o ۷i@sPdZddlmZddlmZddlmZmZee Z GdddeZ dgZ dS) zVipLlava model configuration)PretrainedConfig)logging)CONFIG_MAPPING AutoConfigcsJeZdZdZdZddiZeedZddddd gd d ffd d ZZ S)VipLlavaConfiga This is the configuration class to store the configuration of a [`VipLlavaForConditionalGeneration`]. It is used to instantiate an VipLlava 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 VipLlava-9B. e.g. [ybelkada/vip-llava-7b-hf](https://huggingface.co/ybelkada/vip-llava-7b-hf) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: vision_config (`VipLlavaVisionConfig`, *optional*): Custom vision config or dict text_config (`Union[AutoConfig, dict]`, *optional*): The config object of the text backbone. Can be any of `LlamaConfig` or `MistralConfig`. image_token_index (`int`, *optional*, defaults to 32000): The image token index to encode the image prompt. projector_hidden_act (`str`, *optional*, defaults to `"gelu"`): The activation function used by the multimodal projector. projector_layernorm_eps (`float`, *optional*, defaults to 1e-05): The layer norm epsilon of the projector layernorm vision_feature_layers (`Union[int, list[int]]`, *optional*, defaults to `[-2, -5, -8, -11, 6]`): The vision feature layer, or list of layers to select the vision features from. image_seq_length (`int`, *optional*, defaults to 576): Sequence length of one image embedding. Example: ```python >>> from transformers import VipLlavaForConditionalGeneration, VipLlavaConfig, CLIPVisionConfig, LlamaConfig >>> # Initializing a CLIP-vision config >>> vision_config = CLIPVisionConfig() >>> # Initializing a Llama config >>> text_config = LlamaConfig() >>> # Initializing a VipLlava vipllava-7b style configuration >>> configuration = VipLlavaConfig(vision_config, text_config) >>> # Initializing a model from the vipllava-7b style configuration >>> model = VipLlavaForConditionalGeneration(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```vipllavaimage_token_idimage_token_index) text_config vision_configN}gelugh㈵>)iii@c s||_||_||_||_||_||_t|jtr-|dd|d<t |dd i||_n|dur@t dddddddd d d |_t|trY|dd |d<t |dd i|}n |durbt d }||_ t j d i|dS)N model_typeclip_vision_modeliiiPr i)intermediate_size hidden_size patch_size image_sizenum_hidden_layersnum_attention_heads vocab_sizeprojection_dimllama) r projector_hidden_actprojector_layernorm_epsvision_feature_layersimage_seq_lengthr isinstancedictgetrr super__init__) selfr r r r!r"r#r$kwargs __class__r i/home/ubuntu/vllm_env/lib/python3.10/site-packages/transformers/models/vipllava/configuration_vipllava.pyr)Ns6    zVipLlavaConfig.__init__) __name__ __module__ __qualname____doc__r attribute_mapr sub_configsr) __classcell__r r r,r.rs/ rN) r2configuration_utilsrutilsrautorr get_loggerr/loggerr__all__r r r r.s    b