Qwen3-14B-NVFP4

Model Overview

• Model Architecture: Qwen/Qwen3-14B
  • Input: Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: FP4
  • Activation quantization: FP4
• Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English.
• Release Date: 10/29/2025
• Version: 1.0

This model is a quantized version of Qwen/Qwen3-14B. It was evaluated on a several tasks to assess the its quality in comparison to the unquatized model.

Model Optimizations

This model was obtained by quantizing the weights and activations of Qwen/Qwen3-14B to FP4 data type, ready for inference with vLLM>=0.9.1 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 and activations of the linear operators within transformers blocks are quantized using LLM Compressor.

Deployment

Use with vLLM

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

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "Saktsant/Qwen3-14B-NVFP4"
number_gpus = 1

sampling_params = SamplingParams(temperature=0.6, top_p=0.9, max_tokens=256)

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
    {"role": "user", "content": "Who are you?"},
]

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

This model was created by applying LLM Compressor with calibration samples from UltraChat, as presented in the code snipet below.

from datasets import load_dataset
from transformers import AutoModelForCausalLM, AutoTokenizer
from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import QuantizationModifier
from llmcompressor.utils import dispatch_for_generation


MODEL_ID = "Qwen/Qwen3-14B"


# Load model and tokenizer
model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype="auto")
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
DATASET_ID = "HuggingFaceH4/ultrachat_200k"
DATASET_SPLIT = "train_sft"
NUM_CALIBRATION_SAMPLES = 512  
MAX_SEQUENCE_LENGTH = 2048
# Load dataset and preprocess
ds = load_dataset(DATASET_ID, split=f"{DATASET_SPLIT}[:{NUM_CALIBRATION_SAMPLES}]")
ds = ds.shuffle(seed=42)
def preprocess(example):
    return {
        "text": tokenizer.apply_chat_template(
            example["messages"],
            tokenize=False,
        )
    }
ds = ds.map(preprocess)
# Tokenize inputs
def tokenize(sample):
    return tokenizer(
        sample["text"],
        padding=False,
        max_length=MAX_SEQUENCE_LENGTH,
        truncation=True,
        add_special_tokens=False,
    )
ds = ds.map(tokenize, remove_columns=ds.column_names)
# Configure quantization
recipe = [
    QuantizationModifier(
        ignore=[
            "re:.*lm_head.*",
            "re:.*q_proj.*",
            "re:.*k_proj.*",
            "re:.*v_proj.*",
            "re:.*o_proj.*",
            "re:.*gate_proj.*",
            "re:.*up_proj.*",
            "re:.*down_proj.*",
        ],
        config_groups={
            "group_0": {
                "targets": ["Linear"],
                "weights": {
                    "num_bits": 4,
                    "type": "float",
                    "strategy": "tensor_group",
                    "group_size": 16,
                    "symmetric": True,
                    "observer": "minmax",
                },
                "input_activations": {
                    "num_bits": 4,
                    "type": "float",
                    "strategy": "tensor_group",
                    "group_size": 16,
                    "symmetric": True,
                    "dynamic": "local",
                    "observer": "minmax",
                },
            }
        },
    )
]
# Save directory
SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-NVFP4"
# Apply quantization
oneshot(
    model=model,
    dataset=ds,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
    output_dir=SAVE_DIR,
)
# Re-dispatch for generation (Accelerate handles device placement)
model = dispatch_for_generation(model)
print("\n\n")
print("========== SAMPLE GENERATION ==============")
# Prepare inputs with attention_mask
inputs = tokenizer("Hello my name is", return_tensors="pt")
inputs = {k: v.to("cuda") for k, v in inputs.items()}
# Generate
output = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(output[0]))
print("==========================================\n\n")
# Save compressed model and tokenizer
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)