Instructions to use inference-optimization/Qwen3-30B-A3B-RAZOR20 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use inference-optimization/Qwen3-30B-A3B-RAZOR20 with Transformers:

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

pipe = pipeline("text-generation", model="inference-optimization/Qwen3-30B-A3B-RAZOR20")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("inference-optimization/Qwen3-30B-A3B-RAZOR20")
model = AutoModelForCausalLM.from_pretrained("inference-optimization/Qwen3-30B-A3B-RAZOR20", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use inference-optimization/Qwen3-30B-A3B-RAZOR20 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "inference-optimization/Qwen3-30B-A3B-RAZOR20"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "inference-optimization/Qwen3-30B-A3B-RAZOR20",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/inference-optimization/Qwen3-30B-A3B-RAZOR20

SGLang

How to use inference-optimization/Qwen3-30B-A3B-RAZOR20 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 "inference-optimization/Qwen3-30B-A3B-RAZOR20" \
    --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": "inference-optimization/Qwen3-30B-A3B-RAZOR20",
		"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 "inference-optimization/Qwen3-30B-A3B-RAZOR20" \
        --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": "inference-optimization/Qwen3-30B-A3B-RAZOR20",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use inference-optimization/Qwen3-30B-A3B-RAZOR20 with Docker Model Runner:
```
docker model run hf.co/inference-optimization/Qwen3-30B-A3B-RAZOR20
```

Qwen3-30B-A3B-RAZOR20

Qwen/Qwen3-30B-A3B with 20% of its MoE experts structurally pruned with RAZOR (PRUNE method), produced with llmcompressor.

Every one of the 48 MoE layers keeps its 102 most replaceable-critical experts of the original 128 (26 pruned per layer, 4896 of 6144 retained)
Expert selection is uniform across layers: each layer's 26 lowest-RAZOR-saliency experts were removed, and the routers were sliced to match
config.json records the new expert count (num_experts: 102); expert indices are remapped contiguously
All remaining weights (including attention, embeddings and the shared dense layers) are untouched BF16

RAZOR scores each expert by the damage its deletion would do to the routed mixture output, using forward computation only (no gradients or recovery training); see the RAZOR expert pruning example for the calibration and pruning scripts.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "inference-optimization/Qwen3-30B-A3B-RAZOR20", torch_dtype="bfloat16"
)
tokenizer = AutoTokenizer.from_pretrained("inference-optimization/Qwen3-30B-A3B-RAZOR20")

Downloads last month: 364

Safetensors

Model size

25B params

Tensor type

BF16

Model tree for inference-optimization/Qwen3-30B-A3B-RAZOR20

Base model

Qwen/Qwen3-30B-A3B-Base

Finetuned

Qwen/Qwen3-30B-A3B

Finetuned

(90)

this model