o ei(@sxdZddlmZddlmZddlmZddlmZe e Z GdddeZ Gd d d eZ Gd d d eZd gZd S)zDBRX model configuration)Any)PreTrainedConfig)RopeParameters)loggingc sBeZdZdZdZ   d dededBded effd d ZZ S) DbrxAttentionConfigaConfiguration class for Dbrx Attention. [`DbrxAttention`] class. It is used to instantiate attention layers according to the specified arguments, defining the layers architecture. Configuration objects inherit from [`PreTrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PreTrainedConfig`] for more information. Args: attn_pdrop (`float`, *optional*, defaults to 0.0): The dropout probability for the attention layers. clip_qkv (`float`, *optional*): If set, clip the queries, keys, and values in the attention layer to this value. kv_n_heads (`int`, *optional*, defaults to 1): For grouped_query_attention only, allow user to specify number of kv heads. attn_configN attn_pdropclip_qkv kv_n_headskwargsc s(tjdi|||_||_||_dS)N)super__init__r r r )selfr r r r __class__ri/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/dbrx/configuration_dbrx.pyr.s zDbrxAttentionConfig.__init__)r Nr ) __name__ __module__ __qualname____doc__base_config_keyfloatintrr __classcell__rrrrrsrcsdeZdZdZdZ        dd edBd ed ed ededBdededBdeffdd Z Z S) DbrxFFNConfiga|Configuration class for Dbrx FFN. [`DbrxFFN`] class. It is used to instantiate feedforward layers according to the specified arguments, defining the layers architecture. Configuration objects inherit from [`PreTrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PreTrainedConfig`] for more information. Args: ffn_act_fn (`dict`, *optional*, defaults to `None`): A dict specifying activation function for the FFN. The dict should have a key 'name' with the value being the name of the activation function along with any additional keyword arguments. If `None`, then set to `{"name": "silu"}`. ffn_hidden_size (`int`, *optional*, defaults to 3584): The hidden size of the feedforward network. moe_num_experts (`int`, *optional*, defaults to 4): The number of experts in the mixture of experts layer. moe_top_k (`int`, *optional*, defaults to 1): The number of experts to use in the mixture of experts layer. moe_jitter_eps (`float`, *optional*, defaults to `None`): If not `None`, the jitter epsilon for the mixture of experts layer. moe_loss_weight (`float`, *optional*, defaults to 0.01): The loss weight for the mixture of experts layer. moe_normalize_expert_weights (`float`, *optional*, defaults to 1.0): The normalization factor for the expert weights. ffn_configNr {Gz?? ffn_act_fnffn_hidden_sizemoe_num_experts moe_top_kmoe_jitter_epsmoe_loss_weightmoe_normalize_expert_weightsrc  st|dur ddi}||_||_||_||_||_||_||_||_ dD] } | | vr2| | q't | dkr@t d| dS)Nnamesilu) model_typeattn_implementationexperts_implementationtransformers_version _commit_hash torch_dtypedtyperzFound unknown kwargs=) rr hidden_sizer%r&r'r(r)r*r+poplen ValueError) rr5r%r&r'r(r)r*r+rkrrrrRs$   zDbrxFFNConfig.__init__)r Nr!r"r Nr#r$) rrrrrdictrrrrrrrrrr;s8  rc%seZdZdZdZeedZdddddZ          d"de dBde dBde dBde dBde dBde dBde dBdedBdedBde dBde dBde dBde e ee fBdBde dBde dBde dBde dBdef$fd d! ZZS)# DbrxConfiga This is the configuration class to store the configuration of a [`DbrxModel`]. It is used to instantiate a Dbrx model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a different configuration to that of the [transformers-community/dbrx-instruct](https://huggingface.co/transformers-community/dbrx-instruct) architecture. Note: this link points to a re-upload; the original repository is closed. Configuration objects inherit from [`PreTrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PreTrainedConfig`] for more information. Args: d_model (`int`, *optional*, defaults to 2048): Dimensionality of the embeddings and hidden states. n_heads (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. n_layers (`int`, *optional*, defaults to 24): Number of hidden layers in the Transformer encoder. max_seq_len (`int`, *optional*, defaults to 2048): The maximum sequence length of the model. vocab_size (`int`, *optional*, defaults to 32000): Vocabulary size of the Dbrx model. Defines the maximum number of different tokens that can be represented by the `inputs_ids` passed when calling [`DbrxModel`]. resid_pdrop (`float`, *optional*, defaults to 0.0): The dropout probability applied to the attention output before combining with residual. emb_pdrop (`float`, *optional*, defaults to 0.0): The dropout probability for the embedding layer. attn_config (`dict`, *optional*): A dictionary used to configure the model's attention module. ffn_config (`dict`, *optional*): A dictionary used to configure the model's FFN module. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. output_router_logits (`bool`, *optional*, defaults to `False`): Whether or not the router logits should be returned by the model. Enabling this will also allow the model to output the auxiliary loss. See [here]() for more details. Example: ```python >>> from transformers import DbrxConfig, DbrxModel >>> # Initializing a Dbrx configuration >>> configuration = DbrxConfig(n_layers=2, d_model=256, n_heads=8, vocab_size=128) >>> # Initializing a model (with random weights) from the configuration >>> model = DbrxModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` dbrx)rrn_headsd_modeln_layers max_seq_len)num_attention_headsr5num_hidden_layersmax_position_embeddings}r NT{Gz?F vocab_size resid_pdrop emb_pdroprr use_cacheinitializer_rangeoutput_router_logitsrope_parameters pad_token_id bos_token_id eos_token_idtie_word_embeddingsrc s|dur t|_nt|trtdi||_n||_| dur#t|_nt| tr1tdi| |_n| |_||_||_||_||_ ||_ ||_ ||_ | |_ | |_| |_|jj|_|r]td||_||_||_||_| |_tjdi|dS)Nz5tie_word_embeddings is not supported for DBRX models.r)rr isinstancer:rrr>r=r?r@rIrJrKrLrMrNr num_key_value_headsr8rPrQrRrSrOrr)rr>r=r?r@rIrJrKrrrLrMrNrOrPrQrRrSrrrrrs:     zDbrxConfig.__init__)rDrErFrDrGr r NNTrHFNNNNF)rrrrr.rr sub_configs attribute_maprrboolrr:strrrrrrrrr;ys~7       r;N)rtypingrconfiguration_utilsrmodeling_rope_utilsrutilsr get_loggerrloggerrrr;__all__rrrrs     !> {