Instructions to use Subject-Emu-5259/NeuralAI-Air-135M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Subject-Emu-5259/NeuralAI-Air-135M with Transformers:

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

pipe = pipeline("text-generation", model="Subject-Emu-5259/NeuralAI-Air-135M")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Subject-Emu-5259/NeuralAI-Air-135M", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Subject-Emu-5259/NeuralAI-Air-135M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Subject-Emu-5259/NeuralAI-Air-135M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Subject-Emu-5259/NeuralAI-Air-135M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Subject-Emu-5259/NeuralAI-Air-135M

SGLang

How to use Subject-Emu-5259/NeuralAI-Air-135M 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 "Subject-Emu-5259/NeuralAI-Air-135M" \
    --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": "Subject-Emu-5259/NeuralAI-Air-135M",
		"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 "Subject-Emu-5259/NeuralAI-Air-135M" \
        --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": "Subject-Emu-5259/NeuralAI-Air-135M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Subject-Emu-5259/NeuralAI-Air-135M with Docker Model Runner:
```
docker model run hf.co/Subject-Emu-5259/NeuralAI-Air-135M
```

NeuralAI Air 135M

A compact, custom-built decoder-only language model created for the NeuralAI ecosystem. Designed to be small, fast, and locally runnable while still useful for inference on commodity hardware.

Model Details

Property	Value
Architecture	NeuralAI-Air (custom decoder-only Transformer)
Parameters	~135M
Vocabulary size	32,000
Hidden size	768
Layers	15
Attention heads	12
Key/Value heads	2 (GQA)
Intermediate size	2,560
Max position embeddings	2,048
Tie word embeddings	true
Torch dtype	float32

Tokenizer

The tokenizer files (tokenizer.json, tokenizer_config.json) use a GPT-2/BPE-style tokenizer with a 32,000-token vocabulary.

Special tokens:

bos_token: <s>
eos_token: </s>
pad_token: <pad>

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("Subject-Emu-5259/NeuralAI-Air-135M")
tokenizer = AutoTokenizer.from_pretrained("Subject-Emu-5259/NeuralAI-Air-135M")

NeuralAI Model Family

Model	Role	Repo
NeuralAI	Main production DPO adapter (360M, default)	`Subject-Emu-5259/NeuralAI`
NeuralAI-Air-135M	This repo — compact base model	here
NeuralAI-Air-135M-SFT	Supervised fine-tune of this base model	`Subject-Emu-5259/NeuralAI-Air-135M-SFT`

Family collection: https://huggingface.co/collections/Subject-Emu-5259/neuralai-model-family-6a66ee29c7c5f26e044dee3c
GitHub: https://github.com/Subject-Emu-5259/NeuralAI
Live demo: https://neuralai-web-ui-deandrewharris.zocomputer.io

Downloads last month: 143

Model tree for Subject-Emu-5259/NeuralAI-Air-135M

Finetunes

1 model

Collection including Subject-Emu-5259/NeuralAI-Air-135M

NeuralAI Model Family

Collection

DPO LoRA, NeuralAI-Air 135M base, and 135M-SFT models by De'Andrew Preston Harris. • 3 items • Updated about 8 hours ago