o ۾i:@sDdZddlmZmZddlmZeeZGdddeZ dgZ dS)zQwen3-Next model configuration)PretrainedConfiglayer_type_validation)loggingcseZdZdZdZdgZdddddddddddddd Zdgdgfd d gd gfd gd gfd Z                          !  d$fd"d# ZZ S)%Qwen3NextConfiga6$ This is the configuration class to store the configuration of a [`Qwen3NextModel`]. It is used to instantiate a Qwen3-Next model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of Qwen3-Next-80B-A3B-Instruct [Qwen/Qwen3-Next-80B-A3B-Instruct](https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct). 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 151936): Vocabulary size of the model. Defines the number of different tokens that can be represented by the `inputs_ids`. hidden_size (`int`, *optional*, defaults to 2048): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 5632): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 48): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. 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 checkout [this paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`. hidden_act (`str`, *optional*, defaults to `"silu"`): The non-linear activation function in the decoder. max_position_embeddings (`int`, *optional*, defaults to 32768): 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 1e-06): 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 the model's input and output word embeddings should be tied. rope_parameters (`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_theta` (`float`): The base period of the RoPE embeddings. `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 `partial_rotary_factor` (`float`, *optional*, defaults to 0.25): Percentage of the query and keys which will have rotary embedding. attention_bias (`bool`, *optional*, defaults to `False`): Whether to use a bias in the query, key, value and output projection layers during self-attention. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. head_dim (`int`, *optional*, defaults to 256): Projection weights dimension in multi-head attention. linear_conv_kernel_dim (`int`, *optional*, defaults to 4): Kernel size of the convolution used in linear attention layers. linear_key_head_dim (`int`, *optional*, defaults to 128): Dimension of each key head in linear attention. linear_value_head_dim (`int`, *optional*, defaults to 128): Dimension of each value head in linear attention. linear_num_key_heads (`int`, *optional*, defaults to 16): Number of key heads used in linear attention layers. linear_num_value_heads (`int`, *optional*, defaults to 32): Number of value heads used in linear attention layers. decoder_sparse_step (`int`, *optional*, defaults to 1): The frequency of the MoE layer. moe_intermediate_size (`int`, *optional*, defaults to 512): Intermediate size of the routed expert. shared_expert_intermediate_size (`int`, *optional*, defaults to 512): Intermediate size of the shared expert. num_experts_per_tok (`int`, *optional*, defaults to 10): Number of selected experts. num_experts (`int`, *optional*, defaults to 512): Number of routed experts. norm_topk_prob (`bool`, *optional*, defaults to `True`): Whether to normalize the topk probabilities. 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, including load balancing loss and router z-loss. router_aux_loss_coef (`float`, *optional*, defaults to 0.001): The aux loss factor for the total loss. mlp_only_layers (`list[int]`, *optional*, defaults to `[]`): Indicate which layers use Qwen3NextMLP rather than Qwen3NextSparseMoeBlock The list contains layer index, from 0 to num_layers-1 if we have num_layers layers If `mlp_only_layers` is empty, `decoder_sparse_step` is used to determine the sparsity. layer_types (`list[str]`, *optional*): Types of each layer (attention or linear). ```python >>> from transformers import Qwen3NextModel, Qwen3NextConfig >>> # Initializing a Qwen3Next style configuration >>> configuration = Qwen3NextConfig() >>> # Initializing a model from the Qwen3-Next-80B-A3B style configuration >>> model = Qwen3NextModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` qwen3_nextpast_key_valuescolwiserowwise) zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projz layers.*.mlp.experts.*.gate_projzlayers.*.mlp.experts.*.up_projz layers.*.mlp.experts.*.down_projz%layers.*.mlp.shared_experts.gate_projz#layers.*.mlp.shared_experts.up_projz%layers.*.mlp.shared_experts.down_projzlayers.*.mlp.gate_projzlayers.*.mlp.up_projzlayers.*.mlp.down_proj input_ids inputs_embeds hidden_statesattention_mask) embed_tokenslayersnormQ0silu{Gz?ư>TFN  MbP?c $ s^|durg}tjd d| i| ||_||_||_||_||_||_||_||_ | |_ | |_ | |_ | dd}!|!p?| p?ddi} | dd}"d| vrN|"| d<| dd}#d| vr\|#| d<| |_|#|_||_||_||_||_|jdur~d d t|jD|_t|j||_||_||_||_||_||_||_||_||_||_||_ ||_!||_"||_#dS) Ntie_word_embeddings rope_scaling rope_typedefault rope_thetag@partial_rotary_factorg?cSs$g|]}t|ddrdndqS)r rlinear_attentionfull_attention)bool).0ir/^/home/ubuntu/.local/lib/python3.10/site-packages/vllm/transformers_utils/configs/qwen3_next.py sz,Qwen3NextConfig.__init__..r/)$super__init__ vocab_sizemax_position_embeddings hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headsnum_key_value_heads hidden_actinitializer_range rms_norm_eps use_cachepoprope_parametersr)attention_biasattention_dropouthead_dim layer_typesrangerlinear_conv_kernel_dimlinear_key_head_dimlinear_value_head_dimlinear_num_key_headslinear_num_value_headsdecoder_sparse_stepmoe_intermediate_sizeshared_expert_intermediate_sizenum_experts_per_tok num_expertsnorm_topk_proboutput_router_logitsrouter_aux_loss_coefmlp_only_layers)$selfr4r6r7r8r9r:r;r5r<r=r>r$r@rArBrCrFrGrHrIrJrKrLrMrNrOrPrQrRrSrDkwargsr%r(r) __class__r/r0r3s^#      zQwen3NextConfig.__init__)rrrrrrrrrrTFNFrrrrrrrr r!r!r"r!TFr#NN) __name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inferencebase_model_tp_planbase_model_pp_planr3 __classcell__r/r/rVr0rsn    rN) r[ transformers.configuration_utilsrrtransformers.utilsr get_loggerrXloggerr__all__r/r/r/r0s   }