You can deploy this model and start earning money today!

kerasformers/
gemma-3n-e4b-it

Name: gemma-3n-e4b-it
Availability: InStock

google/gemma-3n-E4B-it

finetune

gemma

Views

No views yet

See our collection for all versions of Gemma 3n.

Run Gemma 3n with Keras 3: JAX, PyTorch, or TensorFlow

GitHub

zeromodels/gemma-3n-e4b-it

Gemma 3n is Google's on-device multimodal (image + audio + text) model. Its decoder layers several on-device innovations on the Gemma shape: AltUp (alternating updates over parallel hidden streams), LAuReL (learned augmented residuals), MatFormer (nested per-layer widths), Per-Layer Embeddings, and activation sparsity, with tail KV-sharing and a 5:1 sliding/global attention schedule. Vision is a MobileNet-V5 encoder and audio a USM conformer, both feeding soft tokens into the decoder. Base checkpoints are for completion; -it variants are instruction-tuned.

For more details, see Google's original model card.

Pure-Keras 3 conversion of google/gemma-3n-E4B-it for zeromodels. One implementation runs unmodified on TensorFlow / Torch / JAX.

This is an instruction-tuned checkpoint: use the chat template via Gemma3nProcessor.

✨ Quick start

Text-only

python

import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from zeromodels.models.gemma3n import Gemma3nTextGenerate, Gemma3nTokenizer

model = Gemma3nTextGenerate.from_weights("zeromodels/gemma-3n-e4b-it")
tokenizer = Gemma3nTokenizer.from_weights("zeromodels/gemma-3n-e4b-it")

inputs = tokenizer([{"role": "user", "content": "Hello, who are you?"}])
outputs = model.generate(**inputs, max_new_tokens=64)
print(tokenizer.decode(outputs[0]))

Image + text

python

import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from PIL import Image
from zeromodels.models.gemma3n import Gemma3nConditionalGenerate, Gemma3nProcessor

model = Gemma3nConditionalGenerate.from_weights("zeromodels/gemma-3n-e4b-it")
processor = Gemma3nProcessor.from_weights("zeromodels/gemma-3n-e4b-it")

inputs = processor(conversation=[
    {"role": "user", "content": [
        {"type": "image", "image": Image.open("cat.jpg")},
        {"type": "text", "text": "Describe this image in one sentence."},
    ]}
])
outputs = model.generate(**inputs, max_new_tokens=64)
print(processor.decode(outputs[0]))

All Gemma 3n variants load the same way with from_weights("zeromodels/<variant>"):

Variant	Hub	Type
`gemma-3n-e2b`	`zeromodels/gemma-3n-e2b`	multimodal / base
`gemma-3n-e2b-it`	`zeromodels/gemma-3n-e2b-it`	multimodal / instruct
`gemma-3n-e4b`	`zeromodels/gemma-3n-e4b`	multimodal / base
`gemma-3n-e4b-it`	`zeromodels/gemma-3n-e4b-it`	multimodal / instruct

Tips

Set KERAS_BACKEND before importing Keras / zeromodels.
Loads in bfloat16 by default (the weights are bf16). Pass load_dtype="float32" for full precision, or quantization="int8" to shrink further.
Gemma 3n is audio-capable too: pass audio content items in the conversation to transcribe / reason over speech.
See Gemma 3n docs and Loading Weights.
Upstream safetensors still work via the hf: prefix, e.g. Gemma3nConditionalGenerate.from_weights("hf:google/gemma-3n-E4B-it").

Special Thanks

A huge thank you to the Google Gemma authors for creating and releasing these models.

License: Gemma (gated). Accept the license on the upstream Hub card before downloading.

kerasformers/gemma-3n-e4b-it