RedHatAI/Qwen3-14B-NVFP4
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
main
Qwen3-14B-NVFP4/README.md
ModelHub XC ae6b7779cf 初始化项目，由ModelHub XC社区提供模型 
Model: RedHatAI/Qwen3-14B-NVFP4
Source: Original Platform
2026-10-23 08:01:13 +08:00

396 lines
9.9 KiB
Markdown
Raw Permalink Blame History

---
tags:
- fp4
- vllm
language:
- en
- de
- fr
- it
- pt
- hi
- es
- th
pipeline_tag: text-generation
license: apache-2.0
base_model: Qwen/Qwen3-14B
---
# 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
- **Model Developers:** RedHatAI
This model is a quantized version of [Qwen/Qwen3-14B](https://huggingface.co/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](https://huggingface.co/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](https://github.com/vllm-project/llm-compressor).
## Deployment
### Use with vLLM
This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend, as shown in the example below.
```python
from vllm import LLM, SamplingParams
from transformers import AutoTokenizer
model_id = "RedHatAI/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](https://docs.vllm.ai/en/latest/) for more details.
## Creation
This model was created by applying [LLM Compressor with calibration samples from UltraChat](https://github.com/vllm-project/llm-compressor/blob/main/examples/quantization_w4a4_fp4/llama3_example.py), as presented in the code snipet below.
<details>
```python
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.
model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype="auto")
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
DATASET_ID = "HuggingFaceH4/ultrachat_200k"
DATASET_SPLIT = "train_sft"
# Select number of samples. 512 samples is a good place to start.
# Increasing the number of samples can improve accuracy.
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 the quantization algorithm and scheme.
# In this case, we:
# * quantize the weights to fp4 with per group 16 via ptq
# * calibrate a global_scale for activations, which will be used to
# quantize activations to fp4 on the fly
recipe = [
QuantizationModifier(
ignore=["re:.*lm_head.*"],
config_groups={
"group_0": {
"targets": ["Linear"],
"weights": {
"num_bits": 4,
"type": "float",
"strategy": "tensor_group",
"group_size": 16,
"symmetric": True,
"observer": "mse",
},
"input_activations": {
"num_bits": 4,
"type": "float",
"strategy": "tensor_group",
"group_size": 16,
"symmetric": True,
"dynamic": "local",
"observer": "minmax",
},
}
},
)
]
# Save to disk in compressed-tensors format.
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,
)
print("\n\n")
print("========== SAMPLE GENERATION ==============")
dispatch_for_generation(model)
input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to("cuda")
output = model.generate(input_ids, max_new_tokens=100)
print(tokenizer.decode(output[0]))
print("==========================================\n\n")
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)
```
</details>
## Evaluation
This model was evaluated on the well-known OpenLLM v1, OpenLLM v2 and HumanEval_64 benchmarks using [lm-evaluation-harness](https://github.com/neuralmagic/lm-evaluation-harness). The Reasoning evals were done using [ligheval](https://github.com/neuralmagic/lighteval).
### Accuracy
<table>
<thead>
<tr>
<th>Category</th>
<th>Metric</th>
<th>Qwen3-14B</th>
<th>Qwen3-14B-NVFP4 (this model)</th>
<th>Recovery</th>
</tr>
</thead>
<tbody>
<!-- OpenLLM V1 -->
<tr>
<td rowspan="7"><b>OpenLLM V1</b></td>
<td>arc_challenge</td>
<td>67.32</td>
<td>67.06</td>
<td>99.61</td>
</tr>
<tr>
<td>gsm8k</td>
<td>88.70</td>
<td>88.25</td>
<td>99.49</td>
</tr>
<tr>
<td>hellaswag</td>
<td>79.62</td>
<td>78.24</td>
<td>98.27</td>
</tr>
<tr>
<td>mmlu</td>
<td>78.86</td>
<td>77.23</td>
<td>97.93</td>
</tr>
<tr>
<td>truthfulqa_mc2</td>
<td>58.59</td>
<td>58.49</td>
<td>99.83</td>
</tr>
<tr>
<td>winogrande</td>
<td>73.72</td>
<td>73.80</td>
<td>100.11</td>
</tr>
<tr>
<td><b>Average</b></td>
<td><b>74.47</b></td>
<td><b>73.85</b></td>
<td><b>99.16</b></td>
</tr>
<tr>
<td rowspan="7"><b>OpenLLM V2</b></td>
<td>BBH (3-shot)</td>
<td>59.45</td>
<td>56.78</td>
<td>95.51</td>
</tr>
<tr>
<td>MMLU-Pro (5-shot)</td>
<td>44.39</td>
<td>41.15</td>
<td>92.70</td>
</tr>
<tr>
<td>MuSR (0-shot)</td>
<td>38.62</td>
<td>37.83</td>
<td>97.95</td>
</tr>
<tr>
<td>IFEval (0-shot)</td>
<td>89.45</td>
<td>90.41</td>
<td>101.07</td>
</tr>
<tr>
<td>GPQA (0-shot)</td>
<td>27.43</td>
<td>26.59</td>
<td>96.94</td>
</tr>
<tr>
<td>Math-|v|-5 (4-shot)</td>
<td>57.33</td>
<td>53.40</td>
<td>93.14</td>
</tr>
<tr>
<td><b>Average</b></td>
<td><b>52.78</b></td>
<td><b>51.03</b></td>
<td><b>96.68</b></td>
</tr>
<tr>
<td rowspan="1"><b>Coding</b></td>
<td>HumanEval_64 pass@2</td>
<td>90.74</td>
<td>89.87</td>
<td>99.04</td>
</tr>
<tr>
<td rowspan="4"><b>Reasoning</b></td>
<td>AIME24 (0-shot)</td>
<td>75.86</td>
<td>65.52</td>
<td>86.34</td>
</tr>
<tr>
<td>AIME25 (0-shot)</td>
<td>68.97</td>
<td>65.52</td>
<td>95.00</td>
</tr>
<tr>
<td>GPQA (Diamond, 0-shot)</td>
<td>64.97</td>
<td>60.40</td>
<td>93.00</td>
</tr>
<tr>
<td><b>Average</b></td>
<td><b>69.93</b></td>
<td><b>63.81</b></td>
<td><b>91.45</b></td>
</tr>
</tbody>
</table>
### Reproduction
The results were obtained using the following commands:
<details>
#### OpenLLM v1
```
lm_eval \
--model vllm \
--model_args pretrained="RedHatAI/Qwen3-14B-NVFP4",dtype=auto,max_model_len=4096,tensor_parallel_size=2,enable_chunked_prefill=True,enforce_eager=True\
--apply_chat_template \
--fewshot_as_multiturn \
--tasks openllm \
--batch_size auto
```
#### OpenLLM v2
```
lm_eval \
--model vllm \
--model_args pretrained="RedHatAI/Qwen3-14B-NVFP4",dtype=auto,max_model_len=4096,tensor_parallel_size=2,enable_chunked_prefill=True,enforce_eager=True\
--apply_chat_template \
--fewshot_as_multiturn \
--tasks leaderboard \
--batch_size auto
```
#### HumanEval_64
```
lm_eval \
--model vllm \
--model_args pretrained="RedHatAI/Qwen3-14B-NVFP4",dtype=auto,max_model_len=4096,tensor_parallel_size=2,enable_chunked_prefill=True,enforce_eager=True\
--apply_chat_template \
--fewshot_as_multiturn \
--tasks humaneval_64_instruct \
--batch_size auto
```
#### LightEval
```
# --- model_args.yaml ---
cat > model_args.yaml <<'YAML'
model_parameters:
model_name: "RedHatAI/Qwen3-14B-NVFP4"
dtype: auto
gpu_memory_utilization: 0.9
tensor_parallel_size: 2
max_model_length: 40960
generation_parameters:
seed: 42
temperature: 0.6
top_k: 20
top_p: 0.95
min_p: 0.0
max_new_tokens: 32768
YAML
lighteval vllm model_args.yaml \
"lighteval|aime24|0,lighteval|aime25|0,lighteval|gpqa:diamond|0" \
--max-samples -1 \
--output-dir out_dir
```
</details>
Reference in New Issue View Git Blame Copy Permalink