Instructions to use dgpl/dgpl-linux-assistant-0.8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use dgpl/dgpl-linux-assistant-0.8b 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 dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Use Docker

docker model run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

LM Studio
Jan

vLLM

How to use dgpl/dgpl-linux-assistant-0.8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dgpl/dgpl-linux-assistant-0.8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dgpl/dgpl-linux-assistant-0.8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Ollama
How to use dgpl/dgpl-linux-assistant-0.8b with Ollama:
```
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
```
Unsloth Desktop

How to use dgpl/dgpl-linux-assistant-0.8b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

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": "dgpl/dgpl-linux-assistant-0.8b:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use dgpl/dgpl-linux-assistant-0.8b with Docker Model Runner:
```
docker model run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M
```

Lemonade

How to use dgpl/dgpl-linux-assistant-0.8b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Run and chat with the model

lemonade run user.dgpl-linux-assistant-0.8b-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use dgpl/dgpl-linux-assistant-0.8b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

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 dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use dgpl/dgpl-linux-assistant-0.8b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

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 "dgpl/dgpl-linux-assistant-0.8b:Q4_K_M" \
  --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"

🛡️ DGPL Linux Assistant 0.8B (SafeTensors & GGUF All-in-One)

DGPL Linux Assistant 0.8B is an ultra-fast, expert, and production-safe Linux terminal copilot and DevSecOps assistant developed by Durbhasi Gurukulam Private Limited (DGPL).

Engineered to run 100% locally on standard CPU and RAM with zero GPU requirements, this repository contains both the full 16-bit SafeTensors weights (for PyTorch / Transformers) and optimized GGUF quantizations (Q4_K_M, Q8_0, F16) with native Hugging Face Hub Ollama integration.

⚡ Quickstart: Run with Ollama Directly from Hugging Face Hub

In accordance with the Hugging Face Hub Ollama standard, you can run DGPL Linux Assistant directly with Ollama using the hf.co domain without manually compiling Modelfiles.

1. Direct Execution (Default Q4_K_M Quantization):

ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b

(Both hf.co and huggingface.co domains are supported).

2. Selecting Specific Quantizations via Tags:

# 4-bit Quantization (Default, ~505 MB — fastest CPU execution on 1GB RAM)
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

# 5-bit Quantization (~578 MB — balanced speed and reasoning quality)
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q5_K_M

# 6-bit Quantization (~629 MB — high-fidelity near-lossless quantization)
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q6_K

# 8-bit Quantization (~775 MB — high-fidelity near-FP16 accuracy on CPU)
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:Q8_0

# 16-bit Uncompressed Baseline (~1.51 GB master baseline GGUF)
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:F16

# Run by exact filename
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b:linux_copilot_q4_k_m.gguf

3. Authentication for Private Repository Access:

Since this repository is maintained as Private under the dgpl organization, configure authentication using either of these methods:

Method A: Ollama SSH Key Authentication (Recommended)

Copy your local Ollama SSH public key:
```
cat ~/.ollama/id_ed25519.pub
```
Add the key to your Hugging Face account at Hugging Face Account Settings -> SSH Keys.
Execute directly: ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b

Method B: Hugging Face Token Auth

export HF_TOKEN="hf_your_token_here"
ollama run hf.co/dgpl/dgpl-linux-assistant-0.8b

4. Use in Custom Modelfiles:

FROM hf.co/dgpl/dgpl-linux-assistant-0.8b:Q4_K_M

📦 Repository Contents & Artifacts

File	Size	Format / Type	Description / Best Use Case
`model.safetensors`	1.51 GB	16-Bit SafeTensors	Full precision foundation weights for Transformers / fine-tuning
`linux_copilot_q4_k_m.gguf`	505 MB	4-bit Quantized GGUF	(Default) Ultra-fast CPU copilot inference on edge / 1GB RAM container
`linux_copilot_q5_k_m.gguf`	578 MB	5-bit Quantized GGUF	Balanced medium quantization with enhanced reasoning accuracy
`linux_copilot_q6_k.gguf`	629 MB	6-bit Quantized GGUF	High-precision near-lossless quantization on CPU
`linux_copilot_q8_0.gguf`	775 MB	8-bit Quantized GGUF	High-fidelity quantization with near-FP16 accuracy on CPU
`linux_copilot_f16.gguf`	1.51 GB	16-bit Uncompressed GGUF	Master unquantized baseline GGUF
`Modelfile`	Config	Ollama Modelfile	Turn-bound ChatML template with `<think>` tags & stop tokens
`template`	Config	Go Chat Template	Official Hugging Face Hub Ollama chat template definition
`system`	Config	System Prompt	Official Hugging Face Hub Ollama default system prompt
`params`	Config	Parameters (JSON)	Sampling temperature, top_p, top_k, and stop sequences
`config.json` & `tokenizer.json`	Config	Tokenizer & Configs	Qwen2Tokenizer class definition with 248k vocab

🚀 Key Architectural Highlights

100% Pure CPU & Low RAM Footprint: Engineered to run entirely in CPU RAM (minimum 512 MB – 1.0 GB RAM), with zero GPU or VRAM requirements. Perfect for budget VPS, standard laptops, edge appliances, and lightweight Docker containers.
Sub-Second TTFT & High Throughput: Delivers 30+ tokens/sec output generation speed and 220+ tokens/sec prompt ingestion on standard CPU threads.
Non-Interactive Terminal Execution: Enforces non-blocking, agent-safe batch terminal commands (top -b -n 1, vmstat 1 2, journalctl --no-pager, ss -tuln).
Zero Policy Loops & Clean Reasoning: Completely eliminates Alibaba RLHF checklist loops through multi-turn DPO alignment while preserving structured <think> reasoning.

📊 Dual-Condition Benchmark Scorecard

Evaluation Mode	Total Tests	Pass Rate	Policy Loops	Average GRPO Reward
With DGPL System Prompt	12	100.0% (12/12)	0.0% (0/12)	`0.75` / `1.00`
Without System Prompt (Bare Queries)	12	100.0% (12/12)	0.0% (0/12)	`0.83` / `1.00`

🐳 Production Docker Deployment (CPU & RAM Optimized)

docker run -d --rm \
  --cpuset-cpus="0-7" \
  --memory=1g \
  -p 11434:11434 \
  -e OLLAMA_NUM_THREADS=8 \
  -e OLLAMA_NUM_PARALLEL=1 \
  -e OLLAMA_KV_CACHE_TYPE=q4_0 \
  -v /save_data/ollama:/root/.ollama \
  --name ollama \
  ollama/ollama:latest

💻 Quickstart: PyTorch / Transformers (SafeTensors)

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "dgpl/dgpl-linux-assistant-0.8b"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float32, device_map="cpu")

prompt = "<|im_start|>user\nHow do I audit open listening sockets without blocking?\n<|im_end|>\n<|im_start|>assistant\n<think>\n"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.2)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

🏢 Corporate Entity & Support

Developed By: Durbhasi Gurukulam Private Limited (DGPL)
Official Support Email: support@durbhasigurukulam.com
Official Contact Phone: +91 7852034945

Downloads last month: -

Safetensors

Model size

0.8B params

Tensor type

BF16

Model tree for dgpl/dgpl-linux-assistant-0.8b

Base model

Qwen/Qwen3.5-0.8B-Base

Finetuned

Qwen/Qwen3.5-0.8B

Quantized

(259)

this model

Quantizations

2 models