.`i8fddlmZGddeZGddeZGddeZdS) )PretrainedConfigcXeZdZdZdZ dfd ZxZS)HunYuanVLVisionConfig hunyuan_vl vision_configgelubilinearh㈵>rNT@ c tjdi|||_||_||_||_||_||_|s||_n||_| |_ | |_ | |_ | |_ | |_ ||_||_||_||_||_||_||_||_||_||_||_||_dS)N)super__init__ hidden_act hidden_sizeintermediate_sizeinterpolate_modelearnable_mlp_pooling_sizenum_attention_headsnum_key_value_heads num_channelsnum_hidden_layersout_hidden_size patch_sizeremove_prenormspatial_merge_sizetemporal_patch_size rms_norm_epsresize_resolutionimg_max_token_nummax_image_sizemin_image_sizevideo_max_image_sizevideo_min_image_sizeanyres_vit_max_image_sizemax_vit_seq_lentext_hidden_size)selfrrrr r+r!r"r#r$r%r&r'r(r)r*r,r-r.r0r1r/r2r3r4kwargs __class__s ~/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/vllm/transformers_utils/configs/hunyuan_vl.pyrzHunYuanVLVisionConfig.__init__ s8 ""6"""$&!2 0*D'#6 " ;':D $ $':D $(!2.$,"4#6 (!2!2,,$8!$8!)B&. 0)rr r r r rr Nrrrr Trrrrrrrrrrr)__name__ __module__ __qualname__ model_typebase_config_keyr __classcell__r7s@r8rr sJ%O##$   "&391919191919191919191r9rcfeZdZdZdZdgZ ddeffd ZdZxZ S)HunYuanVLTextConfiga* This is the configuration class to store the configuration of a [`HunYuanVLTextConfig`]. It is used to instantiate an HunYuan 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 HunYuan-7B. Hunyuan-7B-Instruct [tencent/Hunyuan-7B-Instruct](https://huggingface.co/tencent/Hunyuan-7B-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 290943): Vocabulary size of the HunYuan model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`HunYuanVLTextConfig`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 11008): Dimension of the MLP representations or shared MLP representations. num_hidden_layers (`int`, *optional*, defaults to 32): 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*): 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://huggingface.co/papers/2305.13245). If it is not specified, will default to `num_attention_heads`. 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 2048): 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-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*, defaults to 0): Padding token id. bos_token_id (`int`, *optional*, defaults to 1): Beginning of stream token id. eos_token_id (`int`, *optional*, defaults to 2): End of stream token id. eod_token_id (int, *optional*, defaults to 3): Token ID representing the end-of-document marker. Used to indicate the termination of a text sequence. Example: In multi-document processing, this token helps the model distinguish between separate documents. pretraining_tp (`int`, *optional*, defaults to 1): Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is necessary to ensure exact reproducibility of the pretraining results. Please refer to [this issue](https://github.com/pytorch/pytorch/issues/76232). tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether to tie weight embeddings rope_theta (`float`, *optional*, defaults to 10000.0): The base period of the RoPE embeddings. rope_scaling (`Dict`, *optional*): Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update `max_position_embeddings` to the expected new maximum. See the following thread for more information on how these scaling strategies behave: https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an experimental feature, subject to breaking API changes in future versions. attention_bias (`bool`, defaults to `False`, *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 128): The attention head dimension. hunyuan_vl_textpast_key_valuespr+ Nsilur{Gz?r TrrrrF@rc :||_||_||_||_||_||_||_||}||_||_| |_ | |_ ||_ | |_ ||_ ||_||_||_t#jd| | ||d|dS)N) pad_token_id bos_token_id eos_token_idtie_word_embeddingsr) vocab_sizemax_position_embeddingsrrr%r"head_dimr#rinitializer_ranger+pretraining_tp use_cache rope_theta rope_scalingattention_biasattention_dropoutrr)r5rQrrr%r"r#rrRrTr+rVrMrNrO eod_token_idrUrPrWrXrYrZrSr6r7s r8rzHunYuanVLTextConfig.__init__s4%'>$&!2!2#6    &"5 #6 $!2(,"$(,!2 %%% 3         r9cb|jdSt|jtrt|jdkrt d|j|jdd}|jdd}|jdd}||dvrt d|||t d |-t|t r|d krt d ||-t|t r|d krt d |dSdS) z< Validate the `rope_scaling` configuration. Nrzi`rope_scaling` must be a dictionary with with two fields, `type` and `factor` or `type` and `alpha`, got typefactoralpha)lineardynamiczF`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got zI`rope_scaling`'s factor or alpha field must be have one, got both of noneg?z9`rope_scaling`'s factor field must be a float > 1.0, got z8`rope_scaling`'s alpha field must be a float > 1.0, got )rX isinstancedictlen ValueErrorgetfloat)r5rope_scaling_typerope_scaling_factorrope_scaling_alphas r8_rope_scaling_validationz,HunYuanVLTextConfig._rope_scaling_validations   $ F$+T22 c$:K6L6LPQ6Q6QK7;7HKK !-11&$??"/33HdCC!.227DAA  $(9AV(V(V+(++   &+=+E#   *.66 +:MQT:T:T-*--   )-u55 *9Ks9R9R,),,  * )9R9Rr9)rErrFrGrGNrHrrIr TrrrrrFrJNFrKN) r:r;r<__doc__r=keys_to_ignore_at_inferenceintrrkr?r@s@r8rBrBIsIIV#J#4"5!&  $!/7 7  7 7 7 7 7 7 r&&&&&&&r9rBcXeZdZdZeedZdgZ dfd Zfd Z fd Z xZ S)HunYuanVLConfigr)r text_configrDN6789:;c  tjdi| t|tr|jddi||_n||jd|_t|tr|jddi||_n||jddi| |_||_||_||_ ||_ ||_ ||_ |jj |j_| dd|_dS)Nrrqattn_implementationr)rrrbrc sub_configsrrqimage_token_id im_start_id im_end_id im_newline_idvideo_start_id video_end_idrr4pop_attn_implementation) r5rqrr|r}r{r~rrr6r7s r8rzHunYuanVLConfig.__init__s1 ""6""" mT * * E!B!1/!B!S!S]!S!SD    "!B!1/!B!D!DD  k4 ( ( I>t/ >MMMMD    >t/ >HHHHD ,&"*,(.2.>.J+%+JJ/Dd$K$K!!!r9ctddx} |dvr||jvrt |||dSt||dS)N__dict__rq)dtype_attn_implementation_internal)r__getattribute__rfrsetattr __setattr__)r5keyvaluerqr7s r8rzHunYuanVLConfig.__setattr__,s!GG44Z@@DD]SS S[ EEE{+++ Ke , , , , , GG  U + + + + +r9cdtdvr>|dvr:td}||jvrt||St|S)Nrqr) _name_or_pathr=rr)rrrgetattr)r5rrqr7s r8rz HunYuanVLConfig.__getattribute__7s EGG44Z@@ @ @SQ F F  ''22=AAKk***{C000ww'',,,r9)NNrrrsrtrurvrw) r:r;r<r=rrBrzrmrrrr?r@s@r8rprpsJ.*K$5"5(L(L(L(L(L(LT , , , , , - - - - - - - - -r9rpN) transformersrrrBrprr9r8rs *)))))=1=1=1=1=1,=1=1=1@nnnnn*nnnbH-H-H-H-H-&H-H-H-H-H-r9