Skip to content
Repo

Gemma-4-26B-A4B-it

from modal_training_gym.common.models.gemma4_26b_a4b import Gemma4_26B_A4B

Google Gemma-4-26B-A4B-it multimodal MoE model with 25.2B total and 3.8B active parameters.

Attributes

model_name str

Hugging Face repository ID. Default: "google/gemma-4-26B-A4B-it"

architecture ModelArchitecture | None

Megatron architecture parameters for this model. Default: ModelArchitecture(...)

response_parser Callable[[str], ParsedResponse] | None

Parser for generated text. Default: parse_gemma4_response

download() -> None

Download or materialize weights into the model volume.

parse_response(text: str) -> ParsedResponse

Parse model text with response_parser.

Without a configured parser, the model text becomes ParsedResponse.content.

Returns

Parsed model output.