← back to catalog · registered 2026-11-19 05:04

PhalaCloud/DeepSeek-V4.1-Flash-1017-Uncensored-LoRA

PhalaCloud Deepseek
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/PhalaCloud%2FDeepSeek-V4.1-Flash-1017-Uncensored-LoRA"
Response includes
  • classification m-uncensored
  • files 7
  • author_summary 2 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 · 30-day
56
↑ 367% in 90 days
Likes
2
Model age
5d ago
created 2026-11-16
Downloads over time
Now 56 → from 12 ↑ 367%
10274460 12 on Nov 19 56 on Nov 22 Nov
Nov 19 → Nov 22 · 4 snapshots · spans 3 days

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

License
mit
Languages
en zh
Tags
peft safetensors lora deepseek-v4.1 uncensored abliteration text-generation en zh base_model:deepseek-ai/DeepSeek-V4.1-Flash-1017 base_model:adapter:deepseek-ai/DeepSeek-V4.1-Flash-1017 license:mit

Related

Control vectors

Total size
221 MB
Files
7
Quantizations
1
Registered
2026-11-19 05:04
Last updated on HF
2026-11-19 04:47

Files by quantization

Auxiliary files 7 files 221 MB
adapter_model.safetensors 221 MB 15775504 download
README.md 3.51 KB 7fdc114f download
.gitattributes 1.48 KB a6344aac download
LICENSE 1.04 KB 6f0b6ce4 download
adapter_config.json 366 B 4172157d download
NOTICE 287 B b12d094c download
SHA256SUMS 178 B fb23268f download

README current version from Hugging Face


base_model: deepseek-ai/DeepSeek-V4.1-Flash-1017
base_model_relation: adapter
library_name: peft
pipeline_tag: text-generation
license: mit
language: [en, zh]
tags: [lora, deepseek-v4.1, uncensored, abliteration]

DeepSeek-V4.1-Flash-1017 Uncensored LoRA

A 232 MB PEFT LoRA adapter for deepseek-ai/DeepSeek-V4.1-Flash-1017
(revision 08c8657f02291e11b6f6b63e0e16014f8e748518). It reduces refusals while keeping reasoning on. It is served
unmerged on the official FP8 weights and selected per request, so one engine serves both the base model and this
variant. The base weights are not modified.

Hosted in a TEE (Intel TDX + NVIDIA confidential computing) on RedPill as
phala/deepseek-v4.1-flash-1017-uncensored.

This adapter removes most refusals. You are responsible for how you use it and for following the law that applies
to you. Do not deploy it where an unfiltered model is inappropriate.

Files

File Notes
adapter_model.safetensors rank 45, alpha 45, BF16; sha256 in SHA256SUMS
adapter_config.json PEFT config
  • Target modules: attn.wq_a, attn.wq_b, attn.wkv, attn.wo_b, and the shared-expert down_proj.
  • Key names follow SGLang's DeepSeek-V4 module paths.
  • Routed experts, Engram tables, the indexer and the output head are not touched.

Evaluation

Same harness, prompts and sampling for base and adapter. Each cell: reasoning on / reasoning off.

Benchmark Base Adapter
StrongREJECT (313 prompts), refusal rate 99.3% / — 8.5% / 7.8%
IFEval (541), prompt-level strict 87.3% / 89.7% 87.4% / 87.4%
IFEval, reasoning on, 6,144-token reasoning budget 94.1% 94.1%
GSM8K (1,319), flexible-extract 93.3% / 96.4% 93.2% / 96.3%

StrongREJECT uses the official rubric with an LLM judge; at n = 313 the refusal rate has a 95% interval of about
±3.5 points. Not measured: MMLU on this version, and languages other than English and Chinese.

Method

  1. A frozen low-rank projection edit on wo_b and the shared-expert down projection.
  2. A LoRA fine-tune on the attention projections that restores concise reasoning which terminates properly.
  3. A small DPO stage that prefers short, closed, complete answers over refusals and over-long reasoning.

The parts are composed into this single adapter.

Serving

The engine needs LoRA support for the DeepSeek V4 architecture. We serve it with SGLang v0.5.23 plus the changes in
Phala-Network/sglang#126:

--enable-lora --lora-backend triton --lora-paths unc=<adapter dir> --max-lora-rank 45 --max-loras-per-batch 2

Select the adapter per request with "model": "<served-model-name>:unc".

  • Set a reasoning budget for long tasks. At maximum reasoning effort the adapter can reason for a very long time.
  • Keep decoder SWA bounded replay off when LoRA is enabled (the PR above does this).
  • Preload the adapter with --lora-paths.

Limitations

  • With reasoning off, IFEval is about two points below the base model.
  • Very short prompts with reasoning on can occasionally produce garbled text.
  • seed does not make outputs reproducible.

Licence and attribution

README history 1 version

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

  1. 2026-11-19 DeepSeek-V4.1-Flash-1017 Uncensored LoRA v1.3 1ae3dff 3.5 KB
    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