---
tags:
- vllm
- vision
- w4a16
license: gemma
base_model: google/gemma-3-4b-it
library_name: transformers
---
# gemma-3-4b-it-quantized.w4a16
## Model Overview
- **Model Architecture:** google/gemma-3-4b-it
- **Input:** Vision-Text
- **Output:** Text
- **Model Optimizations:**
- **Weight quantization:** INT4
- **Activation quantization:** FP16
- **Release Date:** 6/4/2025
- **Version:** 1.0
- **Model Developers:** RedHatAI
Quantized version of [google/gemma-3-4b-it](https://huggingface.co/google/gemma-3-4b-it).
### Model Optimizations
This model was obtained by quantizing the weights of [google/gemma-3-4b-it](https://huggingface.co/google/gemma-3-4b-it) to INT4 data type, ready for inference with vLLM >= 0.8.0.
## 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 vllm.assets.image import ImageAsset
from transformers import AutoProcessor
# Define model name once
model_name = "RedHatAI/gemma-3-4b-it-quantized.w4a16"
# Load image and processor
image = ImageAsset("cherry_blossom").pil_image.convert("RGB")
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
# Build multimodal prompt
chat = [
{"role": "user", "content": [{"type": "image"}, {"type": "text", "text": "What is the content of this image?"}]},
{"role": "assistant", "content": []}
]
prompt = processor.apply_chat_template(chat, add_generation_prompt=True)
# Initialize model
llm = LLM(model=model_name, trust_remote_code=True)
# Run inference
inputs = {"prompt": prompt, "multi_modal_data": {"image": [image]}}
outputs = llm.generate(inputs, SamplingParams(temperature=0.2, max_tokens=64))
# Display result
print("RESPONSE:", outputs[0].outputs[0].text)
```
vLLM also supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details.
## Creation
This model was created with [llm-compressor](https://github.com/vllm-project/llm-compressor) by running the code snippet below:
Model Creation Code
```python
import base64
from io import BytesIO
import torch
from datasets import load_dataset
from transformers import AutoProcessor, Gemma3ForConditionalGeneration
from llmcompressor.modifiers.quantization import GPTQModifier
from llmcompressor.transformers import oneshot
# Load model.
model_id = "google/gemma-3-4b-it"
model = Gemma3ForConditionalGeneration.from_pretrained(
model_id,
device_map="auto",
torch_dtype="auto",
)
processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
# Oneshot arguments
DATASET_ID = "neuralmagic/calibration"
DATASET_SPLIT = {"LLM": "train[:512]"}
NUM_CALIBRATION_SAMPLES = 512
MAX_SEQUENCE_LENGTH = 2048
# Load dataset and preprocess.
ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
ds = ds.shuffle(seed=42)
dampening_frac=0.05
def data_collator(batch):
assert len(batch) == 1, "Only batch size of 1 is supported for calibration"
item = batch[0]
collated = {}
import torch
for key, value in item.items():
if isinstance(value, torch.Tensor):
collated[key] = value.unsqueeze(0)
elif isinstance(value, list) and isinstance(value[0][0], int):
# Handle tokenized inputs like input_ids, attention_mask
collated[key] = torch.tensor(value)
elif isinstance(value, list) and isinstance(value[0][0], float):
# Handle possible float sequences
collated[key] = torch.tensor(value)
elif isinstance(value, list) and isinstance(value[0][0], torch.Tensor):
# Handle batched image data (e.g., pixel_values as [C, H, W])
collated[key] = torch.stack(value) # -> [1, C, H, W]
elif isinstance(value, torch.Tensor):
collated[key] = value
else:
print(f"[WARN] Unrecognized type in collator for key={key}, type={type(value)}")
return collated
# Recipe
recipe = [
GPTQModifier(
targets="Linear",
ignore=["re:.*lm_head.*", "re:.*embed_tokens.*", "re:vision_tower.*", "re:multi_modal_projector.*"],
sequential_update=True,
sequential_targets=["Gemma3DecoderLayer"],
dampening_frac=dampening_frac,
)
]
SAVE_DIR=f"{model_id.split('/')[1]}-quantized.w4a16"
# Perform oneshot
oneshot(
model=model,
tokenizer=model_id,
dataset=ds,
recipe=recipe,
max_seq_length=MAX_SEQUENCE_LENGTH,
num_calibration_samples=NUM_CALIBRATION_SAMPLES,
trust_remote_code_model=True,
data_collator=data_collator,
output_dir=SAVE_DIR
)
```
## Evaluation
The model was evaluated using [lm_evaluation_harness](https://github.com/neuralmagic/lm-evaluation-harness) for OpenLLM v1 text benchmark. The evaluations were conducted using the following commands:
Evaluation Commands
### OpenLLM v1
```
lm_eval \
--model vllm \
--model_args pretrained="",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True,enforce_eager=True \
--tasks openllm \
--batch_size auto
```
### Accuracy
| Category |
Metric |
google/gemma-3-4b-it |
RedHatAI/gemma-3-4b-it-quantized.w4a16 |
Recovery (%) |
| OpenLLM V1 |
ARC Challenge |
56.57% |
56.57% |
100.00% |
| GSM8K |
76.12% |
72.33% |
95.02% |
| Hellaswag |
74.96% |
73.35% |
97.86% |
| MMLU |
58.38% |
56.33% |
96.49% |
| Truthfulqa (mc2) |
51.87% |
50.81% |
97.96% |
| Winogrande |
70.32% |
68.82% |
97.87%% |
| Average Score |
64.70% |
63.04% |
97.42% |
| Vision Evals |
MMMU (val) |
39.89% |
40.11% |
100.55% |
| ChartQA |
50.76% |
49.32% |
97.16% |
| Average Score |
45.33% |
44.72% |
98.86% |