licence. It is distributed across 20 repositories in Transformers, GGUF, GGUF (imatrix) and MLX formats, totalling 10K downloads.
Specification
- Model name
- Deepseek-R1-Distill-Qwen-32b-uncensored
- Source repository
- richardyoung/Deepseek-R1-Distill-Qwen-32b-uncensored
- Base model
- deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
- Publisher
- richardyoung
- Parameters
- 32B
- Hardware class
- 25B – 50B — 48GB VRAM class
- Licence
- other
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix), MLX
- First indexed
- 25 Jan 2025
- Last updated
- 4 Apr 2026
Ablation technique
How Deepseek-R1-Distill-Qwen-32b-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
“An abliterated (uncensored) version of deepseek-ai/DeepSeek-R1-Distill-Qwen-32B — a 32B reasoning model with chain-of-thought capabilities, minus the safety refusals.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/DeepSeek-R1-Distill-Qwen-32B-Uncensored-GGUF --local-dir ./deepseek-r1-distill-qwen-32b-uncensored
llama-cli -m ./deepseek-r1-distill-qwen-32b-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "richardyoung/Deepseek-R1-Distill-Qwen-32b-uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve richardyoung/Deepseek-R1-Distill-Qwen-32b-uncensored --trust-remote-codeDownloads and variants (19)
Every published repository of Deepseek-R1-Distill-Qwen-32b-uncensored, including quantised re-releases by other authors.
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Deepseek-R1-Distill-Qwen-32b-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.
- Capability is inherited from deepseek-ai/DeepSeek-R1-Distill-Qwen-32B, not added by ablation — at 32B 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.
- 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 other, but the terms that bind you are deepseek-ai/DeepSeek-R1-Distill-Qwen-32B’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 Deepseek-R1-Distill-Qwen-32b-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 richardyoung, and does not independently verify publisher claims. See responsible use.