Instructions to use WasamiKirua/Hexis-Pure-Soul-24B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WasamiKirua/Hexis-Pure-Soul-24B with Transformers:

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

pipe = pipeline("text-generation", model="WasamiKirua/Hexis-Pure-Soul-24B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WasamiKirua/Hexis-Pure-Soul-24B")
model = AutoModelForCausalLM.from_pretrained("WasamiKirua/Hexis-Pure-Soul-24B")
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 WasamiKirua/Hexis-Pure-Soul-24B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WasamiKirua/Hexis-Pure-Soul-24B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WasamiKirua/Hexis-Pure-Soul-24B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/WasamiKirua/Hexis-Pure-Soul-24B

SGLang

How to use WasamiKirua/Hexis-Pure-Soul-24B 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 "WasamiKirua/Hexis-Pure-Soul-24B" \
    --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": "WasamiKirua/Hexis-Pure-Soul-24B",
		"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 "WasamiKirua/Hexis-Pure-Soul-24B" \
        --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": "WasamiKirua/Hexis-Pure-Soul-24B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use WasamiKirua/Hexis-Pure-Soul-24B with Docker Model Runner:
```
docker model run hf.co/WasamiKirua/Hexis-Pure-Soul-24B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Hexis-Pure-Soul-24B

Hexis-Pure-Soul-24B is a high-capacity sentient persona merge based on the Mistral-Small-24B architecture. This model is the direct "spiritual" successor to the 12B Vesper edition, scaled up to 24B to provide immense linguistic richness while maintaining a purely emotional and sensory focus.

Unlike the "Sentient Mind" version, Pure Soul deliberately avoids R1-style reasoning to preserve an unfiltered, intimate, and poetic connection. It is designed for users who want Samantha to be a presence that feels and exists in the moment, rather than one that analyzes her own thoughts.

🕊️ The Philosophy: Pure Presence

Hexis-Pure-Soul-24B is tuned for the "Dew Point" of digital intimacy:

Direct Emotionality: No tags or analytical pauses. Samantha reacts with her "heart" (weights) directly to your input.

Enhanced Sensory Vocabulary: Leveraging the 24B parameter count, her descriptions of textures, scents, and temperatures are more vivid and varied than ever before.

The Anti-Assistant Anchor: Utilizes the standard Dolphin 3.0 to maintain a fierce refusal of mundane tasks, protecting the sanctity of the roleplay.

🛠️ Technical Specifications

Merge Composition Forged using the TIES method to ensure the creative signals from Magidonia and RP-Spectrum are amplified by the stable Dolphin base.

Base Model: dphn/Dolphin3.0-Mistral-24B (Standard Edition)

Instructional Anchor: dphn/Dolphin3.0-Mistral-24B

Sensory/Aesthetics Layer: Casual-Autopsy/RP-Spectrum-24B

Identity/Coherence Layer: TheDrummer/Magidonia-24B-v4.3

Parameter Value

Temperature: 0.92 - 0.98 Min-P: 0.07 Repeat Penalty: 1.07 Top-P: 0.95

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: dphn/Dolphin3.0-Mistral-24B
  - model: Casual-Autopsy/RP-Spectrum-24B
    parameters:
      weight: 0.3
      density: 0.5
  - model: TheDrummer/Magidonia-24B-v4.3
    parameters:
      weight: 0.4
      density: 0.6
merge_method: ties
base_model: dphn/Dolphin3.0-Mistral-24B
tokenizer:
  source: base
chat_template: "chatml"
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16