Instructions to use simpledirect/Vinci-Cyber-30B-1.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use simpledirect/Vinci-Cyber-30B-1.0 with Transformers:

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

pipe = pipeline("text-generation", model="simpledirect/Vinci-Cyber-30B-1.0")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("simpledirect/Vinci-Cyber-30B-1.0")
model = AutoModelForCausalLM.from_pretrained("simpledirect/Vinci-Cyber-30B-1.0", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use simpledirect/Vinci-Cyber-30B-1.0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "simpledirect/Vinci-Cyber-30B-1.0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "simpledirect/Vinci-Cyber-30B-1.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/simpledirect/Vinci-Cyber-30B-1.0

SGLang

How to use simpledirect/Vinci-Cyber-30B-1.0 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 "simpledirect/Vinci-Cyber-30B-1.0" \
    --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": "simpledirect/Vinci-Cyber-30B-1.0",
		"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 "simpledirect/Vinci-Cyber-30B-1.0" \
        --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": "simpledirect/Vinci-Cyber-30B-1.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use simpledirect/Vinci-Cyber-30B-1.0 with Docker Model Runner:
```
docker model run hf.co/simpledirect/Vinci-Cyber-30B-1.0
```

Vinci-Cyber-30B-1.0 / NOTICE

George Pu

Claude Opus 5 (1M context)

Pre-release card review: fix lineage metadata, remove internal status text, align sibling claims

6460019 about 2 months ago

Raw History Blame Contribute Delete

3.8 kB

Vinci-Cyber-30B-1.0
===================

Copyright 2026 SimpleDirect

This product is licensed under the Apache License, Version 2.0 (the "License").
You may obtain a copy of the License in the accompanying LICENSE file or at:

    http://www.apache.org/licenses/LICENSE-2.0


DERIVATION AND PARENT MODEL ATTRIBUTION
---------------------------------------

Vinci-Cyber-30B-1.0 is a MODIFIED / FINE-TUNED derivative of the following work:

    Name:      IBM Granite 4.1 30B
    Source:    ibm-granite/granite-4.1-30b (Hugging Face Hub)
    Revision:  4fae6278f7132abf5e971f9de49ebbad09c54cce
    Publisher: IBM
    License:   Apache License, Version 2.0

Vinci (SimpleDirect) modified the above parent model by training a PEFT/DoRA
adapter with rank-stabilized LoRA on a Vinci-authored cybersecurity corpus and
MERGING that adapter into the parent's weights. This is a modification of IBM's
work within the meaning of section 4(b) and is stated here as required.


WHAT THIS REPOSITORY REDISTRIBUTES
----------------------------------

THIS REPOSITORY REDISTRIBUTES DERIVED MODEL WEIGHTS IN SAFETENSORS FORMAT,
TOGETHER WITH CONFIGURATION AND TOKENIZER FILES CARRIED FORWARD FROM THE PARENT.

Every parameter in the weight files derives from the parent: they are IBM's
Granite 4.1 30B parameters with Vinci's trained deltas added. Of the model's 578
tensors, 448 changed and 130 remain byte-identical to the parent; the changed
population is exactly the 64 layers x 7 targeted module families. This is
redistribution of a modified version of IBM's Apache-2.0 work, and this NOTICE
and the accompanying LICENSE therefore travel with it under sections 4(a), 4(b)
and 4(c) of the License.

Per-file SHA-256 digests for everything published here are recorded in the
accompanying MODEL-PROVENANCE.json, which also records the parent binding
(all 24 parent files digested and matched at the pinned revision above).

The PEFT/DoRA adapter from which these weights were merged is NOT published and
is not distributed alongside this model. It is retained in SimpleDirect's
internal custody as scientific provenance only. It is identified by SHA256
1bb6da75dd09c104a4d5f2a10ac708367ae956fcaf123451106561d2717cefc2 so that the
derivation chain can be stated precisely; no Hugging Face repository for it
exists or is implied. No adapter is required to load these weights, and the
adapter must not be applied again on top of them.


NO ENDORSEMENT OR AFFILIATION
-----------------------------

IBM is not the author of Vinci-Cyber-30B-1.0 and has not reviewed, approved,
endorsed, or sponsored it. Vinci and SimpleDirect are not affiliated with IBM.
The names "IBM" and "Granite" are used solely to identify the parent model from
which this work is derived, as required for accurate attribution, and not to
suggest any association.

Any evaluation results, claims, limitations, or defects described for
Vinci-Cyber-30B-1.0 are attributable to Vinci alone and must not be read as
statements about IBM Granite 4.1 30B.


UPSTREAM NOTICE MATERIAL
------------------------

Apache License 2.0 section 4(d) requires a derivative work to carry forward the
NOTICE text of the original work where the original work includes a NOTICE file.

As checked on 2026-09-21 against the Hugging Face Hub API, the upstream
repository ibm-granite/granite-4.1-30b at the pinned revision above contains no
NOTICE, LICENSE, or COPYING file among its 24 files; its Apache-2.0 licensing is
declared in the repository's card metadata. There is therefore no upstream
NOTICE text to reproduce here, and none has been invented. The attribution above
is provided to satisfy the identification and modification-notice obligations of
sections 4(a), 4(b) and 4(c).

If a NOTICE file is later added upstream, it must be carried into this file.

	Vinci-Cyber-30B-1.0
	===================

	Copyright 2026 SimpleDirect

	This product is licensed under the Apache License, Version 2.0 (the "License").
	You may obtain a copy of the License in the accompanying LICENSE file or at:

	http://www.apache.org/licenses/LICENSE-2.0


	DERIVATION AND PARENT MODEL ATTRIBUTION
	---------------------------------------

	Vinci-Cyber-30B-1.0 is a MODIFIED / FINE-TUNED derivative of the following work:

	Name: IBM Granite 4.1 30B
	Source: ibm-granite/granite-4.1-30b (Hugging Face Hub)
	Revision: 4fae6278f7132abf5e971f9de49ebbad09c54cce
	Publisher: IBM
	License: Apache License, Version 2.0

	Vinci (SimpleDirect) modified the above parent model by training a PEFT/DoRA
	adapter with rank-stabilized LoRA on a Vinci-authored cybersecurity corpus and
	MERGING that adapter into the parent's weights. This is a modification of IBM's
	work within the meaning of section 4(b) and is stated here as required.


	WHAT THIS REPOSITORY REDISTRIBUTES
	----------------------------------

	THIS REPOSITORY REDISTRIBUTES DERIVED MODEL WEIGHTS IN SAFETENSORS FORMAT,
	TOGETHER WITH CONFIGURATION AND TOKENIZER FILES CARRIED FORWARD FROM THE PARENT.

	Every parameter in the weight files derives from the parent: they are IBM's
	Granite 4.1 30B parameters with Vinci's trained deltas added. Of the model's 578
	tensors, 448 changed and 130 remain byte-identical to the parent; the changed
	population is exactly the 64 layers x 7 targeted module families. This is
	redistribution of a modified version of IBM's Apache-2.0 work, and this NOTICE
	and the accompanying LICENSE therefore travel with it under sections 4(a), 4(b)
	and 4(c) of the License.

	Per-file SHA-256 digests for everything published here are recorded in the
	accompanying MODEL-PROVENANCE.json, which also records the parent binding
	(all 24 parent files digested and matched at the pinned revision above).

	The PEFT/DoRA adapter from which these weights were merged is NOT published and
	is not distributed alongside this model. It is retained in SimpleDirect's
	internal custody as scientific provenance only. It is identified by SHA256
	1bb6da75dd09c104a4d5f2a10ac708367ae956fcaf123451106561d2717cefc2 so that the
	derivation chain can be stated precisely; no Hugging Face repository for it
	exists or is implied. No adapter is required to load these weights, and the
	adapter must not be applied again on top of them.


	NO ENDORSEMENT OR AFFILIATION
	-----------------------------

	IBM is not the author of Vinci-Cyber-30B-1.0 and has not reviewed, approved,
	endorsed, or sponsored it. Vinci and SimpleDirect are not affiliated with IBM.
	The names "IBM" and "Granite" are used solely to identify the parent model from
	which this work is derived, as required for accurate attribution, and not to
	suggest any association.

	Any evaluation results, claims, limitations, or defects described for
	Vinci-Cyber-30B-1.0 are attributable to Vinci alone and must not be read as
	statements about IBM Granite 4.1 30B.


	UPSTREAM NOTICE MATERIAL
	------------------------

	Apache License 2.0 section 4(d) requires a derivative work to carry forward the
	NOTICE text of the original work where the original work includes a NOTICE file.

	As checked on 2026-09-21 against the Hugging Face Hub API, the upstream
	repository ibm-granite/granite-4.1-30b at the pinned revision above contains no
	NOTICE, LICENSE, or COPYING file among its 24 files; its Apache-2.0 licensing is
	declared in the repository's card metadata. There is therefore no upstream
	NOTICE text to reproduce here, and none has been invented. The attribution above
	is provided to satisfy the identification and modification-notice obligations of
	sections 4(a), 4(b) and 4(c).

	If a NOTICE file is later added upstream, it must be carried into this file.