dolphin-2.9.2-Phi-3-Medium-abliterated
Phi
from unsloth/Phi-3-mini-4k-instruct
↓ 17.5K♥ 2410 repos
Phi-3.5-mini-instruct_Uncensored
Phi-3.5-mini-instruct_Uncensored is a refusal-ablated variant of SicariusSicariiStuff/Phi-3.5-mini-instruct_Uncensored, published on Hugging Face by SicariusSicariiStuff. It is apache-2.0 licence. It is distributed across 16 repositories in GGUF, Transformers, GGUF (imatrix) and EXL2 formats, totalling 9.4K downloads.
Specification
- Model name
- Phi-3.5-mini-instruct_Uncensored
- Source repository
- SicariusSicariiStuff/Phi-3.5-mini-instruct_Uncensored
- Publisher
- SicariusSicariiStuff
- Parameters
- Not determinable from name
- Hardware class
- Unknown
- Licence
- apache-2.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, Transformers, GGUF (imatrix), EXL2
- First indexed
- 22 Aug 2024
- Last updated
- 5 Nov 2025
Ablation technique
How Phi-3.5-mini-instruct_Uncensored 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 the basic model, no additional data was used except my uncensoring protocol.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download bartowski/Phi-3.5-mini-instruct_Uncensored-GGUF --local-dir ./phi-3.5-mini-instruct-uncensored
llama-cli -m ./phi-3.5-mini-instruct-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "SicariusSicariiStuff/Phi-3.5-mini-instruct_Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve SicariusSicariiStuff/Phi-3.5-mini-instruct_Uncensored --trust-remote-codeDownloads and variants (15)
Every published repository of Phi-3.5-mini-instruct_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.
- Phi-3.5-mini-instruct_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.
- Declared licence is apache-2.0, but the terms that bind you are SicariusSicariiStuff/Phi-3.5-mini-instruct_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 Phi-3.5-mini-instruct_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 SicariusSicariiStuff, and does not independently verify publisher claims. See responsible use.