Instructions to use promzeus/gh0stx-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use promzeus/gh0stx-bf16 with Transformers:

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

pipe = pipeline("text-generation", model="promzeus/gh0stx-bf16")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("promzeus/gh0stx-bf16")
model = AutoModelForCausalLM.from_pretrained("promzeus/gh0stx-bf16")
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 promzeus/gh0stx-bf16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "promzeus/gh0stx-bf16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "promzeus/gh0stx-bf16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/promzeus/gh0stx-bf16

SGLang

How to use promzeus/gh0stx-bf16 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 "promzeus/gh0stx-bf16" \
    --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": "promzeus/gh0stx-bf16",
		"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 "promzeus/gh0stx-bf16" \
        --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": "promzeus/gh0stx-bf16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use promzeus/gh0stx-bf16 with Docker Model Runner:
```
docker model run hf.co/promzeus/gh0stx-bf16
```

gh0stx-bf16

Expert-pruned bf16 derivative of Qwen3.5-MoE. Full-precision weights — portable to any CUDA stack, quantize to whatever your hardware needs.

Specs

Base: Qwen3.5-MoE (A17B active), expert-pruned to ~141B total params
Format: bf16, ~263 GB (16 shards)
Context: 128K (native up to 256K)
Optional acceleration: MTP head (qwen3_5_mtp) → ~1.7× tokens/s (see below)

Hardware

bf16 needs ~280 GB VRAM → multi-GPU. Typical setups:

8× H100/A100 80GB (TP=8) — runs bf16 directly.
Fewer / smaller cards — quantize first: AWQ / GPTQ (INT4) ≈ ~75–90 GB, or FP8 (H100) ≈ ~140 GB. Use llm-compressor or AutoAWQ.
NVFP4 (single Blackwell card) → see the sibling promzeus/gh0stx-nvfp4.

Serve (vLLM, bf16, tensor-parallel)

vllm serve promzeus/gh0stx-bf16 \
  --served-model-name gh0stx --tensor-parallel-size 8 \
  --trust-remote-code --max-model-len 131072 \
  --gpu-memory-utilization 0.90 \
  --enable-chunked-prefill --enable-prefix-caching \
  --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder

Adjust --tensor-parallel-size to your GPU count.

Enable MTP speculative decoding (optional, ~1.7×)

The MTP head is bf16 and runs on any GPU (incl. H100), not just Blackwell.

Add model-mtp-grafted.safetensors to the model folder (available in the gh0stx-nvfp4 sibling repo — the head weights are format-agnostic).
In config.json set "mtp_num_hidden_layers": 1.
In model.safetensors.index.json add the mtp.* keys → model-mtp-grafted.safetensors (and bump total_size).

Serve with:

--speculative-config '{"method":"qwen3_5_mtp","num_speculative_tokens":3}'

num_speculative_tokens=3 is the sweet spot.

Notes

KV cache: keep bf16 (fp8 KV is unstable with the MTP + linear-attention path).
Thinking mode is ON by default (Qwen3.5). Disable per request via chat_template_kwargs: {"enable_thinking": false}.
Requires a vLLM build with Qwen3.5-MoE support.

Sibling

promzeus/gh0stx-nvfp4 — NVFP4 (~84 GB), single NVIDIA Blackwell card (GB10 / B200 / RTX 50xx), MTP pre-wired.

Downloads last month: 616

Safetensors

Model size

141B params

Tensor type

BF16

Model tree for promzeus/gh0stx-bf16

Base model

Qwen/Qwen3.5-397B-A17B

Finetuned

(38)

this model