qi98ddlmZddlmZGddeZdgZy))PreTrainedConfig)RopeParametersc(feZdZdZdZdgZdddddddZd gd gfd d gd gfd gd gfd Zddddddddddddddddgdddfdedzd edzd!edzd"edzd#edzd$edzd%edzd&e dzd'e dzd(edzd)e dzd*e dzd+e dzd,e dzd-e e e e fzdzd.edzd/eedzd0edzd1e dzf&fd2 ZxZS)3 GlmConfigau This is the configuration class to store the configuration of a [`GlmModel`]. It is used to instantiate an Glm 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 Glm-4-9b-chat. e.g. [THUDM/glm-4-9b-chat](https://huggingface.co/THUDM/glm-4-9b-chat) 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 151552): Vocabulary size of the Glm model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`GlmModel`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 13696): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 40): Number of hidden layers in the Transformer decoder. num_attention_heads (`int`, *optional*, defaults to 32): Number of attention heads for each attention layer in the Transformer decoder. num_key_value_heads (`int`, *optional*, defaults to 2): This is the number of key_value heads that should be used to implement Grouped Query Attention. If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details, check out [this paper](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `num_attention_heads`. head_dim (`int`, *optional*, defaults to 128): The attention head dimension. hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): The legacy activation function. It is overwritten by the `hidden_activation`. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. 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. rms_norm_eps (`float`, *optional*, defaults to 1.5625e-07): The epsilon used by the rms normalization layers. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True`. tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether to tie weight embeddings rope_parameters (`RopeParameters`, *optional*): Dictionary containing the configuration parameters for the RoPE embeddings. The dictionary should contain a value for `rope_theta` and optionally parameters used for scaling in case you want to use RoPE with longer `max_position_embeddings`. pad_token_id (`int`, *optional*, defaults to 151329): Padding token id. eos_token_id (`int` | `list`, *optional*, defaults to `[151329, 151336, 151338]`): End of stream token id. bos_token_id (`int`, *optional*): Beginning of stream token id. attention_bias (`bool`, defaults to `False`, *optional*, defaults to `True`): Whether to use a bias in the query, key, value and output projection layers during self-attention. ```python >>> from transformers import GlmModel, GlmConfig >>> # Initializing a Glm glm-4-9b-chat style configuration >>> configuration = GlmConfig() >>> # Initializing a model from the glm-4-9b-chat style configuration >>> model = GlmModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```glmpast_key_valuescolwiserowwisecolwise_gather_outputrowwise_split_input)zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projzlayers.*.mlp.gate_up_projzlayers.*.mlp.down_proj input_ids inputs_embeds hidden_statesattention_mask) embed_tokenslayersnormiPii5( silugig{Gz?gh㈵>TFN!O)ri(Oi*O vocab_size hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headsnum_key_value_headshead_dim hidden_actattention_dropoutmax_position_embeddingsinitializer_range rms_norm_eps use_cachetie_word_embeddingsrope_parameters pad_token_id eos_token_id bos_token_idattention_biasc R||_| |_||_||_||_||_||_||_||_| |_ | |_ | |_ ||_ | |_ ||_|jdd||_||_||_||_t)|Tdi|y)Npartial_rotary_factorg?)rr#rrrrr rr!r$r%r&r,r"r( setdefaultr'r)r+r*super__init__)selfrrrrrrr r!r"r#r$r%r&r'r(r)r*r+r,kwargs __class__s [/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/glm/configuration_glm.pyr2zGlmConfig.__init__hs.%'>$&!2!2#6   #6 $!2(",!2.137#6 ((( "6")__name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inferencebase_model_tp_planbase_model_pp_planintstrfloatboolrdictlistr2 __classcell__)r5s@r6rrs@ DJ#4"5%.%.%.%.%<"7 &(9:#%568IJ!"_$56"("&(-(**,*+"!'*-.4*.%2!%+0MQ#))A#'&*),#$J,#4Z,#: ,# : ,# !4Z ,#!4Z,#*,#$J,#!4<,#"%t,#!4<,#dl,#$;,#"D[,# ($sN/B*CCdJ!,#"Dj#,#$3i$&%,#&Dj',#(t ),#,#r7rN)configuration_utilsrmodeling_rope_utilsrr__all__r/r7r6rJs&"41# #D -r7