Instructions to use Brain2nd/NeuronSpark-V3-1.1B-SFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Brain2nd/NeuronSpark-V3-1.1B-SFT with Transformers:

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

pipe = pipeline("text-generation", model="Brain2nd/NeuronSpark-V3-1.1B-SFT", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Brain2nd/NeuronSpark-V3-1.1B-SFT", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Brain2nd/NeuronSpark-V3-1.1B-SFT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Brain2nd/NeuronSpark-V3-1.1B-SFT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Brain2nd/NeuronSpark-V3-1.1B-SFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Brain2nd/NeuronSpark-V3-1.1B-SFT

SGLang

How to use Brain2nd/NeuronSpark-V3-1.1B-SFT 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 "Brain2nd/NeuronSpark-V3-1.1B-SFT" \
    --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": "Brain2nd/NeuronSpark-V3-1.1B-SFT",
		"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 "Brain2nd/NeuronSpark-V3-1.1B-SFT" \
        --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": "Brain2nd/NeuronSpark-V3-1.1B-SFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Brain2nd/NeuronSpark-V3-1.1B-SFT with Docker Model Runner:
```
docker model run hf.co/Brain2nd/NeuronSpark-V3-1.1B-SFT
```

NeuronSpark V3 — 1.1B SFT (step 2000)

Architecture: SNN (Spiking Neural Network) decoder with PonderNet-style adaptive K time-step routing. Custom model, not a transformer variant.

Hidden dim D = 1024 · K_max = 12 · 24 layers
~1.24B parameters (model.safetensors ≈ 2.47 GB bf16)
Tokenizer vocab = 128387 (multilingual)
SFT step 2000, base = pretrain step 108000
Optimizer: AdamW + DeepSpeed ZeRO-2 (8 ranks)
SFT data: uniform-bucketed [1000, 2048] tokens, ~35% thinking / 65% non-thinking, multi-turn ZH/EN balanced (OpenThoughts, QwQ, Congliu, smoltalk2, WildChat-1M, no_robots)

Load (inference)

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("Brain2nd/NeuronSpark-V3-1.1B-SFT")
model = AutoModelForCausalLM.from_pretrained(
    "Brain2nd/NeuronSpark-V3-1.1B-SFT",
    trust_remote_code=True,
).cuda().eval()

msgs = [{"role": "user", "content": "What is the capital of France?"}]
text = tokenizer.apply_chat_template(msgs, tokenize=False,
                                     add_generation_prompt=True,
                                     enable_thinking=True)
ids = tokenizer(text, return_tensors="pt").input_ids.cuda()
out = model.generate_cached(input_ids=ids, max_new_tokens=512,
                            temperature=0.7, top_p=0.9, top_k=50, repetition_penalty=1.1)
print(tokenizer.decode(out[0, ids.shape[1]:], skip_special_tokens=False))

Resume SFT

deepspeed/ contains 8-rank ZeRO-2 sharded optimizer state. Use:

deepspeed --num_gpus=8 train_sft.py \
  --deepspeed_config ds_config.json \
  --resume ./ckpt_step2000

Files


`model.safetensors`	HF-format bf16 weights
`config.json` / `generation_config.json`	model config
`tokenizer.json` / `tokenizer_config.json`	tokenizer (vocab=128387)
`modeling_neuronspark.py` / `configuration_neuronspark.py`	custom arch (`trust_remote_code=True`)
`deepspeed/`	ZeRO-2 optimizer state (8 ranks)
`training_state.pth`	step / epoch / tokens_seen
`zero_to_fp32.py`	DeepSpeed helper

Downloads last month: 45

Safetensors

Model size

1B params

Tensor type

F32

BF16