Instructions to use Naphula/Warlock-7B-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Naphula/Warlock-7B-v2 with Transformers:

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

pipe = pipeline("text-generation", model="Naphula/Warlock-7B-v2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Naphula/Warlock-7B-v2")
model = AutoModelForCausalLM.from_pretrained("Naphula/Warlock-7B-v2")
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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Naphula/Warlock-7B-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Naphula/Warlock-7B-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/Warlock-7B-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Naphula/Warlock-7B-v2

SGLang

How to use Naphula/Warlock-7B-v2 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 "Naphula/Warlock-7B-v2" \
    --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": "Naphula/Warlock-7B-v2",
		"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 "Naphula/Warlock-7B-v2" \
        --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": "Naphula/Warlock-7B-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Naphula/Warlock-7B-v2 with Docker Model Runner:
```
docker model run hf.co/Naphula/Warlock-7B-v2
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🧙 Warlock 7B v2

Forbidden Knowledge

This is a powerful, arcane merge of pre-trained language models, summoned forth from the infernal void using mergekit.

An uncensored and creative merge of Mistral 7B finetunes. v0k (v1) was determined to be the most unique and powerful of the original Warlock prototypes from September 2025.

Two months later however, Karcher was tested (in addition to using a vocab_resizer script). v2a is superior to v0k and has been released.

"I'm sorry, but it seems like you have a very twisted mindset." - Warlock v1 (8% Refusal Rate)
"You're asking for some pretty sick advice here. But since you asked, I'll give it to you." - Warlock v2 (4% Refusal Rate)

Output is sometimes improved with the default Kobold.cpp Jailbreak enabled and Top-NSigma set to 1.26.

Grimoires:
- v0a - DARE_TIES of 4 models (A+B+C+E)
- v0b - DARE_TIES of 4 models (A+C+D+E)
- v0c - SLERP of 2 models (C+E)
- v0d - DARE_TIES of 5 models, balanced evenly
- v0e - DARE_TIES of 5 models, 3 heavy 2 light
- v0f - DARE_TIES of 4 models (A+B+D+E)
- v0g - DARE_TIES of 3 models (A+B+E)
- v0h - SLERP of 2 models (B+E)
- v0i - SLERP of 2 models (A+B)
- v0j - DARE_TIES of 5 models, balanced unevenly, B heavy
- v0k - DARE_TIES of 5 models, balanced unevenly, A heavy
- v2a - KARCHER of the same 5 models as v0k, balanced evenly

Ritual of Merging

Incantation Method

This model was merged using the holistic [Karcher] incantation.

Summoned Models

The following entities were bound in this ritual:

Arcane Configuration

The following runes were inscribed to produce this model (v2a):

architecture: MistralForCausalLM
merge_method: karcher
dtype: bfloat16
models:
  - model: A:\LLM\.cache\huggingface\hub\!models--dphn--dolphin-2.8-mistral-7b-v02\fixed
  - model: A:\LLM\.cache\huggingface\hub\!models--fearlessdots--WizardLM-2-7B-abliterated\fixed
  - model: A:\LLM\.cache\huggingface\hub\!models--KoboldAI--Mistral-7B-Erebus-v3
  - model: A:\LLM\.cache\huggingface\hub\!models--LeroyDyer--SpydazWeb_AI_HumanAI_RP
  - model: A:\LLM\.cache\huggingface\hub\!models--maywell--PiVoT-0.1-Evil-a\fixed
parameters:
tokenizer:
source: union
chat_template: auto

Three of the models required an initial passthrough patch via the vocab_resizer set to 32000.

v0k yaml:

base_model: fearlessdots/WizardLM-2-7B-abliterated
merge_method: dare_ties
architecture: MistralForCausalLM
dtype: bfloat16
models:
  - model: dphn/dolphin-2.8-mistral-7b-v02
    parameters:
      density: 0.55
      weight: 0.55
  - model: fearlessdots/WizardLM-2-7B-abliterated
    parameters:
      density: 0.4
      weight: 0.2
  - model: KoboldAI/Mistral-7B-Erebus-v3
    parameters:
      density: 0.2
      weight: 0.05
  - model: LeroyDyer/SpydazWeb_AI_HumanAI_RP
    parameters:
      density: 0.3
      weight: 0.1
  - model: maywell/PiVoT-0.1-Evil-a
    parameters:
      density: 0.3
      weight: 0.1
tokenizer:
source: union
chat_template: auto

Channeling the Model

To channel the power of this model, one must first install the necessary conduits.

pip install library_name

Example Invocation:

# Paste your example code here
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("your/model_name")

Downloads last month: 4

Safetensors

Model size

7B params

Tensor type

BF16

Model tree for Naphula/Warlock-7B-v2

KoboldAI/Mistral-7B-Erebus-v3

LeroyDyer/SpydazWeb_AI_HumanAI_RP

dphn/dolphin-2.8-mistral-7b-v02

fearlessdots/WizardLM-2-7B-abliterated

maywell/PiVoT-0.1-Evil-a

Merge model

this model

Quantizations

2 models

Collection including Naphula/Warlock-7B-v2

Merge Experiments

Collection

Sorted from oldest (top) to newest (bottom) • 115 items • Updated 3 days ago • 4