# Gemma 3 [[gemma3]] [Gemma 3](https://goo.gle/Gemma3Report)는 사전 훈련된 버전과 지시문 조정 버전을 갖춘 멀티모달 모델로, 1B, 13B, 27B 매개변수로 제공됩니다. 아키텍처는 이전 Gemma 버전과 대부분 동일합니다. 주요 차이점은 모든 글로벌 셀프 어텐션 레이어마다 5개의 로컬 슬라이딩 윈도우 셀프 어텐션 레이어를 번갈아 사용하는 점, 128K 토큰의 더 긴 컨텍스트 길이를 지원하는 점, 그리고 고해상도 이미지나 정사각형이 아닌 종횡비의 이미지에서 정보가 사라지는 것을 방지하기 위해 고해상도 이미지를 "패닝 및 스캐닝"할 수 있는 [SigLip](./siglip) 인코더를 사용한다는 점입니다. 지시문 조정 버전은 지식 증류 및 강화 학습으로 후속 학습되었습니다. Gemma 3의 모든 원본 체크포인트는 [Gemma 3](https://huggingface.co/collections/google/gemma-3-release-67c6c6f89c4f76621268bb6d) 릴리스에서 확인할 수 있습니다. > [!팁] > Gemma를 다양한 비전 및 언어 작업에 적용하는 추가 예시를 보려면 오른쪽 사이드바의 Gemma 3 모델을 클릭하세요. 아래 예시는 [Pipeline](/docs/transformers/v5.6.0/ko/main_classes/pipelines#transformers.Pipeline) 또는 [AutoModel](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoModel) 클래스를 사용하여 이미지를 기반으로 텍스트를 생성하는 방법을 보여줍니다. ```py import torch from transformers import pipeline pipeline = pipeline( task="image-text-to-text", model="google/gemma-3-4b-pt", device=0, dtype=torch.bfloat16 ) pipeline( "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg", text=" What is shown in this image?" ) ``` ```py import torch from transformers import AutoProcessor, Gemma3ForConditionalGeneration model = Gemma3ForConditionalGeneration.from_pretrained( "google/gemma-3-4b-it", dtype=torch.bfloat16, device_map="auto", attn_implementation="sdpa" ) processor = AutoProcessor.from_pretrained( "google/gemma-3-4b-it", padding_side="left" ) messages = [ { "role": "system", "content": [ {"type": "text", "text": "You are a helpful assistant."} ] }, { "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?"}, ] }, ] inputs = processor.apply_chat_template( messages, tokenize=True, return_dict=True, return_tensors="pt", add_generation_prompt=True, ).to(model.device) output = model.generate(**inputs, max_new_tokens=50, cache_implementation="static") print(processor.decode(output[0], skip_special_tokens=True)) ``` 양자화는 가중치를 더 낮은 정밀도로 표현하여, 큰 모델의 메모리 부담을 줄여줍니다. 사용 가능한 양자화 백엔드에 대한 더 자세한 내용은 [양자화](../quantization/overview) 개요를 참고하세요. 아래 예제에서는 [torchao](../quantization/torchao)를 사용하여 가중치를 int4로만 양자화합니다. ```py # pip install torchao import torch from transformers import TorchAoConfig, Gemma3ForConditionalGeneration, AutoProcessor quantization_config = TorchAoConfig("int4_weight_only", group_size=128) model = Gemma3ForConditionalGeneration.from_pretrained( "google/gemma-3-27b-it", dtype=torch.bfloat16, device_map="auto", quantization_config=quantization_config ) processor = AutoProcessor.from_pretrained( "google/gemma-3-27b-it", padding_side="left" ) messages = [ { "role": "system", "content": [ {"type": "text", "text": "You are a helpful assistant."} ] }, { "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?"}, ] }, ] inputs = processor.apply_chat_template( messages, tokenize=True, return_dict=True, return_tensors="pt", add_generation_prompt=True, ).to(model.device) output = model.generate(**inputs, max_new_tokens=50, cache_implementation="static") print(processor.decode(output[0], skip_special_tokens=True)) ``` [AttentionMaskVisualizer](https://github.com/huggingface/transformers/blob/beb9b5b02246b9b7ee81ddf938f93f44cfeaad19/src/transformers/utils/attention_visualizer.py#L139)를 사용하여 모델이 주목할 수 있는 토큰과 주목할 수 없는 토큰을 더 잘 이해할 수 있습니다. ```py from transformers.utils.attention_visualizer import AttentionMaskVisualizer visualizer = AttentionMaskVisualizer("google/gemma-3-4b-it") visualizer("What is shown in this image?") ``` ## 노트 [[notes]] - 이미지-텍스트 및 이미지 전용 입력에는 [Gemma3ForConditionalGeneration](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ForConditionalGeneration)을 사용하세요. - Gemma 3는 다중 입력 이미지를 지원하지만, 프로세서에 전달하기 전에 이미지가 올바르게 배치되었는지 확인하세요. 각 배치는 하나 이상의 이미지를 포함한 리스트여야 합니다. ```py url_cow = "https://media.istockphoto.com/id/1192867753/photo/cow-in-berchida-beach-siniscola.jpg?s=612x612&w=0&k=20&c=v0hjjniwsMNfJSuKWZuIn8pssmD5h5bSN1peBd1CmH4=" url_cat = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg" messages =[ { "role": "system", "content": [ {"type": "text", "text": "You are a helpful assistant."} ] }, { "role": "user", "content": [ {"type": "image", "url": url_cow}, {"type": "image", "url": url_cat}, {"type": "text", "text": "Which image is cuter?"}, ] }, ] ``` - 프로세서에 전달되는 텍스트에는 이미지가 삽입되어야 하는 위치마다 `` 토큰이 있어야 합니다. - 프로세서에는 채팅 메시지를 모델 입력으로 변환하는 자체 [apply_chat_template()](/docs/transformers/v5.6.0/ko/main_classes/processors#transformers.ProcessorMixin.apply_chat_template) 메소드가 있습니다. - 기본적으로 이미지는 잘리지 않으며 기본 이미지만 모델로 전달됩니다. 고해상도 이미지나 정사각형이 아닌 종횡비의 이미지에서는 비전 인코더가 896x896의 고정 해상도를 사용하기 때문에 아티팩트가 발생할 수 있습니다. 이러한 아티팩트를 방지하고 추론 중 성능을 향상시키려면, `do_pan_and_scan=True`를 설정하여 이미지를 여러 개의 작은 패치로 자르고 기본 이미지 임베딩과 이어 붙입니다. 더 빠른 추론을 위해 팬과 스캔을 비활성화할 수 있습니다. ```diff inputs = processor.apply_chat_template( messages, tokenize=True, return_dict=True, return_tensors="pt", add_generation_prompt=True, + do_pan_and_scan=True, ).to(model.device) ``` - 텍스트 전용 모드로 훈련된 Gemma-3 1B 체크포인트의 경우, [AutoModelForCausalLM](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoModelForCausalLM)을 대신 사용하세요. ```py import torch from transformers import AutoModelForCausalLM, AutoTokenizer tokenizer = AutoTokenizer.from_pretrained( "google/gemma-3-1b-pt", ) model = AutoModelForCausalLM.from_pretrained( "google/gemma-3-1b-pt", dtype=torch.bfloat16, device_map="auto", attn_implementation="sdpa" ) input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to(model.device) output = model.generate(**input_ids, cache_implementation="static") print(tokenizer.decode(output[0], skip_special_tokens=True)) ``` ## Gemma3ImageProcessor[[transformers.Gemma3ImageProcessor]] #### transformers.Gemma3ImageProcessor[[transformers.Gemma3ImageProcessor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L58) Constructs a Gemma3ImageProcessor image processor. pan_and_scan_batchedtransformers.Gemma3ImageProcessor.pan_and_scan_batchedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L82[{"name": "images", "val": ": torch.Tensor"}, {"name": "pan_and_scan_min_crop_size", "val": ": int"}, {"name": "pan_and_scan_max_num_crops", "val": ": int"}, {"name": "pan_and_scan_min_ratio_to_activate", "val": ": float"}]- **images** (`torch.Tensor`) -- Image to resize. - **pan_and_scan_min_crop_size** (`int`, *optional*) -- Minimum size of each crop in pan and scan. - **pan_and_scan_max_num_crops** (`int`, *optional*) -- Maximum number of crops per image in pan and scan. - **pan_and_scan_min_ratio_to_activate** (`float`, *optional*) -- Minimum aspect ratio to activate pan and scan.0 Pan and Scan an image, by cropping into smaller images when the aspect ratio exceeds minimum allowed ratio. **Parameters:** do_pan_and_scan (`bool`, *kwargs*, *optional*) : Whether to apply `pan_and_scan` to images. pan_and_scan_min_crop_size (`int`, *kwargs*, *optional*) : Minimum size of each crop in pan and scan. pan_and_scan_max_num_crops (`int`, *kwargs*, *optional*) : Maximum number of crops per image in pan and scan. pan_and_scan_min_ratio_to_activate (`float`, *kwargs*, *optional*) : Minimum aspect ratio to activate pan and scan. - ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments. #### preprocess[[transformers.Gemma3ImageProcessor.preprocess]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L78) **Parameters:** images (`Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]]`) : 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`. do_pan_and_scan (`bool`, *kwargs*, *optional*) : Whether to apply `pan_and_scan` to images. pan_and_scan_min_crop_size (`int`, *kwargs*, *optional*) : Minimum size of each crop in pan and scan. pan_and_scan_max_num_crops (`int`, *kwargs*, *optional*) : Maximum number of crops per image in pan and scan. pan_and_scan_min_ratio_to_activate (`float`, *kwargs*, *optional*) : Minimum aspect ratio to activate pan and scan. return_tensors (`str` or [TensorType](/docs/transformers/v5.6.0/ko/internal/file_utils#transformers.TensorType), *optional*) : Returns stacked tensors if set to `'pt'`, otherwise returns a list of tensors. - ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments. **Returns:** ``~image_processing_base.BatchFeature`` - **data** (`dict`) -- Dictionary of lists/arrays/tensors returned by the __call__ method ('pixel_values', etc.). - **tensor_type** (`Union[None, str, TensorType]`, *optional*) -- You can give a tensor_type here to convert the lists of integers in PyTorch/Numpy Tensors at initialization. ## Gemma3ImageProcessorFast[[transformers.Gemma3ImageProcessor]] #### transformers.Gemma3ImageProcessor[[transformers.Gemma3ImageProcessor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L58) Constructs a Gemma3ImageProcessor image processor. pan_and_scan_batchedtransformers.Gemma3ImageProcessor.pan_and_scan_batchedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L82[{"name": "images", "val": ": torch.Tensor"}, {"name": "pan_and_scan_min_crop_size", "val": ": int"}, {"name": "pan_and_scan_max_num_crops", "val": ": int"}, {"name": "pan_and_scan_min_ratio_to_activate", "val": ": float"}]- **images** (`torch.Tensor`) -- Image to resize. - **pan_and_scan_min_crop_size** (`int`, *optional*) -- Minimum size of each crop in pan and scan. - **pan_and_scan_max_num_crops** (`int`, *optional*) -- Maximum number of crops per image in pan and scan. - **pan_and_scan_min_ratio_to_activate** (`float`, *optional*) -- Minimum aspect ratio to activate pan and scan.0 Pan and Scan an image, by cropping into smaller images when the aspect ratio exceeds minimum allowed ratio. **Parameters:** do_pan_and_scan (`bool`, *kwargs*, *optional*) : Whether to apply `pan_and_scan` to images. pan_and_scan_min_crop_size (`int`, *kwargs*, *optional*) : Minimum size of each crop in pan and scan. pan_and_scan_max_num_crops (`int`, *kwargs*, *optional*) : Maximum number of crops per image in pan and scan. pan_and_scan_min_ratio_to_activate (`float`, *kwargs*, *optional*) : Minimum aspect ratio to activate pan and scan. - ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments. #### preprocess[[transformers.Gemma3ImageProcessor.preprocess]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/image_processing_gemma3.py#L78) **Parameters:** images (`Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]]`) : 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`. do_pan_and_scan (`bool`, *kwargs*, *optional*) : Whether to apply `pan_and_scan` to images. pan_and_scan_min_crop_size (`int`, *kwargs*, *optional*) : Minimum size of each crop in pan and scan. pan_and_scan_max_num_crops (`int`, *kwargs*, *optional*) : Maximum number of crops per image in pan and scan. pan_and_scan_min_ratio_to_activate (`float`, *kwargs*, *optional*) : Minimum aspect ratio to activate pan and scan. return_tensors (`str` or [TensorType](/docs/transformers/v5.6.0/ko/internal/file_utils#transformers.TensorType), *optional*) : Returns stacked tensors if set to `'pt'`, otherwise returns a list of tensors. - ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments. **Returns:** ``~image_processing_base.BatchFeature`` - **data** (`dict`) -- Dictionary of lists/arrays/tensors returned by the __call__ method ('pixel_values', etc.). - **tensor_type** (`Union[None, str, TensorType]`, *optional*) -- You can give a tensor_type here to convert the lists of integers in PyTorch/Numpy Tensors at initialization. ## Gemma3Processor[[transformers.Gemma3Processor]] #### transformers.Gemma3Processor[[transformers.Gemma3Processor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/processing_gemma3.py#L41) Constructs a Gemma3Processor which wraps a image processor and a tokenizer into a single processor. [Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) offers all the functionalities of [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) and [GemmaTokenizer](/docs/transformers/v5.6.0/ko/model_doc/gemma#transformers.GemmaTokenizer). See the [~Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) and [~GemmaTokenizer](/docs/transformers/v5.6.0/ko/model_doc/gemma#transformers.GemmaTokenizer) for more information. **Parameters:** image_processor (`Gemma3ImageProcessor`) : The image processor is a required input. tokenizer (`GemmaTokenizer`) : The tokenizer 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 `256`) : 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. ## Gemma3TextConfig[[transformers.Gemma3TextConfig]] #### transformers.Gemma3TextConfig[[transformers.Gemma3TextConfig]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/configuration_gemma3.py#L35) This is the configuration class to store the configuration of a Gemma3Model. It is used to instantiate a Gemma3 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-3-4b-it](https://huggingface.co/google/gemma-3-4b-it) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.6.0/ko/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.6.0/ko/main_classes/configuration#transformers.PreTrainedConfig) for more information. ```python >>> from transformers import Gemma3TextModel, Gemma3TextConfig >>> # Initializing a Gemma3Text gemma3_text-7b style configuration >>> configuration = Gemma3TextConfig() >>> # Initializing a model from the gemma3_text-7b style configuration >>> model = Gemma3TextModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` **Parameters:** vocab_size (`int`, *optional*, defaults to `262208`) : 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 `26`) : 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](https://huggingface.co/papers/2305.13245). 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 `131072`) : 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. query_pre_attn_scalar (`float`, *optional*, defaults to 256) : scaling factor used on the attention scores sliding_window (`int`, *optional*, defaults to `4096`) : 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*) : Scaling factor when applying tanh softcapping on the logits. attn_logit_softcapping (`float`, *optional*) : Scaling factor when applying tanh softcapping on the attention scores. use_bidirectional_attention (`bool`, *optional*, defaults to `False`) : If True, the model will attend to all text tokens instead of using a causal mask. This does not change behavior for vision tokens. ## Gemma3Config[[transformers.Gemma3Config]] #### transformers.Gemma3Config[[transformers.Gemma3Config]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/configuration_gemma3.py#L159) This is the configuration class to store the configuration of a Gemma3Model. It is used to instantiate a Gemma3 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-3-4b-it](https://huggingface.co/google/gemma-3-4b-it) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.6.0/ko/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.6.0/ko/main_classes/configuration#transformers.PreTrainedConfig) for more information. Example: ```python >>> from transformers import Gemma3ForConditionalGeneration, Gemma3Config, SiglipVisionConfig, Gemma3TextConfig >>> # Initializing a Siglip-like vision config >>> vision_config = SiglipVisionConfig() >>> # Initializing a Gemma3 Text config >>> text_config = Gemma3TextConfig() >>> # Initializing a Gemma3 gemma-3-4b style configuration >>> configuration = Gemma3Config(vision_config, text_config) >>> # Initializing a model from the gemma-3-4b style configuration >>> model = Gemma3TextConfig(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` **Parameters:** text_config (`Union[~models.gemma3.configuration_gemma3.Gemma3TextConfig, dict[str, Any]]`, *optional*) : The config object or dictionary of the text backbone. vision_config (`Union[~models.siglip.configuration_siglip.SiglipVisionConfig, dict[str, Any]]`, *optional*) : The config object or dictionary of the vision backbone. mm_tokens_per_image (`int`, *optional*, defaults to 256) : The number of tokens per image embedding. boi_token_index (`int`, *optional*, defaults to 255999) : The begin-of-image token index to wrap the image prompt. eoi_token_index (`int`, *optional*, defaults to 256000) : The end-of-image token index to wrap the image prompt. image_token_index (`int`, *optional*, defaults to `262144`) : The image token index used as a placeholder for input images. 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. ## Gemma3TextModel[[transformers.Gemma3TextModel]] #### transformers.Gemma3TextModel[[transformers.Gemma3TextModel]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L495) The bare Gemma3 Text Model outputting raw hidden-states without any specific head on to. This model inherits from [PreTrainedModel](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Gemma3TextModel.forwardhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L518[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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`).0[BaseModelOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPast) or `tuple(torch.FloatTensor)`A [BaseModelOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPast) 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. The [Gemma3TextModel](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3TextModel) 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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-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. **Parameters:** config ([Gemma3TextConfig](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3TextConfig)) : 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()](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** `[BaseModelOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPast) or `tuple(torch.FloatTensor)`` A [BaseModelOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPast) 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. ## Gemma3Model[[transformers.Gemma3Model]] #### transformers.Gemma3Model[[transformers.Gemma3Model]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L779) The Base Gemma3 model which consists of a vision backbone and a language model without language modeling head., This model inherits from [PreTrainedModel](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Gemma3Model.forwardhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L834[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "pixel_values", "val": ": torch.FloatTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "token_type_ids", "val": ": torch.LongTensor | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "**lm_kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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 [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor). See `Gemma3ImageProcessor.__call__()` for details ([Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) uses [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) for processing images). - **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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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)`. - **token_type_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) -- Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`: - 0 corresponds to a *sentence A* token, - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) - **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.text_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.text_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`).0`Gemma3ModelOutputWithPast` or `tuple(torch.FloatTensor)`A `Gemma3ModelOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. The [Gemma3Model](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Model) 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)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **past_key_values** (`~cache_utils.Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a [Cache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-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. Example: ```python >>> from PIL import Image >>> import httpx >>> from io import BytesIO >>> from transformers import AutoProcessor, Gemma3ForConditionalGeneration >>> model = Gemma3ForConditionalGeneration.from_pretrained("google/gemma32-3b-mix-224") >>> processor = AutoProcessor.from_pretrained("google/gemma32-3b-mix-224") >>> prompt = "Where is the cat standing?" >>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg" >>> with httpx.stream("GET", url) as response: ... image = Image.open(BytesIO(response.read())) >>> inputs = processor(images=image, text=prompt, return_tensors="pt") >>> # Generate >>> generate_ids = model.generate(**inputs,) >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Where is the cat standing?\nsnow" ``` **Parameters:** config ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) : 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()](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``Gemma3ModelOutputWithPast` or `tuple(torch.FloatTensor)`` A `Gemma3ModelOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. #### get_image_features[[transformers.Gemma3Model.get_image_features]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L799) Projects the last hidden state from the vision model into language model space. - **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. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) -- Last layer hidden-state of the first token of the sequence (classification token) after further processing through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns the classification token after processing through a linear layer and a tanh activation function. The linear layer weights are trained from the next sentence prediction (classification) objective during pretraining. - **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. **Parameters:** pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input images. Pixel values can be obtained using [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor). See `Gemma3ImageProcessor.__call__()` for details ([Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) uses [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) for processing images). **Returns:** `[BaseModelOutputWithPooling](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) or `tuple(torch.FloatTensor)`` A [BaseModelOutputWithPooling](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. #### get_placeholder_mask[[transformers.Gemma3Model.get_placeholder_mask]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L810) Obtains multimodal placeholder mask from `input_ids` or `inputs_embeds`, and checks that the placeholder token count is equal to the length of multimodal features. If the lengths are different, an error is raised. ## Gemma3ForCausalLM[[transformers.Gemma3ForCausalLM]] #### transformers.Gemma3ForCausalLM[[transformers.Gemma3ForCausalLM]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L592) The Gemma3 Model for causal language modeling. This model inherits from [PreTrainedModel](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Gemma3ForCausalLM.forwardhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L607[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "logits_to_keep", "val": ": int | torch.Tensor = 0"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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).0[CausalLMOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.CausalLMOutputWithPast) or `tuple(torch.FloatTensor)`A [CausalLMOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.CausalLMOutputWithPast) 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. The [Gemma3ForCausalLM](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ForCausalLM) 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.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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). 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. Example: ```python >>> from transformers import AutoTokenizer, Gemma3ForCausalLM >>> model = Gemma3ForCausalLM.from_pretrained("google/gemma-2-9b") >>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") >>> 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?" ``` **Parameters:** config ([Gemma3TextConfig](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3TextConfig)) : 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()](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** `[CausalLMOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.CausalLMOutputWithPast) or `tuple(torch.FloatTensor)`` A [CausalLMOutputWithPast](/docs/transformers/v5.6.0/ko/main_classes/output#transformers.modeling_outputs.CausalLMOutputWithPast) 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. ## Gemma3ForConditionalGeneration[[transformers.Gemma3ForConditionalGeneration]] #### transformers.Gemma3ForConditionalGeneration[[transformers.Gemma3ForConditionalGeneration]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L936) The Base Gemma3 model which consists of a vision backbone and a language model without language modeling head., This model inherits from [PreTrainedModel](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Gemma3ForConditionalGeneration.forwardhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L952[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "pixel_values", "val": ": torch.FloatTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "token_type_ids", "val": ": torch.LongTensor | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "logits_to_keep", "val": ": int | torch.Tensor = 0"}, {"name": "**lm_kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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 [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor). See `Gemma3ImageProcessor.__call__()` for details ([Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) uses [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) for processing images). - **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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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)`. - **token_type_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) -- Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`: - 0 corresponds to a *sentence A* token, - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) - **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.text_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.text_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).0`Gemma3CausalLMOutputWithPast` or `tuple(torch.FloatTensor)`A `Gemma3CausalLMOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. The [Gemma3ForConditionalGeneration](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ForConditionalGeneration) 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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). 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. Example: ```python >>> from PIL import Image >>> import httpx >>> from io import BytesIO >>> from transformers import AutoProcessor, Gemma3ForConditionalGeneration >>> model = Gemma3ForConditionalGeneration.from_pretrained("google/gemma-3-4b-it") >>> processor = AutoProcessor.from_pretrained("google/gemma-3-4b-it") >>> messages = [ ... { ... "role": "system", ... "content": [ ... {"type": "text", "text": "You are a helpful assistant."} ... ] ... }, ... { ... "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, clean_up_tokenization_spaces=False)[0] "user\nYou are a helpful assistant.\n\n\n\n\n\nWhere is the cat standing?\nmodel\nBased on the image, the cat is standing in a snowy area, likely outdoors. It appears to" ``` **Parameters:** config ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) : 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()](/docs/transformers/v5.6.0/ko/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``Gemma3CausalLMOutputWithPast` or `tuple(torch.FloatTensor)`` A `Gemma3CausalLMOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. ## Gemma3ForSequenceClassification[[transformers.Gemma3ForSequenceClassification]] #### transformers.Gemma3ForSequenceClassification[[transformers.Gemma3ForSequenceClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L1129) forwardtransformers.Gemma3ForSequenceClassification.forwardhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/gemma3/modeling_gemma3.py#L1145[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "pixel_values", "val": ": torch.FloatTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "token_type_ids", "val": ": torch.LongTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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 [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor). See `Gemma3ImageProcessor.__call__()` for details ([Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) uses [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) for processing images). - **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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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. - **token_type_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) -- Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`: - 0 corresponds to a *sentence A* token, - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) - **labels** (`torch.LongTensor` of shape `(batch_size,)`, *optional*) -- Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). - **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`).0`SequenceClassifierOutputWithPast` or `tuple(torch.FloatTensor)`A `SequenceClassifierOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs. The [Gemma3ForSequenceClassification](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ForSequenceClassification) 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) -- Classification (or regression if config.num_labels==1) loss. - **logits** (`torch.FloatTensor` of shape `(batch_size, config.num_labels)`) -- Classification (or regression if config.num_labels==1) scores (before SoftMax). - **past_key_values** (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a [Cache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). 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. Example of single-label classification: ```python >>> import torch >>> from transformers import AutoTokenizer, Gemma3ForSequenceClassification >>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-3-4b-it") >>> model = Gemma3ForSequenceClassification.from_pretrained("google/gemma-3-4b-it") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_class_id = logits.argmax().item() >>> model.config.id2label[predicted_class_id] ... >>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)` >>> num_labels = len(model.config.id2label) >>> model = Gemma3ForSequenceClassification.from_pretrained("google/gemma-3-4b-it", num_labels=num_labels) >>> labels = torch.tensor([1]) >>> loss = model(**inputs, labels=labels).loss >>> round(loss.item(), 2) ... ``` Example of multi-label classification: ```python >>> import torch >>> from transformers import AutoTokenizer, Gemma3ForSequenceClassification >>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-3-4b-it") >>> model = Gemma3ForSequenceClassification.from_pretrained("google/gemma-3-4b-it", problem_type="multi_label_classification") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_class_ids = torch.arange(0, logits.shape[-1])[torch.sigmoid(logits).squeeze(dim=0) > 0.5] >>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)` >>> num_labels = len(model.config.id2label) >>> model = Gemma3ForSequenceClassification.from_pretrained( ... "google/gemma-3-4b-it", num_labels=num_labels, problem_type="multi_label_classification" ... ) >>> labels = torch.sum( ... torch.nn.functional.one_hot(predicted_class_ids[None, :].clone(), num_classes=num_labels), dim=1 ... ).to(torch.float) >>> loss = model(**inputs, labels=labels).loss ``` **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](/docs/transformers/v5.6.0/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.6.0/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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 [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor). See `Gemma3ImageProcessor.__call__()` for details ([Gemma3Processor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Processor) uses [Gemma3ImageProcessor](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3ImageProcessor) for processing images). 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?](../glossary#attention-mask) 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?](../glossary#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](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.6.0/ko/internal/generation_utils#transformers.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. token_type_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) : Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`: - 0 corresponds to a *sentence A* token, - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*) : Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). 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:** ``SequenceClassifierOutputWithPast` or `tuple(torch.FloatTensor)`` A `SequenceClassifierOutputWithPast` 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 ([Gemma3Config](/docs/transformers/v5.6.0/ko/model_doc/gemma3#transformers.Gemma3Config)) and inputs.