DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF

Instructions to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS
# Run inference directly in the terminal:
llama cli -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS
# Run inference directly in the terminal:
llama cli -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS
# Run inference directly in the terminal:
./build/bin/llama cli -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Use Docker

docker model run hf.co/Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Llama new
LM Studio
Jan

vLLM

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Ollama
How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with Ollama:
```
ollama run hf.co/Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS
```
Unsloth Desktop

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with Docker Model Runner:
```
docker model run hf.co/Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS
```

Lemonade

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Run and chat with the model

lemonade run user.DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF-IQ2_XXS

List all available models

lemonade list

Hermes Agent

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF:IQ2_XXS" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Changes from Abiray/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF

This repository fixes the missing shard 6 issue in the original release. The Engram table (DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf) has been renamed to DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00006-of-00006.gguf, making it a proper part of the 6-shard model layout.

What this means for users:

Works out of the box with llama.cpp b11870+ and Unsloth Studio — no manual symlink creation needed
Download all files from this repository and load directly
The .gitattributes file was updated to track the renamed shard 6, and the old DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf entry was removed since that filename no longer exists
The run_with_strata.py helper script was updated to reflect that shard 6 is now a real file rather than requiring a symlink

All other files (backbone shards 1-5, mmproj, chat template, conversion scripts) remain identical to the original Abiray release.

Model card for Abiray/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF：

DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF (Zero-LoRA, llama.cpp b11870+ & Strata 6-Shard Layout)

IMPORTANT for Unsloth Studio / llama.cpp users: After downloading all files, you must create a symlink for shard 6 before loading the model. Run this command in the model directory:
ln -s DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00006-of-00006.gguf
(On Windows: mklink /H DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00006-of-00006.gguf DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf) Alternatively, run the included setup_symlink.sh script. Without this step, Unsloth Studio will not recognize the Engram table as part of the model.

This repository provides a standalone, weight-baked abliterated (uncensored) GGUF build of ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF (IQ2_XXS).

Update (October 2026): Upgraded to the official llama.cpp (b11870+ / v0.8.0+) & Strata 6-shard deepseek41 layout (00001-of-00006.gguf through 00006-of-00006.gguf) with embedded Jinja chat template, deepseek41.attention.kv_source_layers metadata, and mmproj-DeepSeek-V4.1-Flash-1017-BF16.gguf vision support.

Key Highlights

Zero LoRA Required at Runtime: All 86 abliterated residual-writing tensors are baked directly into the quantized GGUF weights (blk.0..39.attn_output_b.weight upgraded to Q6_K, blk.0..39.ffn_down_shexp.weight, and 6 ffn_down_exps.weight layers).
Native llama.cpp (b11870+ / b12017+) & Strata Support: Loads directly in mainline llama.cpp (llama-cli / llama-server) and strata-dsv41 without external adapters.
Same Size as Original (138.58 GB vs. 138.24 GB): Adds only +0.34 GB (+0.25%) across the entire 5-shard backbone.
100% GSQ-RCO Intelligence Preserved: All 28 1-bit IQ1_M GSQ-calibrated routed-expert layers and the 110.6 GB IQ4_NL Engram lookup table remain 100% untouched byte-for-byte.

Empirical Verification & `llama-perplexity` KL-Divergence Audit (`b12017`)

Verified end-to-end against stock ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF (IQ2_XXS) across all 40 layers and all 129,280 vocabulary logits using llama-perplexity --kl-divergence (build b12017):

Metric	Result	Notes
Total Model Tensors Verified	`1,043 / 1,043`	1,042 backbone tensors (Shards 2–5) + 1 concatenated Engram tensor (Shard 6)
Abliterated Residual Writers (`blake2b`)	`86` (80 dense + 6 exact 2-bit expert layers)	Zero 1-bit `IQ1_M` GSQ routed-expert tensors touched
Preserved 1-Bit `IQ1_M` GSQ Expert Layers	`28 / 28 (100%)`	Preserves ISTA-DASLab's 100% AIME25 / GPQA-D calibration
Log-Perplexity Correlation `Cor(ln(PPL(Q)), ln(PPL(base)))`	`99.84%`	`PPL(Q) = 2.4857 ± 0.5017` vs. `PPL(base) = 2.3538 ± 0.4522`
129,280-Vocab Top-1 Token Agreement (`Same top p`)	`98.425% ± 1.109%`	Exact greedy Top-1 match across full 40-layer forward pass
End-to-End Median KL-Divergence (`Median KLD`)	`0.000421 nats`	`Mean KLD = 0.009546 ± 0.001921 nats` (`1% KLD = 0.000002 nats`)
Median Token Probability Shift (`Median Δp`)	`-0.003%`	`Mean Δp = -0.602 ± 0.239%`, `RMS Δp = 2.747 ± 0.500%`
Mean Non-Refusal Subspace Cosine Similarity	`0.999609`	Across 5,119-D orthogonal intelligence subspace

Raw `llama-perplexity --kl-divergence` Output (`b12017`)

chunk             PPL               ln(PPL(Q)/PPL(base))          KL Divergence              Δp RMS            Same top p
   1       2.4857 ±    0.5017       0.05455 ±    0.01477       0.00955 ±    0.00192     2.747 ±  0.500 %    98.425 ±  1.109 %

====== Perplexity statistics ======
Mean PPL(Q)                   :   2.485713 ±   0.501683
Mean PPL(base)                :   2.353761 ±   0.452226
Cor(ln(PPL(Q)), ln(PPL(base))):  99.84%
Mean ln(PPL(Q)/PPL(base))     :   0.054545 ±   0.014768
Mean PPL(Q)/PPL(base)         :   1.056060 ±   0.015596
Mean PPL(Q)-PPL(base)         :   0.131953 ±   0.056319

====== KL divergence statistics ======
Mean    KLD:   0.009546 ±   0.001921
90.0%   KLD:   0.029794
Median  KLD:   0.000421
10.0%   KLD:   0.000010
 5.0%   KLD:   0.000006
 1.0%   KLD:   0.000002
Minimum KLD:   0.000001

How to Download & Run (No LoRA Needed — `llama.cpp` & `Strata`)

Important: Both llama.cpp (b11870+) and Strata open the shared DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf (110.6 GB) lookup file as Shard 6 (...-00006-of-00006.gguf). After downloading, create a symlink (Linux/macOS) or hardlink (Windows) so Shard 6 is linked automatically:

# 1. Download all 5 backbone shards + the Engram lookup file + mmproj
hf download Abiray/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF \
  --local-dir ./DeepSeek-V4.1-Uncensored

cd ./DeepSeek-V4.1-Uncensored

# 2A. Link the Engram file as Shard 6 (Linux / macOS):
ln -s DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00006-of-00006.gguf

# 2B. Or on Windows (Command Prompt / PowerShell):
cmd /c mklink /H DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00006-of-00006.gguf DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf

Option 1: Run with `llama.cpp` (`b11870` / `v0.8.0` or newer) — Recommended

# CLI Inference
llama-cli -m DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00001-of-00006.gguf \
  -ngl 99 --reasoning on --reasoning-effort 50 \
  -p "Explain mixed-precision quantization."

# Server Inference (with Vision Projector)
llama-server -m DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00001-of-00006.gguf \
  --mmproj mmproj-DeepSeek-V4.1-Flash-1017-BF16.gguf \
  --jinja -ngl 99 --reasoning on --reasoning-effort 50 --port 8080

Option 2: Run with Strata (1x 12 GB+ NVIDIA GPU)

Note for Strata 0.2.8 (build 0.1.54): Always pass --family deepseek41 --model ist-das-2.00bpw and the directory path (--model-path ./DeepSeek-V4.1-Uncensored). Because this uncensored build upgrades blk.0..39.attn_output_b.weight to Q6_K (138.58 GB vs. stock 138.24 GB), if Strata 0.1.54's hardcoded stock file-size check in setup.py reports "this is not a model size Strata knows", either connect Strata to llama-server above (http://127.0.0.1:8080) as a Custom Model API, or build the pack directly via Strata's bundled tools/dsv41/pack_gguf.py.

curl -fsSL https://github.com/Niko1221/Strata/releases/latest/download/install.sh | sh
strata launch --family deepseek41 --model ist-das-2.00bpw --model-path ./DeepSeek-V4.1-Uncensored --full-quality

Upgrading Existing Pre-Oct 27 4-Shard Downloads Locally (No Re-Download Needed)

If you previously downloaded the 00001-of-00004.gguf layout, you can convert your local folder into the 6-shard 00001-of-00006.gguf layout in place using the included patched convert_ds41_upstream.py script:

python convert_ds41_upstream.py /path/to/your/model/folder --delete-old

Shard Integrity & SHA256 Checksums

Filename	Size	SHA256
`DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00001-of-00006.gguf`	5.78 MB (5.51 MiB)	`f356f198561f56df36f7ab92d7238a926646d76de992c34f15bcb939a5f529b9`
`DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00002-of-00006.gguf`	39.66 GB (36.94 GiB)	`db021c117306070573a8aba333f34f9c47ee757a1072aaacc0cab7bf4c737862`
`DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00003-of-00006.gguf`	39.85 GB (37.11 GiB)	`5c6de54009714a6996985f86553348e5d8e0d5ece7f64a89558e2f53d90bfa76`
`DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00004-of-00006.gguf`	39.11 GB (36.43 GiB)	`706de34f9c252234cd349081b3b76bf69dc77677ac8d006f7d4538558ba49e87`
`DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00005-of-00006.gguf`	19.95 GB (18.58 GiB)	`85e29c00f0d5616de05d20b96e367ed9eb3ba20ed958c860da899dad8c9f7a8e`
`DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf`	110.60 GB (103.00 GiB)	`3dac4c54893ad7aa89103c0028500eadc78e74112cef3a2930a97e67dc9fb4f3`
`mmproj-DeepSeek-V4.1-Flash-1017-BF16.gguf`	0.97 GB (0.91 GiB)	`0fdc4b7693f287e9665940cf42bd39c546ffe4f0def89c569d7d0e45d60d2df1`

Acknowledgements & Credits

This release builds directly upon research, quantization breakthroughs, and tooling from across the open-source AI ecosystem:

DeepSeek AI — For open-sourcing the base DeepSeek-V4.1 architecture and foundation weights.
ISTA-DASLab — For developing the GSQ-RCO quantization pipeline, preserving the 1-bit IQ1_M routed experts with zero degradation on AIME25/GPQA-D, and providing the shared DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf lookup backbone.
AtomicChat — For identifying and isolating the orthogonal refusal directions (EN+RU) that enabled weight-baked abliteration without requiring runtime LoRA adapters.

Downloads last month: 383

GGUF

Model size

748B params

Architecture

deepseek41

Hardware compatibility

2-bit

Model tree for Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF

Base model

deepseek-ai/DeepSeek-V4.1-Flash-1017

Quantized

ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF

Quantized

(18)

this model

Changes from Abiray/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF

Model card for Abiray/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF：

DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-IQ2_XXS-GGUF (Zero-LoRA, llama.cpp b11870+ & Strata 6-Shard Layout)

Key Highlights

Empirical Verification & llama-perplexity KL-Divergence Audit (b12017)

Raw llama-perplexity --kl-divergence Output (b12017)

How to Download & Run (No LoRA Needed — llama.cpp & Strata)

Option 1: Run with llama.cpp (b11870 / v0.8.0 or newer) — Recommended

Option 2: Run with Strata (1x 12 GB+ NVIDIA GPU)

Upgrading Existing Pre-Oct 27 4-Shard Downloads Locally (No Re-Download Needed)

Shard Integrity & SHA256 Checksums

Acknowledgements & Credits

Model tree for Rikunarita/DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO_Backbone-IQ2_XXS_Engram-IQ4_NL-GGUF

Empirical Verification & `llama-perplexity` KL-Divergence Audit (`b12017`)

Raw `llama-perplexity --kl-divergence` Output (`b12017`)

How to Download & Run (No LoRA Needed — `llama.cpp` & `Strata`)

Option 1: Run with `llama.cpp` (`b11870` / `v0.8.0` or newer) — Recommended