o i@sDddlmZddlmZGdddeZGdddeZddgZdS) )PretrainedConfig) Qwen2ConfigcsveZdZdZdZ           d dededededededededededeffdd ZZ S)!Ovis2VisionConfiga This is the configuration class to store the configuration of a [`Ovis2VisionModel`]. It is used to instantiate a Ovis2VisionModel model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of Ovis2. Args: hidden_size (`int`, *optional*, defaults to 1024): Dimensionality of the encoder layers and the pooler layer. intermediate_size (`int`, *optional*, defaults to 2816): Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder. num_hidden_layers (`int`, *optional*, defaults to 24): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 8): Number of attention heads for each attention layer in the Transformer encoder. num_channels (`int`, *optional*, defaults to 3): Number of channels in the input images. image_size (`int`, *optional*, defaults to 224): The size (resolution) of each image. patch_size (`int`, *optional*, defaults to 14): The size (resolution) of each patch. rms_norm_eps (`float`, *optional*, defaults to 1e-05): The epsilon used by the RMSNorm layers. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. qkv_bias (`bool`, *optional*, defaults to `False`): Whether to add a learnable bias to the query, key, and value sequences at each attention head. mlp_bias (`bool`, *optional*, defaults to `False`): Whether to add a learnable bias to the MLP layers. hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` `"quick_gelu"` are supported. vocab_size (`int`, *optional*, defaults to 16384): Vocabulary size of the Vision Transformer. hidden_stride (`int`, *optional*, defaults to 1): The stride of the hidden layer in the Vision Transformer. num_visual_indicator_tokens (`int`, *optional*, defaults to 5): Number of visual indicator tokens. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated normal initializer for initializing all weight matrices. tokenize_function (`str`, *optional*, defaults to `"softmax"`): The function used to tokenize the visual indicator tokens. vision_config rh㈵>Fsilu@{Gz?softmax hidden_sizeintermediate_sizenum_hidden_layersnum_attention_heads num_channels image_size patch_size rms_norm_epsattention_dropoutqkv_biasmlp_biasc s|tjdi|||_||_||_||_||_||_||_| |_ | |_ | |_ | |_ ||_ | |_||_||_||_||_dS)N)super__init__rrrrrrrr hidden_actrrr vocab_size hidden_stridenum_visual_indicator_tokenstokenize_functioninitializer_range)selfrrrrrrrrrrrr#r$r%r&r(r'kwargs __class__r a/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/ovis2/configuration_ovis2.pyr"Bs$ zOvis2VisionConfig.__init__)rrr r rr r r rFFrrrrrr) __name__ __module__ __qualname____doc__base_config_keyintfloatboolr" __classcell__r r r+r-rsV+    rcs@eZdZdZdZeedZdddgdddffd d ZZ S) Ovis2ConfigaQ This is the configuration class to store the configuration of a [`Ovis2ForConditionalGeneration`]. It is used to instantiate a Ovis2 model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of Ovis2. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. e.g. [thisisiron/Ovis2-1B-hf](https://huggingface.co/thisisiron/Ovis2-1B-hf) Args: vision_config (`Union[AutoConfig, dict]`, *optional*, defaults to `Ovis2VisionConfig`): The config object or dictionary of the vision backbone. text_config (`Union[AutoConfig, dict]`, *optional*, defaults to `Qwen2Config`): The config object or dictionary of the text backbone. image_token_id (`int`, *optional*, defaults to 151665): The image token id to encode the image prompt. visual_indicator_token_ids (`List[int]`, *optional*, defaults to `[151666, 151667, 151668, 151669, 151670]`): The visual indicator token ids to encode the image prompt. vocab_size (`int`, *optional*, defaults to 151643): Vocabulary size of the text model. hidden_size (`int`, *optional*, defaults to 1536): Dimensionality of the encoder layers and the pooler layer. ```python >>> from transformers import Ovis2ForConditionalGeneration, Ovis2Config >>> # Initializing a Ovis2 style configuration >>> configuration = Ovis2Config() >>> # Initializing a model from the Ovis2-2B style configuration >>> model = Ovis2ForConditionalGeneration(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` ovis2) text_configrNiqP)irPisPitPiuPivPi[Pic st|trtdi||_nt|tr||_|dur"tt|d|_t|tr0tdi||_nt|tr9||_n|durAt|_||_||_||_ ||_ t j di|dS)N)r&r ) isinstancedictrrlenrr9r$rimage_token_idvisual_indicator_token_idsr!r")r)rr9r=r>r$rr*r+r r-r"s"    zOvis2Config.__init__) r.r/r0r1 model_typerr sub_configsr"r6r r r+r-r7ms& r7N)configuration_utilsrqwen2.configuration_qwen2rrr7__all__r r r r-s  Y I