Instructions to use FINAL-Bench/AETHER-7B-7Attn-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FINAL-Bench/AETHER-7B-7Attn-base with Transformers:

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

pipe = pipeline("text-generation", model="FINAL-Bench/AETHER-7B-7Attn-base", device_map="auto")

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("FINAL-Bench/AETHER-7B-7Attn-base", dtype="auto", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use FINAL-Bench/AETHER-7B-7Attn-base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FINAL-Bench/AETHER-7B-7Attn-base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/AETHER-7B-7Attn-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/FINAL-Bench/AETHER-7B-7Attn-base

SGLang

How to use FINAL-Bench/AETHER-7B-7Attn-base 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 "FINAL-Bench/AETHER-7B-7Attn-base" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/AETHER-7B-7Attn-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "FINAL-Bench/AETHER-7B-7Attn-base" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/AETHER-7B-7Attn-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use FINAL-Bench/AETHER-7B-7Attn-base with Docker Model Runner:
```
docker model run hf.co/FINAL-Bench/AETHER-7B-7Attn-base
```

AETHER-7B-7Attn-base

Open-weight release — model weights only.

To our knowledge, the first foundation model to implement seven heterogeneous attention mechanisms within a single architecture.

Attention layout

Seven distinct attention mechanisms (A–G) are distributed across 49 layers in a 7×7 Latin square — every mechanism appears exactly once in each row and each column, so no single mechanism dominates any depth band.

Mechanism identities and the assignment recipe are proprietary and not disclosed.

Model


Parameters	6.59B total / ~3B active (Mixture-of-Experts)
Type	Decoder-only language model, heterogeneous multi-attention architecture
Precision	bfloat16
Languages	Korean, English
Stage	Pretrained base (no instruction tuning)
Developer	VIDRAFT

Unlike conventional models that repeat a single attention mechanism across all layers, AETHER-7B-7Attn composes seven distinct attention mechanisms inside one network. This heterogeneous design is the defining characteristic of the AETHER architecture.

Release scope

This repository distributes model weights only.

The following are proprietary and not disclosed:

the attention composition recipe (which mechanisms, and how they are arranged)
layer layout and architectural configuration details
modeling / inference implementation code
training data, data mixture, and hyperparameters

Loading these weights requires the corresponding architecture implementation, which is not distributed in this repository. For access or licensing inquiries, please contact VIDRAFT.

Intended use

Released for research, evaluation and benchmarking of the released weights. This is a base (pretrained) model — it continues text rather than following instructions.

License

Apache-2.0 — the released weights are free to use, modify and redistribute under the terms of the Apache License 2.0.

Note that the architecture implementation, attention composition recipe and training pipeline are not part of this release and remain proprietary to VIDRAFT.

VIDRAFT · Sovereign AI

Downloads last month: -

Safetensors

Model size

7B params

Tensor type

BF16

Collection including FINAL-Bench/AETHER-7B-7Attn-base

'Aether' Foundation Model

Collection

6 items • Updated 1 day ago • 30