Instructions to use ApollonLabs/Heliactis-1-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ApollonLabs/Heliactis-1-4B with Transformers:

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

pipe = pipeline("text-generation", model="ApollonLabs/Heliactis-1-4B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("ApollonLabs/Heliactis-1-4B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ApollonLabs/Heliactis-1-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ApollonLabs/Heliactis-1-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ApollonLabs/Heliactis-1-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ApollonLabs/Heliactis-1-4B

SGLang

How to use ApollonLabs/Heliactis-1-4B 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 "ApollonLabs/Heliactis-1-4B" \
    --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": "ApollonLabs/Heliactis-1-4B",
		"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 "ApollonLabs/Heliactis-1-4B" \
        --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": "ApollonLabs/Heliactis-1-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ApollonLabs/Heliactis-1-4B with Docker Model Runner:
```
docker model run hf.co/ApollonLabs/Heliactis-1-4B
```

ApollonLabsAI — Towards the Sun

Heliactis-1-4B

Heliactis means "sunbeam". It is the ancient-style form of the living Greek word ηλιαχτίδα (ηλιακτίδα), built from the ancient hḗlios ("sun") and aktís ("ray").

Full-precision weights (fp16, safetensors) of Heliactis 1, an Apollon Labs fine-tune of XHToken/Spark-X2.5-4B. These are the same weights as the F16 file in ApollonLabs/Heliactis-1-4B-GGUF, which holds the full model card: measured results, the base-model defect we reduced, and the honest limitations. Please read it before use.

Usage

The architecture is not in transformers, so it ships as custom code:

from transformers import AutoModelForCausalLM, AutoTokenizer
tok = AutoTokenizer.from_pretrained("ApollonLabs/Heliactis-1-4B")
model = AutoModelForCausalLM.from_pretrained("ApollonLabs/Heliactis-1-4B", trust_remote_code=True)

Trained and evaluated with thinking disabled (enable_thinking=False in the chat template). To block the 502 orphan-byte tokens, pass their ids (ban-ids.json in the GGUF repo) as a bad-words or logit-bias list at generation time.

License and attribution

Apache 2.0, inherited from the base model XHToken/Spark-X2.5-4B. Derivative work by Apollon Labs. The custom code files come unchanged from the base model.

Apollon Labs — Towards the Sun.

Downloads last month: 384

Safetensors

Model size

4B params

Tensor type

F16

Model tree for ApollonLabs/Heliactis-1-4B

Base model

XHToken/Spark-X2.5-4B-Base

Finetuned

XHToken/Spark-X2.5-4B

Finetuned

(24)

this model