o ei#@s^ddlmZddlmZddlmZmZeeZ GdddeZ GdddeZ ddgZ d S) )PretrainedConfig)logging)CONFIG_MAPPING AutoConfigcsPeZdZdZdZddddddZ     dfdd ZZS)AudioFlamingo3EncoderConfiga This is the configuration class to store the configuration of an [`AudioFlamingo3Encoder`]. It is used to instantiate an AudioFlamingo3 audio encoder according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the audio encoder of the AudioFlamingo3 architecture. e.g. [nvidia/audio-flamingo-3-hf](https://huggingface.co/nvidia/audio-flamingo-3-hf) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: num_mel_bins (`int`, *optional*, defaults to 128): Number of mel features used per input features. Should correspond to the value used in the `AudioFlamingo3Processor` class. num_hidden_layers (`int`, *optional*, defaults to 32): Number of encoder layers. num_attention_heads (`int`, *optional*, defaults to 20): Number of attention heads for each attention layer in the Transformer encoder. intermediate_size (`int`, *optional*, defaults to 5120): Dimensionality of the "intermediate" (often named feed-forward) layer in encoder. layerdrop (`float`, *optional*, defaults to 0.0): The LayerDrop probability for the encoder. See the [LayerDrop paper](https://huggingface.co/papers/1909.11556) for more details. activation_function (`str`, *optional*, defaults to `"gelu"`): The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`, `"relu"`, `"silu"` and `"gelu_new"` are supported. hidden_size (`int`, *optional*, defaults to 1280): Dimensionality of the layers. dropout (`float`, *optional*, defaults to 0.0): The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. activation_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for activations inside the fully connected layer. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. scale_embedding (`bool`, *optional*, defaults to `False`): Scale embeddings by dividing by sqrt(hidden_size). max_source_positions (`int`, *optional*, defaults to 1500): The maximum sequence length of log-mel filter-bank features that this model might ever be used with. Example: ```python >>> from transformers import AudioFlamingo3EncoderConfig, AudioFlamingo3Encoder >>> # Initializing an AudioFlamingo3EncoderConfig >>> configuration = AudioFlamingo3EncoderConfig() >>> # Initializing an AudioFlamingo3Encoder (with random weights) >>> model = AudioFlamingo3Encoder(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```audioflamingo3_encoder hidden_sizenum_hidden_layersnum_attention_headsintermediate_size layerdrop)d_modelencoder_layersencoder_attention_headsencoder_ffn_dimencoder_layerdrop gelu{Gz?Fc sjtjdi|||_||_||_||_||_||_| |_| |_ ||_ | |_ ||_ ||_| |_ | |_dS)N)super__init__ num_mel_binsr r r r dropoutattention_dropoutactivation_dropoutactivation_functioninitializer_ranger scale_embeddingmax_source_positions)selfrr r r r r#r r r!r"r$r%r&kwargs __class__r}/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/audioflamingo3/configuration_audioflamingo3.pyr\s z$AudioFlamingo3EncoderConfig.__init__) rrrrrrrrrrrFr)__name__ __module__ __qualname____doc__ model_type attribute_mapr __classcell__rrr)r+rs.9 rcs:eZdZdZdZeedZ     d fdd ZZ S) AudioFlamingo3Configa This is the configuration class to store the configuration of an [`AudioFlamingo3ForConditionalGeneration`]. It is used to instantiate an AudioFlamingo3 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 AudioFlamingo3. e.g. [nvidia/audio-flamingo-3-hf](https://huggingface.co/nvidia/audio-flamingo-3-hf) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: audio_config (`Union[AudioFlamingo3EncoderConfig, dict]`, *optional*, defaults to `AudioFlamingo3EncoderConfig`): The config object or dictionary of the audio backbone. text_config (`Union[AutoConfig, dict]`, *optional*, defaults to `Qwen2Config`): The config object or dictionary of the text backbone. audio_token_id (`int`, *optional*, defaults to 151669): The audio token index to encode the audio prompt. projector_hidden_act (`str`, *optional*, defaults to `"gelu"`): Activation function used in the projector. projector_bias (`bool`, *optional*, defaults to `True`): Whether to include bias terms in the projector. Example: ```python >>> from transformers import AudioFlamingo3ForConditionalGeneration, AudioFlamingo3Config, AudioFlamingo3EncoderConfig, Qwen2Config >>> # Initializing an AudioFlamingo3Encoder config >>> audio_config = AudioFlamingo3EncoderConfig() >>> # Initializing a Qwen2 config >>> text_config = Qwen2Config() >>> # Initializing an AudioFlamingo3 configuration >>> configuration = AudioFlamingo3Config(audio_config, text_config) >>> # Initializing a model from the audioflamingo3 style configuration >>> model = AudioFlamingo3ForConditionalGeneration(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```audioflamingo3) audio_config text_configNuPrTc s||_t|tr|dd|d<t|ddi|}n |dur%td}||_t|trA|dd|d<t|ddi|}n |durJtd}||_||_||_t j di|dS)Nr0rqwen2r) audio_token_id isinstancedictgetrr5r6projector_hidden_actprojector_biasrr)r'r5r6r9r=r>r(r)rr+rs     zAudioFlamingo3Config.__init__)NNr7rT) r,r-r.r/r0rr sub_configsrr2rrr)r+r3s+r3N) configuration_utilsrutilsrautorr get_loggerr,loggerrr3__all__rrrr+s   g R