`i!GdZddlmZddlmZddlmZddlmZej e Z GddeZ Gd d eZ Gd d eZd gZd S)zMllama model configuration)Optional)PretrainedConfig)rope_config_validation)loggingc eZdZdZdZdZ d$dedededededededededededede e ede e e ed effd! Z e d"efd#Z xZS)%MllamaVisionConfiga+ This is the configuration class to store the configuration of a [`MllamaVisionModel`]. It is used to instantiate an Mllama vision 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 Mllama-11B. e.g. [meta-llama/Llama-3.2-11B-Vision](https://huggingface.co/meta-llama/Llama-3.2-11B-Vision) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: hidden_size (`int`, *optional*, defaults to 1280): Dimensionality of the encoder layers and the pooler layer. hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`): 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. num_hidden_layers (`int`, *optional*, defaults to 32): Number of hidden layers in the Transformer encoder. num_global_layers (`int`, *optional*, defaults to 8): Number of global layers in the Transformer encoder. Vision model has a second transformer encoder, called global. num_attention_heads (`int`, *optional*, defaults to 16): 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 image. intermediate_size (`int`, *optional*, defaults to 5120): Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder. vision_output_dim (`int`, *optional*, defaults to 7680): Dimensionality of the vision model output. Includes output of transformer encoder with intermediate layers and global transformer encoder. image_size (`int`, *optional*, defaults to 448): The size (resolution) of each image *tile*. patch_size (`int`, *optional*, defaults to 14): The size (resolution) of each patch. norm_eps (`float`, *optional*, defaults to 1e-05): The epsilon used by the layer normalization layers. max_num_tiles (`int`, *optional*, defaults to 4): Maximum number of tiles for image splitting. intermediate_layers_indices (`list[int]`, *optional*, defaults to [3, 7, 15, 23, 30]): Indices of intermediate layers of transformer encoder from which to extract and output features. These output features are concatenated with final hidden state of transformer encoder. supported_aspect_ratios (`list[list[int]]`, *optional*): List of supported aspect ratios for image splitting. If not specified, the default supported aspect ratios are [[1, 1], [1, 2], [1, 3], [1, 4], [2, 1], [2, 2], [3, 1], [4, 1]] for `max_num_tiles=4`. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. Example: ```python >>> from transformers import MllamaVisionConfig, MllamaVisionModel >>> # Initializing a Llama config >>> config = MllamaVisionConfig() >>> # Initializing a vision model from the mllama-11b style configuration >>> model = MllamaVisionModel(config) >>> # Accessing the model configuration >>> configuration = model.config ```mllama_vision_model vision_configgelu rh㈵>N{Gz? hidden_size hidden_actnum_hidden_layersnum_global_layersnum_attention_heads num_channelsintermediate_sizevision_output_dim image_size patch_sizenorm_eps max_num_tilesintermediate_layers_indicessupported_aspect_ratiosinitializer_rangec z|/| dkrtdddgddgddgddgddgddgddgddgg}| gd} ||_||_||_||_||_| |_||_| |_| |_ ||_ | |_ | |_ ||_ ||_||_t!jdi|dS)Nrz;max_num_tiles must be 4 for default supported aspect ratiosr)r) ValueErrorrrrrrr rr!r$rr#r"attention_headsr%r&super__init__)selfrrrrrrrrr r!r"r#r$r%r&kwargs __class__s /home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/mllama/configuration_mllama.pyr2zMllamaVisionConfig.__init__\s & # *!! !^___()1v1v1v1v1vPQSTvXY[\W]`acd_e&f # & .*<*<*< '&$!2(!2$!2$+F(!2*  2'>$!2""6"""""returnc*t|jS)N)lenr%)r3s r6max_aspect_ratio_idz&MllamaVisionConfig.max_aspect_ratio_ids4/000r7)r r rrrrrrrrrrNNr)__name__ __module__ __qualname____doc__ model_typebase_config_keyintstrfloatrlistr2propertyr; __classcell__r5s@r6r r sZ< < |'J%O  !#!"#%!%!%;?=A#'!*#*#*#*# *#  *# ! *#*#*#*#*#*#*#*#&.d3i%8*#"*$tCy/!:*# !!*#*#*#*#*#*#X1S111X11111r7r c(eZdZdZdZdZ d*dedededededededede e ded ed!ed"e d#e d$e e ed%ed&ed'ed(e ef&fd) Z xZS)+MllamaTextConfiga( This is the configuration class to store the configuration of a [`MllamaTextModel`]. It is used to instantiate an Mllama text 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 Mllama-11B. e.g. [meta-llama/Llama-3.2-11B-Vision](https://huggingface.co/meta-llama/Llama-3.2-11B-Vision) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: vocab_size (`int`, *optional*, defaults to 128256): Vocabulary size of the Mllama text model. Defines the maximum number of different tokens that can be represented by the `inputs_ids` passed when calling [`MllamaTextModel`]. hidden_size (`int`, *optional*, defaults to 4096): Dimensionality of the embeddings and hidden states. hidden_act (`str` or `Callable`, *optional*, defaults to `"silu"`): The non-linear activation function (function or string) in the encoder and pooler. num_hidden_layers (`int`, *optional*, defaults to 40): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 32): Number of attention heads for each attention layer in the Transformer encoder. num_key_value_heads (`int`, *optional*, defaults to 8): This is the number of key_value heads that should be used to implement Grouped Query Attention. If not specified, will default to `num_attention_heads`. intermediate_size (`int`, *optional*, defaults to 14336): Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder. rope_theta (`float`, *optional*, defaults to `500000.0`): The base period of the RoPE embeddings. rope_scaling (`Dict`, *optional*): Dictionary containing the scaling configuration for the RoPE embeddings. NOTE: if you apply new rope type and you expect the model to work on longer `max_position_embeddings`, we recommend you to update this value accordingly. Expected contents: `rope_type` (`str`): The sub-variant of RoPE to use. Can be one of ['default', 'linear', 'dynamic', 'yarn', 'longrope', 'llama3'], with 'default' being the original RoPE implementation. `factor` (`float`, *optional*): Used with all rope types except 'default'. The scaling factor to apply to the RoPE embeddings. In most scaling types, a `factor` of x will enable the model to handle sequences of length x * original maximum pre-trained length. `original_max_position_embeddings` (`int`, *optional*): Used with 'dynamic', 'longrope' and 'llama3'. The original max position embeddings used during pretraining. `attention_factor` (`float`, *optional*): Used with 'yarn' and 'longrope'. The scaling factor to be applied on the attention computation. If unspecified, it defaults to value recommended by the implementation, using the `factor` field to infer the suggested value. `beta_fast` (`float`, *optional*): Only used with 'yarn'. Parameter to set the boundary for extrapolation (only) in the linear ramp function. If unspecified, it defaults to 32. `beta_slow` (`float`, *optional*): Only used with 'yarn'. Parameter to set the boundary for interpolation (only) in the linear ramp function. If unspecified, it defaults to 1. `short_factor` (`list[float]`, *optional*): Only used with 'longrope'. The scaling factor to be applied to short contexts (< `original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2 `long_factor` (`list[float]`, *optional*): Only used with 'longrope'. The scaling factor to be applied to long contexts (< `original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2 `low_freq_factor` (`float`, *optional*): Only used with 'llama3'. Scaling factor applied to low frequency components of the RoPE `high_freq_factor` (`float`, *optional*): Only used with 'llama3'. Scaling factor applied to high frequency components of the RoPE rms_norm_eps (`float`, *optional*, defaults to 1e-05): The epsilon used by the rms normalization layers. 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. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions. tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether to tie weight embeddings cross_attention_layers (`list[int]`, *optional*): Indices of the cross attention layers. If not specified, will default to [3, 8, 13, 18, 23, 28, 33, 38]. dropout (`float`, *optional*, defaults to 0): The dropout probability for self- and cross-attention layers. bos_token_id (`int`, *optional*, defaults to 128000): The id of the beginning of sentence token. eos_token_id (`int`, *optional*, defaults to 128001): The id of the end of sentence token. pad_token_id (`int`, *optional*, defaults to 128004): The id of the padding token. Example: ```python >>> from transformers import MllamaTextModel, MllamaTextConfig >>> # Initializing a Mllama text config >>> config = MllamaTextConfig() >>> # Initializing a model from the Mllama text configuration >>> model = MllamaTextModel(config) >>> # Accessing the model configuration >>> configuration = model.config ```mllama_text_model text_configsilu(rr8 NrrTFr vocab_sizerrrrnum_key_value_headsr rope_theta rope_scaling rms_norm_epsmax_position_embeddingsr& use_cachetie_word_embeddingscross_attention_layersdropout bos_token_id eos_token_id pad_token_idc @|gd}||_||_||_||_||_||_| |_| |_||_| |_ ||_ ||_ ||_ | |_ | |_t|t!jd||||d|dS)N)rr r,!&)rcrarbr^r.)rWrr_rrrXr&r]rYr[rr`rrZr\rr1r2)r3rWrrrrrXrrYrZr[r\r&r]r^r_r`rarbrcr4r5s r6r2zMllamaTextConfig.__init__s. " )%C%C%C "$!2&<#&#6 #6 !2"$(!2 $('>$t$$$ %%% 3         r7)rMrNrOrPrrrQrRNrrSrTFNrrTrUrV)r<r=r>r?r@rArBrCrDrdictboolrEr2rGrHs@r6rJrJscd d L%J#O! !##%#$!'#'+"'.#'$)6:""&,)1 1 1 1  1  1 ! 1 !1 1 1 tn1 1 "%1 !1 1 "1 !)c 3!1 "#1 $%1 &'1 (sm)1 1 1 1 1 1 1 1 1 1 r7rJc@eZdZdZdZddiZeedZ d fd Z xZ S) MllamaConfiga This is the configuration class to store the configuration of a [`MllamaForConditionalGeneration`]. It is used to instantiate an Mllama 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 Mllama-9B. e.g. [meta-llama/Llama-3.2-11B-Vision](https://huggingface.co/meta-llama/Llama-3.2-11B-Vision) 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 (`Union[AutoConfig, dict]`, *optional*, defaults to `MllamaVisionConfig`): The config object or dictionary of the vision backbone. text_config (`Union[AutoConfig, dict]`, *optional*, defaults to `MllamaTextConfig`): The config object or dictionary of the text backbone. image_token_index (`int`, *optional*, defaults to 128256): The image token index to encode the image prompt. Example: ```python >>> from transformers import MllamaForConditionalGeneration, MllamaConfig, MllamaVisionConfig, MllamaTextConfig >>> # Initializing a CLIP-vision config >>> vision_config = MllamaVisionConfig() >>> # Initializing a Llama config >>> text_config = MllamaTextConfig() >>> # Initializing a mllama-11b style configuration >>> configuration = MllamaConfig(vision_config, text_config) >>> # Initializing a model from the mllama-11b style configuration >>> model = MllamaForConditionalGeneration(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```mllamaimage_token_idimage_token_index)rLr NrMc |.t|_tdnCt |t rtdi||_nt |tr||_||_|.t|_tdnCt |t rtdi||_nt |tr||_tj di|dS)Nz9vision_config is None, using default mllama vision configz5text_config is None, using default mllama text configr.) r r loggerinfo isinstancerjrprJrLr1r2)r3r rLrpr4r5s r6r2zMllamaConfig.__init__Ys  !3!5!5D  KKS T T T T  t , , /!3!D!Dm!D!DD    '9 : : /!.D !2  /11D  KKO P P P P  T * * +/>>+>>D    %5 6 6 +*D ""6"""""r7)NNrM) r<r=r>r?r@ attribute_maprJr sub_configsr2rGrHs@r6rmrm+sw% % NJ-M#3EWXXK ##########r7rmN)r?typingrconfiguration_utilsrmodeling_rope_utilsrutilsr get_loggerr<rrr rJrm__all__r.r7r6r}s! 333333999999  H % %p1p1p1p1p1)p1p1p1f[ [ [ [ [ '[ [ [ |G#G#G#G#G##G#G#G#T  r7