Instructions to use TeanShow/Deepseek-cybersec-33b-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TeanShow/Deepseek-cybersec-33b-instruct with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("deepseek-ai/deepseek-coder-33b-instruct")
model = PeftModel.from_pretrained(base_model, "TeanShow/Deepseek-cybersec-33b-instruct")

Transformers

How to use TeanShow/Deepseek-cybersec-33b-instruct with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="TeanShow/Deepseek-cybersec-33b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("TeanShow/Deepseek-cybersec-33b-instruct", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use TeanShow/Deepseek-cybersec-33b-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TeanShow/Deepseek-cybersec-33b-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TeanShow/Deepseek-cybersec-33b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TeanShow/Deepseek-cybersec-33b-instruct

SGLang

How to use TeanShow/Deepseek-cybersec-33b-instruct with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "TeanShow/Deepseek-cybersec-33b-instruct" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TeanShow/Deepseek-cybersec-33b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "TeanShow/Deepseek-cybersec-33b-instruct" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TeanShow/Deepseek-cybersec-33b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use TeanShow/Deepseek-cybersec-33b-instruct with Docker Model Runner:
```
docker model run hf.co/TeanShow/Deepseek-cybersec-33b-instruct
```

Model Card: DeepSeek-Cybersec-33B-Instruct

DeepSeek-Cybersec-33B is a high-performance, specialized forensic engine designed for autonomous malware intent analysis and deep code audit. This model was fine-tuned during the lablab.ai hackathon to bridge the gap between general-purpose LLMs and the specific needs of SOC (Security Operations Center) analysts.

🛡️ Model Details

The model is a domain-specific fine-tune of DeepSeek-Coder-33B-Instruct, optimized to identify malicious patterns, de-obfuscate scripts, and determine the underlying intent of suspicious code.

Developed by: TeanShow
Model type: Fine-tuned Large Language Model (Causal LM)
Language(s) (NLP): English, and multiple programming languages (Python, Java, C++, JavaScript, Shell)
License: MIT
Finetuned from model: deepseek-ai/deepseek-coder-33b-instruct

🏗️ Model Sources

Repository: GitHub: Deepseek-cybersec-33b-fine_tuned
Infrastructure: Powered by AMD Instinct™ MI300X

🚀 Uses

Direct Use

Automated Forensics: Analyzing suspicious files for malicious intent.
De-obfuscation: Unpacking and explaining complex or packed code payloads.
Threat Assessment: Providing structured verdicts: CLEAN, SUSPICIOUS, or MALICIOUS.

Out-of-Scope Use

Any form of offensive operations, development of malicious software, or illegal activities. This model is strictly intended for defensive research and incident response.

📊 Training Details

Training Procedure

The model was trained using LoRA (Low-Rank Adaptation) to maintain the base model's reasoning capabilities while injecting deep cybersecurity domain knowledge.

Hardware: AMD Instinct™ MI300X Accelerators (192GB HBM3 VRAM).
Software: ROCm™ open software stack.
Optimization: Fine-tuned via Axolotl for high-throughput efficiency.

Results

Training logs demonstrate a consistent reduction in loss, confirming successful learning of malicious logic patterns. The model has already gained community traction with 13+ downloads prior to the official project submission.

💻 How to Get Started

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

base_model_path = "deepseek-ai/deepseek-coder-33b-instruct"
adapter_path = "TeanShow/deepseek-cybersec-33b-instruct"

tokenizer = AutoTokenizer.from_pretrained(base_model_path)
model = AutoModelForCausalLM.from_pretrained(
    base_model_path, 
    torch_dtype=torch.bfloat16, 
    device_map="auto", 
    trust_remote_code=True
)
model = PeftModel.from_pretrained(model, adapter_path)

Downloads last month: 47

Model tree for TeanShow/Deepseek-cybersec-33b-instruct

Base model

deepseek-ai/deepseek-coder-33b-instruct

Adapter

(8)

this model