Team Ai
Modelpublic

RedHatAI/Qwen3-30B-A3B-quantized.w4a16

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
7likes1.7kdownloads
Model Card

Qwen3-30B-A3B-quantized.w4a16

Model Overview

  • —Model Architecture: Qwen3ForCausalLM
  • —Input: Text
  • —Output: Text
  • —Model Optimizations:
  • —Weight quantization: INT4
  • —Intended Use Cases:
  • —Reasoning.
  • —Function calling.
  • —Subject matter experts via fine-tuning.
  • —Multilingual instruction following.
  • —Translation.
  • —Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws).
  • —Release Date: 05/05/2025
  • —Version: 1.0
  • —Model Developers: RedHat (Neural Magic)

Model Optimizations

This model was obtained by quantizing the weights of Qwen3-30B-A3B to INT4 data type. This optimization reduces the number of bits per parameter from 16 to 4, reducing the disk size and GPU memory requirements by approximately 75%.

Only the weights of the linear operators within transformers blocks are quantized. Weights are quantized using a symmetric per-group scheme, with group size 128. The GPTQ algorithm is applied for quantization, as implemented in the llm-compressor library.

Deployment

This model can be deployed efficiently using the vLLM backend, as shown in the example below.

python
from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "RedHatAI/Qwen3-30B-A3B-quantized.w4a16"
number_gpus = 1
sampling_params = SamplingParams(temperature=0.6, top_p=0.95, top_k=20, min_p=0, max_tokens=256)

messages = [
    {"role": "user", "content": prompt}
]

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [{"role": "user", "content": "Give me a short introduction to large language model."}]

prompts = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)

llm = LLM(model=model_id, tensor_parallel_size=number_gpus)

outputs = llm.generate(prompts, sampling_params)

generated_text = outputs[0].outputs[0].text
print(generated_text)

vLLM aslo supports OpenAI-compatible serving. See the documentation for more details.

Creation

<details> <summary>Creation details</summary> This model was created with llm-compressor by running the code snippet below.

python
  from llmcompressor.modifiers.quantization import GPTQModifier
  from llmcompressor.transformers import oneshot
  from transformers import AutoModelForCausalLM, AutoTokenizer
  
  # Load model
  model_stub = "Qwen/Qwen3-30B-A3B"
  model_name = model_stub.split("/")[-1]

  num_samples = 1024
  max_seq_len = 8192

  model = AutoModelForCausalLM.from_pretrained(model_stub)

  tokenizer = AutoTokenizer.from_pretrained(model_stub)

  def preprocess_fn(example):
    return {"text": tokenizer.apply_chat_template(example["messages"], add_generation_prompt=False, tokenize=False)}
  
  ds = load_dataset("neuralmagic/LLM_compression_calibration", split="train")
  ds = ds.map(preprocess_fn)

  # Configure the quantization algorithm and scheme
  recipe = GPTQModifier(
      ignore: ["lm_head", "re:.*gate$"]
      sequential_targets=["Qwen3DecoderLayer"],
      targets="Linear",
      scheme="W4A16",
      dampening_frac=0.01,
  )

  # Apply quantization
  oneshot(
      model=model,
      dataset=ds, 
      recipe=recipe,
      max_seq_length=max_seq_len,
      num_calibration_samples=num_samples,
  )
  
  # Save to disk in compressed-tensors format
  save_path = model_name + "-quantized.w4a16"
  model.save_pretrained(save_path)
  tokenizer.save_pretrained(save_path)
  print(f"Model and tokenizer saved to: {save_path}")

</details>

Evaluation

The model was evaluated on the OpenLLM leaderboard tasks (versions 1 and 2), using lm-evaluation-harness, and on reasoning tasks using lighteval. vLLM was used for all evaluations.

<details> <summary>Evaluation details</summary>

lm-evaluation-harness

  lm_eval \
    --model vllm \
    --model_args pretrained="RedHatAI/Qwen3-30B-A3B-quantized.w4a16",dtype=auto,gpu_memory_utilization=0.5,max_model_len=8192,enable_chunk_prefill=True,tensor_parallel_size=1 \
    --tasks openllm \
    --apply_chat_template\
    --fewshot_as_multiturn \
    --batch_size auto
  lm_eval \
    --model vllm \
    --model_args pretrained="RedHatAI/Qwen3-30B-A3B-quantized.w4a16",dtype=auto,gpu_memory_utilization=0.5,max_model_len=8192,enable_chunk_prefill=True,tensor_parallel_size=1 \
    --tasks mgsm \
    --apply_chat_template\
    --batch_size auto
  lm_eval \
    --model vllm \
    --model_args pretrained="RedHatAI/Qwen3-30B-A3B-quantized.w4a16",dtype=auto,gpu_memory_utilization=0.5,max_model_len=16384,enable_chunk_prefill=True,tensor_parallel_size=1 \
    --tasks leaderboard \
    --apply_chat_template\
    --fewshot_as_multiturn \
    --batch_size auto

lighteval

lightevalmodelarguments.yaml

yaml
  model_parameters:
    model_name: RedHatAI/Qwen3-30B-A3B-quantized.w4a16
    dtype: auto
    gpu_memory_utilization: 0.9
    max_model_length: 40960
    generation_parameters:
      temperature: 0.6
      top_k: 20
      min_p: 0.0
      top_p: 0.95
      max_new_tokens: 32768
  lighteval vllm \
    --model_args lighteval_model_arguments.yaml \
    --tasks lighteval|aime24|0|0 \
    --use_chat_template = true
  lighteval vllm \
    --model_args lighteval_model_arguments.yaml \
    --tasks lighteval|aime25|0|0 \
    --use_chat_template = true
  lighteval vllm \
    --model_args lighteval_model_arguments.yaml \
    --tasks lighteval|math_500|0|0 \
    --use_chat_template = true
  lighteval vllm \
    --model_args lighteval_model_arguments.yaml \
    --tasks lighteval|gpqa:diamond|0|0 \
    --use_chat_template = true
  lighteval vllm \
    --model_args lighteval_model_arguments.yaml \
    --tasks extended|lcb:codegeneration \
    --use_chat_template = true

</details>

Accuracy

<table> <tr> <th>Category </th> <th>Benchmark </th> <th>Qwen3-30B-A3B </th> <th>Qwen3-30B-A3B-quantized.w4a16<br>(this model) </th> <th>Recovery </th> </tr> <tr> <td rowspan="7" ><strong>OpenLLM v1</strong> </td> <td>MMLU (5-shot) </td> <td>77.67 </td> <td>76.11 </td> <td>98.00% </td> </tr> <tr> <td>ARC Challenge (25-shot) </td> <td>63.40 </td> <td>62.97 </td> <td>99.3% </td> </tr> <tr> <td>GSM-8K (5-shot, strict-match) </td> <td>87.26 </td> <td>86.66 </td> <td>99.3% </td> </tr> <tr> <td>Hellaswag (10-shot) </td> <td>54.33 </td> <td>54.76 </td> <td>100.8% </td> </tr> <tr> <td>Winogrande (5-shot) </td> <td>66.77 </td> <td>64.33 </td> <td>96.3% </td> </tr> <tr> <td>TruthfulQA (0-shot, mc2) </td> <td>56.27 </td> <td>54.76 </td> <td>97.3% </td> </tr> <tr> <td><strong>Average</strong> </td> <td><strong>67.62</strong> </td> <td><strong>66.60</strong> </td> <td><strong>98.5%</strong> </td> </tr> <tr> <td rowspan="7" ><strong>OpenLLM v2</strong> </td> <td>MMLU-Pro (5-shot) </td> <td>47.45 </td> <td>45.38 </td> <td>95.6% </td> </tr> <tr> <td>IFEval (0-shot) </td> <td>86.26 </td> <td>84.86 </td> <td>98.4% </td> </tr> <tr> <td>BBH (3-shot) </td> <td>34.81 </td> <td>28.12 </td> <td>80.8% </td> </tr> <tr> <td>Math-lvl-5 (4-shot) </td> <td>52.14 </td> <td>56.99 </td> <td>109.3% </td> </tr> <tr> <td>GPQA (0-shot) </td> <td>0.31 </td> <td>0.60 </td> <td>--- </td> </tr> <tr> <td>MuSR (0-shot) </td> <td>8.09 </td> <td>9.05 </td> <td>--- </td> </tr> <tr> <td><strong>Average</strong> </td> <td><strong>38.18</strong> </td> <td><strong>37.50</strong> </td> <td><strong>98.2%</strong> </td> </tr> <tr> <td><strong>Multilingual</strong> </td> <td>MGSM (0-shot) </td> <td>32.27 </td> <td>33,890 </td> <td>104.8% </td> </tr> <tr> <td rowspan="6" ><strong>Reasoning<br>(generation)</strong> </td> <td>AIME 2024 </td> <td>78.33 </td> <td>78.54 </td> <td>100.3% </td> </tr> <tr> <td>AIME 2025 </td> <td>71.46 </td> <td>70.31 </td> <td>98.4% </td> </tr> <tr> <td>GPQA diamond </td> <td>62.63 </td> <td>62.12 </td> <td>99.2% </td> </tr> <tr> <td>Math-lvl-5 </td> <td>97.60 </td> <td>97.20 </td> <td>99.6% </td> </tr> <tr> <td>LiveCodeBench </td> <td>60.66 </td> <td>58.75 </td> <td>96.9% </td> </tr> </table>