Instructions to use pulkitrai/SOC7 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use pulkitrai/SOC7 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-0.6B")
model = PeftModel.from_pretrained(base_model, "pulkitrai/SOC7")

Transformers

How to use pulkitrai/SOC7 with Transformers:

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

pipe = pipeline("text-generation", model="pulkitrai/SOC7")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("pulkitrai/SOC7", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use pulkitrai/SOC7 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "pulkitrai/SOC7"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "pulkitrai/SOC7",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/pulkitrai/SOC7

SGLang

How to use pulkitrai/SOC7 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 "pulkitrai/SOC7" \
    --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": "pulkitrai/SOC7",
		"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 "pulkitrai/SOC7" \
        --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": "pulkitrai/SOC7",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use pulkitrai/SOC7 with Docker Model Runner:
```
docker model run hf.co/pulkitrai/SOC7
```

SOC7 — Natural Language SOC Assistant

SOC7 is a cybersecurity-focused conversational AI assistant designed to help Security Operations Center (SOC) analysts interact with Microsoft Sentinel using natural-language prompts.

Instead of requiring an analyst to manually construct every Kusto Query Language (KQL) query, SOC7 is designed to understand questions written in plain English and help translate them into Sentinel-oriented investigations.

For example:

"Show me failed login attempts for the last 24 hours."

SOC7 can help generate the appropriate Sentinel/KQL investigation query and explain what the query is looking for.

Important: SOC7 is a language-model adapter. It does not automatically connect to or access your Microsoft Sentinel environment by itself. A separate application, API integration, or agent workflow is required to execute generated queries against Sentinel.

Model Details

Model Description

SOC7 is a lightweight LoRA adapter fine-tuned from Qwen/Qwen3-0.6B for cybersecurity and SOC-oriented conversational tasks.

The model is intended to assist analysts with:

Natural-language to KQL translation
Microsoft Sentinel investigation queries
Security log analysis
Alert investigation
Threat-hunting questions
Authentication and sign-in investigations
IP and user investigation
Basic incident-response assistance
Explaining KQL queries in simple language
Identifying useful Sentinel tables for an investigation
Structuring SOC investigation workflows

Example

Analyst:

Find users with more than 5 failed sign-ins in the last hour.

SOC7:

SigninLogs
| where TimeGenerated >= ago(1h)
| where ResultType != "0"
| summarize FailedAttempts=count() by UserPrincipalName
| where FailedAttempts > 5
| order by FailedAttempts desc

Downloads last month: 30

Model tree for pulkitrai/SOC7

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Adapter

(616)

this model