Instructions to use AbteeXAILab/lumynax-longctx-prolong-512k-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AbteeXAILab/lumynax-longctx-prolong-512k-instruct with Transformers:

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

pipe = pipeline("text-generation", model="AbteeXAILab/lumynax-longctx-prolong-512k-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("AbteeXAILab/lumynax-longctx-prolong-512k-instruct")
model = AutoModelForCausalLM.from_pretrained("AbteeXAILab/lumynax-longctx-prolong-512k-instruct", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AbteeXAILab/lumynax-longctx-prolong-512k-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AbteeXAILab/lumynax-longctx-prolong-512k-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AbteeXAILab/lumynax-longctx-prolong-512k-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AbteeXAILab/lumynax-longctx-prolong-512k-instruct

SGLang

How to use AbteeXAILab/lumynax-longctx-prolong-512k-instruct 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 "AbteeXAILab/lumynax-longctx-prolong-512k-instruct" \
    --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": "AbteeXAILab/lumynax-longctx-prolong-512k-instruct",
		"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 "AbteeXAILab/lumynax-longctx-prolong-512k-instruct" \
        --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": "AbteeXAILab/lumynax-longctx-prolong-512k-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AbteeXAILab/lumynax-longctx-prolong-512k-instruct with Docker Model Runner:
```
docker model run hf.co/AbteeXAILab/lumynax-longctx-prolong-512k-instruct
```

LumynaX Long-Context ProLong-512K Instruct

Legacy release · Outdated research artifact

This model card documents an early LumynaX experiment. It is no longer maintained, is not recommended for production, and does not represent the current capabilities, architecture, or safety standards of AbteeX AI Labs.

How infusion works

LumynaX Core is the core intelligence model. It governs the inference path and integrates selected open-source models as specialised execution layers.

Prompt  →  LumynaX Core  →  Infused model / MoE experts  →  LumynaX Core  →  Response

LumynaX infusion is the controlled composition of LumynaX Core with a compatible open-source model. Depending on the model family and deployment objective, the integration can operate in two ways:

Routed infusion — LumynaX Core directs inference through the selected model without modifying its weights.
MoE infusion — when required by the architecture, compatible model weights can be composed as specialised experts within a Mixture-of-Experts design.

In both cases, LumynaX Core remains the primary intelligence and orchestration layer, applying sovereignty controls, context, agentic planning, and inference optimisation around model execution. Infusion does not automatically imply a weight merge; each release manifest records the method used by that pack.

This release


Infused model	`princeton-nlp/Llama-3-8B-ProLong-512k-Instruct`
Infusion method	Routed runtime and identity integration
Weight composition	None — this pack preserves the source-model weights
Runtime	Transformers
Release	v0.1.0
Status	Outdated and retained for research provenance only

This package predates the current LumynaX Core implementation. Its included identity, runtime, or deployment wrappers are historical release components—not the complete modern LumynaX pipeline.

Archive access

The artifacts remain available for reproducibility. Before evaluation, verify checksums.sha256, inspect release_export_manifest.json, and review LICENSE.txt.

AbteeX AI Labs · Aotearoa New Zealand