o ¾e¦iP1ã@s6ddlmZmZddlmZGdd„deƒZdgZdS)é)ÚPreTrainedConfigÚlayer_type_validation)ÚRopeParameterscBsÄeZdZdZdZdgZdZdddddddd œZd gd gfd d gd gfd gd gfdœZddiZ                         !      dCd"e dBd#e dBd$e dBd%e dBd&e dBd'e dBd(e dBd)e dBd*e dBd+e dBd,e dBd-e dBd.e dBd/e dBd0e dBd1e dBd2e dBd3e dBd4e dBde dBd5e dBd6e dBd7e dBd8eee efBdBd9ee dBd:e dBd;e dBde dBd?e dBd@e dBf@‡fdAdB„ Z‡ZS)DÚ MiniMaxConfiga³ This is the configuration class to store the configuration of a [`MiniMaxModel`]. It is used to instantiate an MiniMax 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 MiniMax. [MiniMaxAI/MiniMax-Text-01-hf](https://huggingface.co/MiniMaxAI/MiniMax-Text-01-hf) 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 32000): Vocabulary size of the MiniMax model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`MiniMaxModel`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 14336): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 32): 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 `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 `8`. head_dim (`int`, *optional*, defaults to `hidden_size // num_attention_heads`): The attention head dimension. hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): The non-linear activation function (function or string) in the decoder. max_position_embeddings (`int`, *optional*, defaults to `4096*32`): The maximum sequence length that this model might ever be used with. MiniMax's sliding window attention allows sequence of up to 4096*32 tokens. 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 1e-05): 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`. pad_token_id (`int`, *optional*): The id of the padding token. bos_token_id (`int`, *optional*, defaults to 1): The id of the "beginning-of-sequence" token. eos_token_id (`int`, *optional*, defaults to 2): The id of the "end-of-sequence" token. tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether the model's input and output word embeddings should be tied. sliding_window (`int`, *optional*): Sliding window attention window size. If not specified, will default to `4096`. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. num_experts_per_tok (`int`, *optional*, defaults to 2): The number of experts to route per-token, can be also interpreted as the `top-k` routing parameter num_local_experts (`int`, *optional*, defaults to 8): Number of experts per Sparse MLP layer. 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 router_aux_loss_coef (`float`, *optional*, defaults to 0.001): The aux loss factor for the total loss. router_jitter_noise (`float`, *optional*, defaults to 0.0): Amount of noise to add to the router. 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`. layer_types (`list`, *optional*): Attention pattern for each layer. block_size (`int`, *optional*, defaults to 256): The length of each attention block, determining how queries, keys, and values are grouped and processed for intra- and inter-block attention. full_attn_alpha_factor (`float`, *optional*, defaults to 1): Weight for residual value in residual connection after normal attention. full_attn_beta_factor (`float`, *optional*, defaults to 1): Weight for hidden state value in residual connection after normal attention. linear_attn_alpha_factor (`float`, *optional*, defaults to 1): Weight for residual value in residual connection after lightning attention. linear_attn_beta_factor (`float`, *optional*, defaults to 1): Weight for hidden state value in residual connection after lightning attention. mlp_alpha_factor (`float`, *optional*, defaults to 1): Weight for residual value in residual connection after MLP. mlp_beta_factor (`float`, *optional*, defaults to 1): Weight for hidden state value in residual connection after MLP. ```python >>> from transformers import MiniMaxModel, MiniMaxConfig >>> # Initializing a MiniMax style configuration >>> configuration = MiniMaxConfig() >>> # Initializing a model from the MiniMax style configuration >>> model = MiniMaxModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```ÚminimaxÚpast_key_valuesg€„.AÚcolwiseÚrowwiseÚcolwise_gather_outputÚpacked_colwise)zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projzlayers.*.mlp.gatez!layers.*.mlp.experts.gate_up_projzlayers.*.mlp.experts.down_projÚ input_idsÚ inputs_embedsÚ hidden_statesÚattention_mask)Ú embed_tokensÚlayersÚnormÚ num_expertsÚnum_local_expertsé}éé8é éNÚsiluéç{®Gáz”?çñh㈵øä>TééFççü©ñÒMbP?éÚ vocab_sizeÚ hidden_sizeÚintermediate_sizeÚnum_hidden_layersÚnum_attention_headsÚnum_key_value_headsÚhead_dimÚ hidden_actÚmax_position_embeddingsÚinitializer_rangeÚ rms_norm_epsÚ use_cacheÚ pad_token_idÚ bos_token_idÚ eos_token_idÚtie_word_embeddingsÚsliding_windowÚattention_dropoutÚnum_experts_per_tokÚoutput_router_logitsÚrouter_aux_loss_coefÚrouter_jitter_noiseÚrope_parametersÚ layer_typesÚ block_sizeÚfull_attn_alpha_factorÚfull_attn_beta_factorÚlinear_attn_alpha_factorÚlinear_attn_beta_factorÚmlp_alpha_factorÚmlp_beta_factorc!" s||_| |_||_||_||_||_||_|dur|}||_||_| |_ | |_ | |_ ||_ ||_ ||_||_||_||_||_||_| |_||_||_||_||_||_||_||_||_||_| |_|jdursdd„t|jƒDƒ|_t |j|jƒ||_!t"ƒj#di|!¤ŽdS)NcSs$g|]}t|ddƒrdnd‘qS)rrÚfull_attentionÚlinear_attention)Úbool)Ú.0Úi©rGúo/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/minimax/configuration_minimax.pyÚ âsÿz*MiniMaxConfig.__init__..rG)$r#r+r$r%r&r'r3r(r*r,r-r.r4r)r5rr6r7r8r2r/r0r1r:r;r<r=r>r?r@rAÚrangerr9ÚsuperÚ__init__)"Úselfr#r$r%r&r'r(r)r*r+r,r-r.r/r0r1r2r3r4r5rr6r7r8r9r:r;r<r=r>r?r@rAÚkwargs©Ú __class__rGrHrL–sP$ ÿzMiniMaxConfig.__init__) rrrrrrNrrrrTNrrFNr rrFr!r NNr"rrrrrr)Ú__name__Ú __module__Ú __qualname__Ú__doc__Ú model_typeÚkeys_to_ignore_at_inferenceÚ default_thetaÚbase_model_tp_planÚbase_model_pp_planÚ attribute_mapÚintÚstrÚfloatrDrÚdictÚlistrLÚ __classcell__rGrGrOrHrsègù  ýÿßþýüûúùø ÷ ö õ ô óòñðïîíìëêéèç æåäãâá à!ßrN)Úconfiguration_utilsrrÚmodeling_rope_utilsrrÚ__all__rGrGrGrHÚs   S