Hugging Face's logo

Transformers documentation

Gemma4 Unified

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v5.14.0).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

This model was contributed to Hugging Face Transformers on 2026-06-03.

Gemma4 Unified

Overview

Gemma 4 12B Unified is an encoder-free multimodal model with pretrained and instruction-tuned variants. Unlike standard Gemma 4, which uses dedicated encoder towers, Gemma 4 12B Unified projects raw inputs directly into the language model’s embedding space through lightweight linear pipelines. This results in a simpler architecture while maintaining strong multimodal performance.

Key differences from standard Gemma 4:

  • No Vision Tower: Raw pixel patches are projected directly into LM space via a Dense + LayerNorm pipeline with factorized 2D positional embeddings, replacing the vision encoder.
  • No Audio Tower: Raw 16 kHz waveform samples are chunked into fixed-length frames and projected through a simple RMSNorm → Linear pipeline, replacing the mel spectrogram + Conformer encoder.
  • Shared Multimodal Pipeline: Both vision and audio use the same Gemma4UnifiedMultimodalEmbedder (RMSNorm → Linear) for the final projection to text hidden space.

You can find the original Gemma 4 12B Unified checkpoints under the Gemma 4 release.

Encoder-Free Vision Pipeline

The key architectural difference from standard Gemma 4 is the removal of the vision encoder tower. Instead, Gemma 4 12B Unified processes images through a lightweight pipeline:

  1. Patchification: Images are split into 16×16 pixel patches
  2. Patch Merging: Adjacent 3×3 patches are merged into 48×48 model patches, each with 48² × 3 = 6,912 raw pixel channels
  3. Projection: LayerNorm → Dense → LayerNorm projects each merged patch into the LM embedding dimension
  4. Positional Embedding: Factorized 2D positional embeddings are added (separate learned embeddings for x and y axes, summed together)
  5. Final Norm: A final LayerNorm is applied
  6. Multimodal Embedder: RMSNorm → Linear projects to the text hidden size

Like standard Gemma 4, the model processes images of different sizes using a fixed-budget number of tokens. The same constraints apply:

  • The total number of pixels must fit within a patch budget
  • Both height and width must be divisible by 48 (= patch size 16 × pooling kernel 3)

Gemma 4 12B Unified does not apply mean/std normalization. The model’s own patch embedding layer handles the final scaling internally.

The number of soft tokens per image is configurable. The supported options and default (280 soft tokens) are:

Soft TokensPatches (before pooling)Approx. Image Area
70630~161K pixels
1401,260~323K pixels
2802,520~645K pixels
5605,040~1.3M pixels
1,12010,080~2.6M pixels

Encoder-Free Audio Pipeline

The audio pipeline is similarly simplified. Instead of computing mel spectrograms and processing them through a Conformer encoder, raw 16 kHz waveform samples are:

  1. Chunked into fixed-length frames of 640 samples each (40ms per frame at 16 kHz)
  2. Projected directly through RMSNorm → Linear via the shared Gemma4UnifiedMultimodalEmbedder

Since there is no downsampling, the number of output soft tokens equals the number of input frames: ceil(num_samples / 640).

Usage examples

The example below demonstrates how to generate text based on an image and an audio sample with Pipeline or the AutoModel class.

Pipeline
AutoModel
from transformers import pipeline


pipe = pipeline(
    task="any-to-any",
    model="google/gemma-4-12B-it",
)

image_messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
            },
            {
                "type": "text",
                "text": "What is shown in this image?"
            }
        ]
    }
]

image_output = pipe(image_messages, return_full_text=False)
print(image_output[0]["generated_text"])

audio_messages = [
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Please transcribe the following audio:"},
            {
                "type": "audio",
                "url": "https://huggingface.co/datasets/eustlb/audio-samples/resolve/main/bcn_weather.mp3",
            },
        ],
    }
]

audio_output = pipe(audio_messages, return_full_text=False)
print(audio_output[0]["generated_text"])

Gemma4UnifiedAudioConfig

class transformers.Gemma4UnifiedAudioConfig

< >

( transformers_version: str | None = Nonearchitectures: list[str] | None = Noneoutput_hidden_states: bool | None = Falsereturn_dict: bool | None = Truedtype: typing.Union[str, ForwardRef('torch.dtype'), NoneType] = Nonechunk_size_feed_forward: int = 0is_encoder_decoder: bool = Falseid2label: dict[int, str] | dict[str, str] | None = Nonelabel2id: dict[str, int] | dict[str, str] | None = Noneproblem_type: typing.Optional[typing.Literal['regression', 'single_label_classification', 'multi_label_classification']] = Noneaudio_embed_dim: int = 640rms_norm_eps: float = 1e-06initializer_range: float = 0.02 )

Parameters

  • audio_embed_dim (int, defaults to 640) — Dimension of audio features input to the multimodal embedder. Each audio soft token is a raw waveform frame of audio_samples_per_token samples, so audio_embed_dim == audio_samples_per_token.
  • rms_norm_eps (float, optional, defaults to 1e-06) — The epsilon used by the rms normalization layers.
  • initializer_range (float, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

This is the configuration class to store the configuration of a Gemma4UnifiedModel. It is used to instantiate a Gemma4 Unified model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the google/gemma-4-12B-it

Configuration objects inherit from PreTrainedConfig and can be used to control the model outputs. Read the documentation from PreTrainedConfig for more information.

Gemma4UnifiedConfig

class transformers.Gemma4UnifiedConfig

< >

( transformers_version: str | None = Nonearchitectures: list[str] | None = Noneoutput_hidden_states: bool | None = Falsereturn_dict: bool | None = Truedtype: typing.Union[str, ForwardRef('torch.dtype'), NoneType] = Nonechunk_size_feed_forward: int = 0is_encoder_decoder: bool = Falseid2label: dict[int, str] | dict[str, str] | None = Nonelabel2id: dict[str, int] | dict[str, str] | None = Noneproblem_type: typing.Optional[typing.Literal['regression', 'single_label_classification', 'multi_label_classification']] = Nonetext_config: transformers.models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedTextConfig | dict[str, typing.Any] | None = Nonevision_config: transformers.models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedVisionConfig | dict[str, typing.Any] | None = Noneaudio_config: transformers.models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedAudioConfig | dict[str, typing.Any] | None = Noneboi_token_id: int | None = 255999eoi_token_id: int | None = 258882image_token_id: int | None = 258880video_token_id: int | None = 258884boa_token_id: int | None = 256000eoa_token_index: int | None = 258883audio_token_id: int | None = 258881initializer_range: float | None = 0.02tie_word_embeddings: bool = True )

Parameters

  • text_config (Union[~models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedTextConfig, dict[str, Any]], optional) — The config object or dictionary of the text backbone.
  • vision_config (Union[~models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedVisionConfig, dict[str, Any]], optional) — The config object or dictionary of the vision backbone.
  • audio_config (Union[~models.gemma4_unified.configuration_gemma4_unified.Gemma4UnifiedAudioConfig, dict[str, Any]], optional) — The config object or dictionary of the audio backbone.
  • boi_token_id (int, optional, defaults to 255999) — The begin-of-image token index to wrap the image prompt.
  • eoi_token_id (int, optional, defaults to 258882) — The end-of-image token index to wrap the image prompt.
  • image_token_id (int, optional, defaults to 258880) — The image token index used as a placeholder for input images.
  • video_token_id (int, optional, defaults to 258884) — The video token index used as a placeholder for input videos.
  • boa_token_id (int, optional, defaults to 256000) — The begin-of-audio token index to wrap the audio prompt.
  • eoa_token_index (int, optional, defaults to 258883) — The end-of-audio token index to wrap the audio prompt.
  • audio_token_id (int, optional, defaults to 258881) — The audio token index used as a placeholder for input audio.
  • initializer_range (float, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
  • tie_word_embeddings (bool, optional, defaults to True) — Whether to tie weight embeddings according to model’s tied_weights_keys mapping.

This is the configuration class to store the configuration of a Gemma4UnifiedModel. It is used to instantiate a Gemma4 Unified model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the google/gemma-4-12B-it

Configuration objects inherit from PreTrainedConfig and can be used to control the model outputs. Read the documentation from PreTrainedConfig for more information.

Example:

>>> from transformers import (
>>>     Gemma4UnifiedAudioConfig,
>>>     Gemma4UnifiedConfig,
>>>     Gemma4UnifiedForConditionalGeneration,
>>>     Gemma4UnifiedTextConfig,
>>>     Gemma4UnifiedVisionConfig,
>>> )

>>> # Initializing a Gemma 4 Audio config.
>>> audio_config = Gemma4UnifiedAudioConfig()

>>> # Initializing a Gemma 4 Text config.
>>> text_config = Gemma4UnifiedTextConfig()

>>> # Initializing a Gemma 4 vision config.
>>> vision_config = Gemma4UnifiedVisionConfig()

>>> # Initializing a Gemma 4 config similar to google/gemma-4-e2b-it
>>> configuration = Gemma4UnifiedConfig(text_config, vision_config, audio_config)

>>> # Initializing a model from the google/gemma-4-e2b-it configuration
>>> model = Gemma4UnifiedForConditionalGeneration(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config

Gemma4UnifiedTextConfig

class transformers.Gemma4UnifiedTextConfig

< >

( transformers_version: str | None = Nonearchitectures: list[str] | None = Noneoutput_hidden_states: bool | None = Falsereturn_dict: bool | None = Truedtype: typing.Union[str, ForwardRef('torch.dtype'), NoneType] = Nonechunk_size_feed_forward: int = 0is_encoder_decoder: bool = Falseid2label: dict[int, str] | dict[str, str] | None = Nonelabel2id: dict[str, int] | dict[str, str] | None = Noneproblem_type: typing.Optional[typing.Literal['regression', 'single_label_classification', 'multi_label_classification']] = Nonevocab_size: int = 262144hidden_size: int = 2304intermediate_size: int = 9216num_hidden_layers: int = 30num_attention_heads: int = 8num_key_value_heads: int = 4head_dim: int = 256hidden_activation: str = 'gelu_pytorch_tanh'max_position_embeddings: int = 262144initializer_range: float = 0.02rms_norm_eps: float = 1e-06use_cache: bool = Truepad_token_id: int | None = 0eos_token_id: int | list[int] | None = 1bos_token_id: int | None = 2tie_word_embeddings: bool = Truerope_parameters: dict | None = Noneattention_bias: bool = Falseattention_dropout: int | float | None = 0.0sliding_window: int = 1024layer_types: list[str] | None = Nonefinal_logit_softcapping: float | None = Noneuse_bidirectional_attention: typing.Optional[typing.Literal['all', 'vision']] = 'vision'attention_k_eq_v: bool = Falsenum_kv_shared_layers: int = 0use_double_wide_mlp: bool = False )

Parameters

  • vocab_size (int, optional, defaults to 262144) — Vocabulary size of the model. Defines the number of different tokens that can be represented by the input_ids.
  • hidden_size (int, optional, defaults to 2304) — Dimension of the hidden representations.
  • intermediate_size (int, optional, defaults to 9216) — Dimension of the MLP representations.
  • num_hidden_layers (int, optional, defaults to 30) — Number of hidden layers in the Transformer decoder.
  • num_attention_heads (int, optional, defaults to 8) — Number of attention heads for each attention layer in the Transformer decoder.
  • num_key_value_heads (int, optional, defaults to 4) — This is the number of key_value heads that should be used to implement Grouped Query Attention. If num_key_value_heads=num_attention_heads, the model will use Multi Head Attention (MHA), if num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details, check out this paper. If it is not specified, will default to num_attention_heads.
  • head_dim (int, optional, defaults to 256) — The attention head dimension. If None, it will default to hidden_size // num_attention_heads
  • hidden_activation (str, optional, defaults to gelu_pytorch_tanh) — The non-linear activation function (function or string) in the decoder. For example, "gelu", "relu", "silu", etc.
  • max_position_embeddings (int, optional, defaults to 262144) — The maximum sequence length that this model might ever be used with.
  • initializer_range (float, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
  • rms_norm_eps (float, optional, defaults to 1e-06) — The epsilon used by the rms normalization layers.
  • use_cache (bool, optional, defaults to True) — Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if config.is_decoder=True or when the model is a decoder-only generative model.
  • pad_token_id (int, optional, defaults to 0) — Token id used for padding in the vocabulary.
  • eos_token_id (Union[int, list[int]], optional, defaults to 1) — Token id used for end-of-stream in the vocabulary.
  • bos_token_id (int, optional, defaults to 2) — Token id used for beginning-of-stream in the vocabulary.
  • tie_word_embeddings (bool, optional, defaults to True) — Whether to tie weight embeddings according to model’s tied_weights_keys mapping.
  • rope_parameters (dict, optional) — Dictionary containing the configuration parameters for the RoPE embeddings. The dictionary should contain a value for rope_theta and optionally parameters used for scaling in case you want to use RoPE with longer max_position_embeddings.
  • attention_bias (bool, optional, defaults to False) — Whether to use a bias in the query, key, value and output projection layers during self-attention.
  • attention_dropout (Union[int, float], optional, defaults to 0.0) — The dropout ratio for the attention probabilities.
  • sliding_window (int, optional, defaults to 1024) — Sliding window attention window size. If None, no sliding window is applied.
  • layer_types (list[str], optional) — A list that explicitly maps each layer index with its layer type. If not provided, it will be automatically generated based on config values.
  • final_logit_softcapping (float, optional) — Soft-capping value applied to the final logits before computing the probability distribution. Logits are scaled by tanh(logit / cap) * cap.
  • use_bidirectional_attention (str, optional) — Controls bidirectional attention behavior. When set to "vision", vision tokens attend bidirectionally while text tokens use causal attention. When set to "all", all tokens use bidirectional attention.
  • attention_k_eq_v (bool, defaults to False) — Whether keys and values share the same projection weights. When True, the key projection output is reused as the value projection.
  • num_kv_shared_layers (int, defaults to 0) — Number of consecutive decoder layers that share the same key-value projections. A value of 0 means no sharing (each layer has independent KV projections).
  • use_double_wide_mlp (bool, defaults to False) — Whether to use a double-width MLP with fused gate and up projections.

This is the configuration class to store the configuration of a Gemma4UnifiedModel. It is used to instantiate a Gemma4 Unified model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the google/gemma-4-12B-it

Configuration objects inherit from PreTrainedConfig and can be used to control the model outputs. Read the documentation from PreTrainedConfig for more information.

Gemma4UnifiedVisionConfig

class transformers.Gemma4UnifiedVisionConfig

< >

( transformers_version: str | None = Nonearchitectures: list[str] | None = Noneoutput_hidden_states: bool | None = Falsereturn_dict: bool | None = Truedtype: typing.Union[str, ForwardRef('torch.dtype'), NoneType] = Nonechunk_size_feed_forward: int = 0is_encoder_decoder: bool = Falseid2label: dict[int, str] | dict[str, str] | None = Nonelabel2id: dict[str, int] | dict[str, str] | None = Noneproblem_type: typing.Optional[typing.Literal['regression', 'single_label_classification', 'multi_label_classification']] = Nonepatch_size: int = 16pooling_kernel_size: int = 3mm_embed_dim: int = 3840mm_posemb_size: int = 1120rms_norm_eps: float = 1e-06output_proj_dims: int = 3840initializer_range: float = 0.02 )

Parameters

  • patch_size (int, defaults to 16) — Size of the image patches in pixels. Images are first patchified at this resolution.
  • pooling_kernel_size (int, defaults to 3) — Kernel size for merging patches into model patches. A 3×3 merge produces model patches of size patch_size * pooling_kernel_size = 48 pixels.
  • mm_embed_dim (int, defaults to 3840) — Hidden dimension for the patch embedding Dense projection (matches the text model hidden_size).
  • mm_posemb_size (int, defaults to 1120) — Size of the factorized 2D positional embedding table. The table has shape (mm_posemb_size, 2, mm_embed_dim) and is looked up per-axis.
  • rms_norm_eps (float, optional, defaults to 1e-06) — The epsilon used by the rms normalization layers.
  • output_proj_dims (int, defaults to 3840) — Output dimension of the multimodal embedder projection (maps to text hidden size). This is set by the composite config’s text_config.hidden_size at runtime.
  • initializer_range (float, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

This is the configuration class to store the configuration of a Gemma4UnifiedModel. It is used to instantiate a Gemma4 Unified model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the google/gemma-4-12B-it

Configuration objects inherit from PreTrainedConfig and can be used to control the model outputs. Read the documentation from PreTrainedConfig for more information.

Gemma4UnifiedAudioFeatureExtractor

class transformers.Gemma4UnifiedAudioFeatureExtractor

< >

( feature_size: int = 640sampling_rate: int = 16000padding_value: float = 0.0audio_samples_per_token: int = 640**kwargs )

Parameters

  • feature_size (int, optional, defaults to 640) — The feature dimension of the extracted features (samples per token).
  • sampling_rate (int, optional, defaults to 16000) — The sampling rate at which the audio files should be digitalized expressed in hertz (Hz).
  • padding_value (float, optional, defaults to 0.0) — Padding value used to pad the audio.
  • audio_samples_per_token (int, optional, defaults to 640) — Number of raw audio samples per output token. At 16 kHz, 640 samples = 40ms.

Encoder-free audio feature extractor that chunks raw waveform into frames.

Unlike the standard Gemma4 audio feature extractor which computes mel spectrograms, this unified version simply chunks raw 16 kHz audio into fixed-length frames of audio_samples_per_token samples each. Each frame becomes a single audio soft token with the raw waveform samples as its features.

__call__

< >

( raw_speech: numpy.ndarray | list[float] | list[numpy.ndarray] | list[list[float]]padding: bool | str = 'longest'max_length: int | None = Nonetruncation: bool = Truereturn_tensors: str | transformers.utils.generic.TensorType | None = None**kwargs )

Parameters

  • raw_speech — The raw audio waveform(s) to process.
  • padding (str, optional, defaults to "longest") — Padding strategy for batches with different lengths.
  • max_length (int, optional) — Maximum number of tokens to produce per audio.
  • truncation (bool, optional, defaults to True) — Whether to truncate audio above max_length tokens.
  • return_tensors (str, optional) — The type of tensors to return.

Chunk raw audio waveforms into fixed-length frames for the unified model.

Gemma4UnifiedImageProcessor

class transformers.Gemma4UnifiedImageProcessor

< >

( **kwargs: Unpack )

Parameters

  • do_convert_rgb (bool, kwargs, optional, defaults to True) — Whether to convert the image to RGB.
  • do_resize (bool, kwargs, optional, defaults to True) — Whether to resize the image.
  • size (Annotated[int | list[int] | tuple[int, ...] | dict[str, int] | None, None], kwargs, defaults to None) — Describes the maximum input dimensions to the model.
  • default_to_square (bool, kwargs, optional, defaults to True) — Whether to default to a square image when resizing, if size is an int.
  • crop_size (Annotated[int | list[int] | tuple[int, ...] | dict[str, int] | None, None], kwargs) — Size of the output image after applying center_crop.
  • resample (Annotated[Union[int, PILImageResampling, NoneType], None], kwargs, defaults to Resampling.BICUBIC) — Resampling filter to use if resizing the image. This can be one of the enum PILImageResampling. Only has an effect if do_resize is set to True.
  • do_rescale (bool, kwargs, optional, defaults to True) — Whether to rescale the image.
  • rescale_factor (float, kwargs, optional, defaults to 0.00392156862745098) — Rescale factor to rescale the image by if do_rescale is set to True.
  • do_normalize (bool, kwargs, optional, defaults to False) — Whether to normalize the image.
  • image_mean (Union[float, list[float], tuple[float, ...]], kwargs, optional, defaults to [0.0, 0.0, 0.0]) — Image mean to use for normalization. Only has an effect if do_normalize is set to True.
  • image_std (Union[float, list[float], tuple[float, ...]], kwargs, optional, defaults to [1.0, 1.0, 1.0]) — Image standard deviation to use for normalization. Only has an effect if do_normalize is set to True.
  • do_pad (bool, kwargs, optional) — Whether to pad the image. Padding is done either to the largest size in the batch or to a fixed square size per image. The exact padding strategy depends on the model.
  • pad_size (Annotated[int | list[int] | tuple[int, ...] | dict[str, int] | None, None], kwargs) — The size in {"height": int, "width" int} to pad the images to. Must be larger than any image size provided for preprocessing. If pad_size is not provided, images will be padded to the largest height and width in the batch. Applied only when do_pad=True.
  • do_center_crop (bool, kwargs, optional) — Whether to center crop the image.
  • data_format (Union[str, ~image_utils.ChannelDimension], kwargs, optional) — Only ChannelDimension.FIRST is supported. Added for compatibility with slow processors.
  • input_data_format (Union[str, ~image_utils.ChannelDimension], kwargs, optional) — The channel dimension format for the input image. If unset, the channel dimension format is inferred from the input image. Can be one of:
    • "channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format.
    • "channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format.
    • "none" or ChannelDimension.NONE: image in (height, width) format.
  • device (Annotated[Union[str, torch.device, NoneType], None], kwargs) — The device to process the videos on. If unset, the device is inferred from the input videos.
  • return_tensors (Annotated[str | ~utils.generic.TensorType | None, None], kwargs) — Returns stacked tensors if set to 'pt', otherwise returns a list of tensors.
  • disable_grouping (bool, kwargs, optional) — Whether to disable grouping of images by size to process them individually and not in batches. If None, will be set to True if the images are on CPU, and False otherwise. This choice is based on empirical observations, as detailed here: https://github.com/huggingface/transformers/pull/38157
  • image_seq_length (int, kwargs, optional) — The number of image tokens to be used for each image in the input. Added for backward compatibility but this should be set as a processor attribute in future models.
  • patch_size (int, kwargs, optional) — Size of each teacher image patch in pixels (before merging).
  • max_soft_tokens (int, kwargs, optional) — Maximum number of soft (vision) tokens per image after patch merging. Must be one of {70, 140, 280, 560, 1120}.
  • pooling_kernel_size (int, kwargs, optional) — Kernel size for merging teacher patches into model patches.

Constructs a Gemma4 unified image processor.

Gemma4UnifiedVideoProcessor

class transformers.Gemma4UnifiedVideoProcessor

< >

( **kwargs: Unpack )

Parameters

  • do_convert_rgb (bool, kwargs, optional, defaults to True) — Whether to convert the image to RGB.
  • do_resize (bool, kwargs, optional, defaults to True) — Whether to resize the image.
  • size (Annotated[int | list[int] | tuple[int, ...] | dict[str, int] | None, None], kwargs, defaults to None) — Describes the maximum input dimensions to the model.
  • default_to_square (bool, kwargs, optional, defaults to True) — Whether to default to a square image when resizing, if size is an int.
  • resample (Annotated[Union[int, PILImageResampling, NoneType], None], kwargs, defaults to Resampling.BICUBIC) — Resampling filter to use if resizing the image. This can be one of the enum PILImageResampling. Only has an effect if do_resize is set to True.
  • do_rescale (bool, kwargs, optional, defaults to True) — Whether to rescale the image.
  • rescale_factor (float, kwargs, optional, defaults to 0.00392156862745098) — Rescale factor to rescale the image by if do_rescale is set to True.
  • do_normalize (bool, kwargs, optional, defaults to True) — Whether to normalize the image.
  • image_mean (Union[float, list[float], tuple[float, ...]], kwargs, optional, defaults to [0.0, 0.0, 0.0]) — Image mean to use for normalization. Only has an effect if do_normalize is set to True.
  • image_std (Union[float, list[float], tuple[float, ...]], kwargs, optional, defaults to [1.0, 1.0, 1.0]) — Image standard deviation to use for normalization. Only has an effect if do_normalize is set to True.
  • do_center_crop (bool, kwargs, optional, defaults to None) — Whether to center crop the image.
  • do_pad (bool, kwargs, optional) — Whether to pad the image. Padding is done either to the largest size in the batch or to a fixed square size per image. The exact padding strategy depends on the model.
  • crop_size (Annotated[int | list[int] | tuple[int, ...] | dict[str, int] | None, None], kwargs, defaults to None) — Size of the output image after applying center_crop.
  • data_format (Union[str, ~image_utils.ChannelDimension], kwargs, optional) — Only ChannelDimension.FIRST is supported. Added for compatibility with slow processors.
  • input_data_format (Union[str, ~image_utils.ChannelDimension], kwargs, optional) — The channel dimension format for the input image. If unset, the channel dimension format is inferred from the input image. Can be one of:
    • "channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format.
    • "channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format.
    • "none" or ChannelDimension.NONE: image in (height, width) format.
  • device (Annotated[Union[str, torch.device, NoneType], None], kwargs) — The device to process the videos on. If unset, the device is inferred from the input videos.
  • do_sample_frames (bool, kwargs, optional, defaults to True) — Whether to sample frames from the video before processing or to process the whole video.
  • video_metadata (Annotated[~video_utils.VideoMetadata | dict | list[dict | ~video_utils.VideoMetadata] | list[list[dict | ~video_utils.VideoMetadata]] | None, None], kwargs, defaults to None) — Metadata of the video containing information about total duration, fps and total number of frames. It will be used to sample frames from video or compute timestamps. Don’t pass any metadata unless you are trying to decode the video manually before processing
  • fps (Annotated[int | float | None, None], kwargs, defaults to None) — Target frames to sample per second when do_sample_frames=True.
  • num_frames (Annotated[int | None, None], kwargs, defaults to 32) — Maximum number of frames to sample when do_sample_frames=True.
  • return_metadata (bool, kwargs, optional, defaults to False) — Whether to return video metadata or not. Video metadats is an object containing info about video duration, fps, decoding backend, etc.
  • return_tensors (Annotated[str | ~utils.generic.TensorType | None, None], kwargs) — Returns stacked tensors if set to 'pt', otherwise returns a list of tensors.
  • patch_size (int, kwargs, optional) — Size of each image patch in pixels.
  • max_soft_tokens (int, kwargs, optional) — Maximum number of soft (vision) tokens per video frame. Must be one of {70, 140, 280, 560, 1120}.
  • pooling_kernel_size (int, kwargs, optional) — Spatial pooling kernel size applied after patchification.

Constructs a Gemma4UnifiedVideoProcessor video processor.

Gemma4UnifiedProcessor

class transformers.Gemma4UnifiedProcessor

< >

( feature_extractorimage_processortokenizervideo_processorchat_template = Noneimage_seq_length: int = 280audio_seq_length: int = 750audio_ms_per_token: int = 40**kwargs )

Parameters

  • feature_extractor (Gemma4UnifiedAudioFeatureExtractor) — The feature extractor is a required input.
  • image_processor (Gemma4UnifiedImageProcessor) — The image processor is a required input.
  • tokenizer (tokenizer_class) — The tokenizer is a required input.
  • video_processor (Gemma4UnifiedVideoProcessor) — The video processor is a required input.
  • chat_template (str) — A Jinja template to convert lists of messages in a chat into a tokenizable string.
  • image_seq_length (int, optional, defaults to 280) — The number of soft tokens per image used for placeholder expansion.
  • audio_seq_length (int, optional, defaults to 750) — The maximum number of audio soft tokens per audio segment. Serves as an upper-bound cap when dynamic audio token counts are computed.
  • audio_ms_per_token (int, optional, defaults to 40) — Milliseconds of audio per output soft token. Used to dynamically compute the number of audio placeholder tokens as ceil(duration_ms / audio_ms_per_token). The default of 40 comes from the SSCP convolution’s 4× time reduction on 10ms frames.

Constructs a Gemma4UnifiedProcessor which wraps a feature extractor, a image processor, a tokenizer, and a video processor into a single processor.

[Gemma4UnifiedProcessor] offers all the functionalities of [Gemma4UnifiedAudioFeatureExtractor], [Gemma4UnifiedImageProcessor], [tokenizer_class], and [Gemma4UnifiedVideoProcessor]. See the [~Gemma4UnifiedAudioFeatureExtractor], [~Gemma4UnifiedImageProcessor], [~tokenizer_class], and [~Gemma4UnifiedVideoProcessor] for more information.

__call__

< >

( images: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), list['PIL.Image.Image'], list[numpy.ndarray], list['torch.Tensor'], NoneType] = Nonetext: str | list[str] | list[list[str]] | None = Nonevideos: typing.Union[list['PIL.Image.Image'], numpy.ndarray, ForwardRef('torch.Tensor'), list[numpy.ndarray], list['torch.Tensor'], list[list['PIL.Image.Image']], list[list[numpy.ndarray]], list[list['torch.Tensor']], transformers.video_utils.URL, list[transformers.video_utils.URL], list[list[transformers.video_utils.URL]], transformers.video_utils.Path, list[transformers.video_utils.Path], list[list[transformers.video_utils.Path]], NoneType] = Noneaudio: typing.Union[numpy.ndarray, ForwardRef('torch.Tensor'), collections.abc.Sequence[numpy.ndarray], collections.abc.Sequence['torch.Tensor'], NoneType] = None**kwargs: Unpack )

Parameters

  • images (Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]], optional) — Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If passing in images with pixel values between 0 and 1, set do_rescale=False.
  • text (Union[str, list[str], list[list[str]]], optional) — The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings (pretokenized string). If you pass a pretokenized input, set is_split_into_words=True to avoid ambiguity with batched inputs.
  • videos (Union[list[PIL.Image.Image], numpy.ndarray, torch.Tensor, list[numpy.ndarray], list[torch.Tensor], list[list[PIL.Image.Image]], list[list[numpy.ndarray]], list[list[torch.Tensor]], ~video_utils.URL, list[~video_utils.URL], list[list[~video_utils.URL]], ~video_utils.Path, list[~video_utils.Path], list[list[~video_utils.Path]]], optional) — Video to preprocess. Expects a single or batch of videos with pixel values ranging from 0 to 255. If passing in videos with pixel values between 0 and 1, set do_rescale=False.
  • audio (Union[numpy.ndarray, torch.Tensor, collections.abc.Sequence[numpy.ndarray], collections.abc.Sequence[torch.Tensor]], optional) — The audio or batch of audios to be prepared. Each audio can be a NumPy array or PyTorch tensor. In case of a NumPy array/PyTorch tensor, each audio should be of shape (C, T), where C is a number of channels, and T is the sample length of the audio.
  • return_tensors (str or TensorType, optional) — If set, will return tensors of a particular framework. Acceptable values are:

    • 'pt': Return PyTorch torch.Tensor objects.
    • 'np': Return NumPy np.ndarray objects.
  • **kwargs (ProcessingKwargs, optional) — Additional processing options for each modality (text, images, videos, audio). Model-specific parameters are listed above; see the TypedDict class for the complete list of supported arguments.

Gemma4UnifiedPreTrainedModel

class transformers.Gemma4UnifiedPreTrainedModel

< >

( config: PreTrainedConfig*inputs**kwargs )

Parameters

  • config (PreTrainedConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

( *args**kwargs )

A mock value for a dotted path (e.g. torch.float32): attribute access chains, calls behave as pass-through decorators, repr is the dotted path, and using it as a base class substitutes a plain-type base (PEP 560 __mro_entries__), so real subclasses keep a normal metaclass and inspect.signature reads their real __init__ instead of a mock’s.

Gemma4UnifiedModel

class transformers.Gemma4UnifiedModel

< >

( config: Gemma4UnifiedConfig )

Parameters

  • config (Gemma4UnifiedConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

The base Gemma 4 model comprising a vision backbone, an audio backbone, and a language model without a language modeling head.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: typing.Optional[torch.LongTensor] = Nonepixel_values: typing.Optional[torch.FloatTensor] = Nonepixel_values_videos: typing.Optional[torch.FloatTensor] = Noneinput_features: typing.Optional[torch.FloatTensor] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneinput_features_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Nonemm_token_type_ids: typing.Optional[torch.LongTensor] = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Noneuse_cache: bool | None = Noneimage_position_ids: typing.Optional[torch.LongTensor] = Nonevideo_position_ids: typing.Optional[torch.LongTensor] = None**kwargs: Unpack ) → Gemma4UnifiedModelOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • pixel_values (torch.FloatTensor of shape (batch_size, num_channels, image_size, image_size), optional) — The tensors corresponding to the input images. Pixel values can be obtained using Gemma4UnifiedImageProcessor. See Gemma4UnifiedImageProcessor.__call__() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedImageProcessor for processing images).
  • pixel_values_videos (torch.FloatTensor of shape (batch_size, num_frames, num_channels, frame_size, frame_size), optional) — The tensors corresponding to the input video. Pixel values for videos can be obtained using Gemma4UnifiedVideoProcessor. See Gemma4UnifiedVideoProcessor.__call__() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedVideoProcessor for processing videos).
  • input_features (torch.FloatTensor of shape (batch_size, sequence_length, feature_dim), optional) — The tensors corresponding to the input audio features. Audio features can be obtained using Gemma4UnifiedAudioFeatureExtractor. See Gemma4UnifiedAudioFeatureExtractor.call() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedAudioFeatureExtractor for processing audios).
  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • input_features_mask (torch.FloatTensor of shape (num_images, seq_length)) — The attention mask for the input audio.
  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • mm_token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2). Multimodal token type ids can be obtained using AutoProcessor. See ProcessorMixin.call() for details.
  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • use_cache (bool, optional) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
  • image_position_ids (torch.LongTensor of shape (batch_size, max_patches, 2), optional) — 2D patch position coordinates from the image processor, with (-1, -1) indicating padding. Passed through to the vision encoder for positional embedding computation.
  • video_position_ids (torch.LongTensor of shape (num_videos, num_frames, max_patches, 2), optional) — 2D patch position coordinates from the video processor, with (-1, -1) indicating padding. Passed through to the vision encoder for positional embedding computation.

Returns

Gemma4UnifiedModelOutputWithPast or tuple(torch.FloatTensor)

A Gemma4UnifiedModelOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Gemma4UnifiedConfig) and inputs.

The Gemma4UnifiedModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

  • last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size)) — Sequence of hidden-states at the output of the last layer of the model.

    If past_key_values is used only the last hidden-state of the sequences of shape (batch_size, 1, hidden_size) is output.

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

  • image_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). image_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.

  • audio_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). audio_hidden_states of the model produced by the audio encoder and after projecting the last hidden state.

  • shared_kv_states (dict, optional) — Dictionary mapping layer type strings to tuples of (key_states, value_states) tensors. Used to pass shared KV states between layers during KV sharing.

Gemma4UnifiedTextModel

class transformers.Gemma4UnifiedTextModel

< >

( config: Gemma4UnifiedTextConfig )

Parameters

  • config (Gemma4UnifiedTextConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

The base Gemma 4 unified language model without a language modeling head.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: typing.Optional[torch.LongTensor] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Noneuse_cache: bool | None = None**kwargs: Unpack ) → Gemma4UnifiedTextModelOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • use_cache (bool, optional) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).

Returns

Gemma4UnifiedTextModelOutputWithPast or tuple(torch.FloatTensor)

A Gemma4UnifiedTextModelOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Gemma4UnifiedConfig) and inputs.

The Gemma4UnifiedTextModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

  • shared_kv_states (dict, optional) — Dictionary mapping layer type strings to tuples of (key_states, value_states) tensors. Used to pass shared KV states between layers during KV sharing.

Gemma4UnifiedForCausalLM

class transformers.Gemma4UnifiedForCausalLM

< >

( config: Gemma4UnifiedTextConfig )

Parameters

  • config (Gemma4UnifiedTextConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

The base Gemma 4 language model with a language modeling head.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: typing.Optional[torch.LongTensor] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Nonelabels: typing.Optional[torch.LongTensor] = Noneuse_cache: bool | None = Nonelogits_to_keep: typing.Union[int, torch.Tensor] = 0**kwargs: Unpack ) → Gemma4UnifiedCausalLMOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • labels (torch.LongTensor of shape (batch_size, sequence_length), optional) — Labels for computing the masked language modeling loss. Indices should either be in [0, ..., config.vocab_size] or -100 (see input_ids docstring). Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size].
  • use_cache (bool, optional) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
  • logits_to_keep (Union[int, torch.Tensor], optional, defaults to 0) — If an int, compute logits for the last logits_to_keep tokens. If 0, calculate logits for all input_ids (special case). Only last token logits are needed for generation, and calculating them only for that token can save memory, which becomes pretty significant for long sequences or large vocabulary size. If a torch.Tensor, must be 1D corresponding to the indices to keep in the sequence length dimension. This is useful when using packed tensor format (single dimension for batch and sequence length).

Returns

Gemma4UnifiedCausalLMOutputWithPast or tuple(torch.FloatTensor)

A Gemma4UnifiedCausalLMOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Gemma4UnifiedConfig) and inputs.

The Gemma4UnifiedForCausalLM forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss (for next-token prediction).

  • logits (torch.FloatTensor of shape (batch_size, sequence_length, config.text_config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple[torch.FloatTensor], optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple[torch.FloatTensor], optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

  • image_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). image_hidden_states of the model produced by the vision encoder after projecting last hidden state.

  • audio_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). audio_hidden_states of the model produced by the audio encoder and after projecting the last hidden state.

  • shared_kv_states (dict, optional) — Dictionary mapping layer type strings to tuples of (key_states, value_states) tensors. Used to pass shared KV states between layers during KV sharing.

Example:

>>> from transformers import AutoTokenizer, Gemma4UnifiedForCausalLM

>>> model = Gemma4UnifiedForCausalLM.from_pretrained("google/gemma-4-12B-it")
>>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-4-12B-it")

>>> prompt = "What is your favorite condiment?"
>>> inputs = tokenizer(prompt, return_tensors="pt")

>>> # Generate
>>> generate_ids = model.generate(inputs.input_ids, max_length=30)
>>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
"What is your favorite condiment?"

Gemma4UnifiedForConditionalGeneration

class transformers.Gemma4UnifiedForConditionalGeneration

< >

( config: Gemma4UnifiedConfig )

Parameters

  • config (Gemma4UnifiedConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

The base Gemma 4 model comprising a vision backbone, an audio backbone, a language model, and a language modeling head.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: typing.Optional[torch.LongTensor] = Nonepixel_values: typing.Optional[torch.FloatTensor] = Nonepixel_values_videos: typing.Optional[torch.FloatTensor] = Noneinput_features: typing.Optional[torch.FloatTensor] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneinput_features_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Noneimage_position_ids: typing.Optional[torch.LongTensor] = Nonevideo_position_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Nonemm_token_type_ids: typing.Optional[torch.LongTensor] = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Nonelabels: typing.Optional[torch.LongTensor] = Noneuse_cache: bool | None = Nonelogits_to_keep: typing.Union[int, torch.Tensor] = 0**kwargs: Unpack ) → Gemma4UnifiedCausalLMOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • pixel_values (torch.FloatTensor of shape (batch_size, num_channels, image_size, image_size), optional) — The tensors corresponding to the input images. Pixel values can be obtained using Gemma4UnifiedImageProcessor. See Gemma4UnifiedImageProcessor.__call__() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedImageProcessor for processing images).
  • pixel_values_videos (torch.FloatTensor of shape (batch_size, num_frames, num_channels, frame_size, frame_size), optional) — The tensors corresponding to the input video. Pixel values for videos can be obtained using Gemma4UnifiedVideoProcessor. See Gemma4UnifiedVideoProcessor.__call__() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedVideoProcessor for processing videos).
  • input_features (torch.FloatTensor of shape (batch_size, sequence_length, feature_dim), optional) — The tensors corresponding to the input audio features. Audio features can be obtained using Gemma4UnifiedAudioFeatureExtractor. See Gemma4UnifiedAudioFeatureExtractor.call() for details (Gemma4UnifiedProcessor uses Gemma4UnifiedAudioFeatureExtractor for processing audios).
  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • input_features_mask (torch.FloatTensor of shape (num_images, seq_length)) — The attention mask for the input audio.
  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • image_position_ids (torch.LongTensor of shape (batch_size, max_patches, 2), optional) — 2D patch position coordinates from the image processor, with (-1, -1) indicating padding. Passed through to the vision encoder for positional embedding computation.
  • video_position_ids (torch.LongTensor of shape (num_videos, num_frames, max_patches, 2), optional) — 2D patch position coordinates from the video processor, with (-1, -1) indicating padding. Passed through to the vision encoder for positional embedding computation.
  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • mm_token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2). Multimodal token type ids can be obtained using AutoProcessor. See ProcessorMixin.call() for details.
  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • labels (torch.LongTensor of shape (batch_size, sequence_length), optional) — Labels for computing the masked language modeling loss. Indices should either be in [0, ..., config.vocab_size] or -100 (see input_ids docstring). Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size].
  • use_cache (bool, optional) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
  • logits_to_keep (Union[int, torch.Tensor], optional, defaults to 0) — If an int, compute logits for the last logits_to_keep tokens. If 0, calculate logits for all input_ids (special case). Only last token logits are needed for generation, and calculating them only for that token can save memory, which becomes pretty significant for long sequences or large vocabulary size. If a torch.Tensor, must be 1D corresponding to the indices to keep in the sequence length dimension. This is useful when using packed tensor format (single dimension for batch and sequence length).

Returns

Gemma4UnifiedCausalLMOutputWithPast or tuple(torch.FloatTensor)

A Gemma4UnifiedCausalLMOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Gemma4UnifiedConfig) and inputs.

The Gemma4UnifiedForConditionalGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss (for next-token prediction).

  • logits (torch.FloatTensor of shape (batch_size, sequence_length, config.text_config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple[torch.FloatTensor], optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple[torch.FloatTensor], optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

  • image_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). image_hidden_states of the model produced by the vision encoder after projecting last hidden state.

  • audio_hidden_states (torch.FloatTensor, optional) — A torch.FloatTensor of size (batch_size, num_images, sequence_length, hidden_size). audio_hidden_states of the model produced by the audio encoder and after projecting the last hidden state.

  • shared_kv_states (dict, optional) — Dictionary mapping layer type strings to tuples of (key_states, value_states) tensors. Used to pass shared KV states between layers during KV sharing.

Example:

>>> from PIL import Image
>>> from transformers import AutoProcessor, Gemma4UnifiedForConditionalGeneration

>>> model = Gemma4UnifiedForConditionalGeneration.from_pretrained("google/gemma-4-12B-it")
>>> processor = AutoProcessor.from_pretrained("google/gemma-4-12B-it")

>>> messages = [
...     {
...         "role": "user", "content": [
...             {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"},
...             {"type": "text", "text": "Where is the cat standing?"},
...         ]
...     },
... ]

>>> inputs = processor.apply_chat_template(
...     messages,
...     tokenize=True,
...     return_dict=True,
...     return_tensors="pt",
...     add_generation_prompt=True
... )
>>> # Generate
>>> generate_ids = model.generate(**inputs)
>>> processor.batch_decode(generate_ids, skip_special_tokens=True)[0]
Update on GitHub

←Gemma4 GIT→