qi<|ddlmZddlZddlmZddlmZddlmZm Z ddl m Z m Z ddl mZmZddlmZdd lmZdd lmZdd lmZd d lmZd dlmZmZmZmZmZm Z m!Z!m"Z"Gdde Z#Gdde Z$GddeZ%GddeZ&GddeZ'GddeZ(GddeZ)GddeZ*gdZ+y) )CallableN)TransformersKwargs)Cache DynamicCache)PreTrainedConfiglayer_type_validation)create_causal_mask!create_sliding_window_causal_mask)BaseModelOutputWithPast)RopeParameters)ALL_ATTENTION_FUNCTIONS)Unpack)Gemma2RotaryEmbedding)Olmo2AttentionOlmo2DecoderLayerOlmo2ForCausalLM Olmo2ModelOlmo2PreTrainedModel Olmo2RMSNormapply_rotary_pos_embeager_attention_forwardc*peZdZdZdZdgZddddddddZd gd gfd d gd gfd gd gfd Z d$dedzdedzdedzdedzdedzdedzde dzdedzde dzde dzdedzdedzdedzde dzde e e e fzdzde dzde dzd e dzd!edzd"ee dzf(fd# ZxZS)% Olmo3Configa This is the configuration class to store the configuration of a [`Olmo3Model`]. It is used to instantiate an OLMo3 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 [allenai/OLMo-3-0725-1B](https://huggingface.co/allenai/OLMo-3-0725-1B). 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 50304): Vocabulary size of the Olmo3 model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`Olmo3Model`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 11008): Dimension of the 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, check out [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. 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 1): Padding token id. bos_token_id (`int`, *optional*): Beginning of stream token id. eos_token_id (`int`, *optional*, defaults to 50279): End of stream token id. 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`. 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. rms_norm_eps (`float`, *optional*, defaults to 1e-05): The epsilon used by the rms normalization layers. sliding_window (`int`, *optional*, defaults to 4096): Size of the sliding window for sliding window attention. layer_types (`list`, *optional*): Attention pattern for each layer. Defaults to sliding window attention for 3 out of 4 layers, and full attention for every 4th layer. ```python >>> from transformers import Olmo3Model, Olmo3Config >>> # Initializing a Olmo3 7B style configuration >>> configuration = Olmo3Config() >>> # Initializing a model from the Olmo3 7B style configuration >>> model = Olmo3Model(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` olmo3past_key_valuescolwise_gather_outputrowwise_split_inputcolwiserowwise)zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projzlayers.*.mlp.gate_projzlayers.*.mlp.up_projzlayers.*.mlp.down_proj input_ids inputs_embeds hidden_statesattention_mask) embed_tokenslayersnormN vocab_size hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headsnum_key_value_heads hidden_actmax_position_embeddingsinitializer_range use_cache pad_token_id bos_token_id eos_token_idtie_word_embeddingsrope_parametersattention_biasattention_dropout rms_norm_epssliding_window layer_typesc ||_||_||_||_||_||_||}||_||_| |_| |_ ||_ ||_ ||_ | |_ | |_| |_||_||_||_|j$5t'|jDcgc]}|dzdzdk7rdndc}|_t)|j$|j||_t-|\di|ycc}w)Nrsliding_attentionfull_attention)r)r0r*r+r,r-r.r/r1r2r8r9r6r3r4r5r:r;r<ranger r7super__init__)selfr)r*r+r,r-r.r/r0r1r2r3r4r5r6r7r8r9r:r;r<kwargsi __class__s Y/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/olmo3/modular_olmo3.pyrEzOlmo3Config.__init__s#0%'>$&!2!2#6   &"5 #6 $!2",!2#6 ((((,&    #W\]a]s]sWt RSA{a'7#=MM D  d..0F0FG. "6" s.D)ii+ rLNsiluig{Gz?Tr>NigFNFgh㈵>rKN)__name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inferencebase_model_tp_planbase_model_pp_planintstrfloatboolr dictlistrE __classcell__rIs@rJrr*sJXJ#4"5%<%<%<%:"+ )"+&(9:#%568IJ!"_$56"'"&(-(**,*.!'.2*.!%#$#'#(+0MQ&+*-%)%)(,+9#$J9#4Z9#: 9# : 9# !4Z 9#!4Z9#$J9#"%t9#!4<9#$;9#Dj9#Dj9#Dj9#"D[9# ($sN/B*CCdJ!9#"t #9#$!4<%9#&dl'9#(d )9#*#Y%+9#9#rc eZdZy) Olmo3RMSNormNrOrPrQrBr_rJrarar_raceZdZdedeffd Z d dejdeejejfdejdzde dzd ejdzd e e d eejejdzffd Z xZS)Olmo3Attentionconfig layer_idxct||||j||_|jdk(r|j|_yd|_y)N)rgr@)rDrEr<attention_typer;rFrfrgrIs rJrEzOlmo3Attention.__init__sL 95$00;7;7J7JNa7af33gkr_Nr$position_embeddingsr%rcache_positionrGreturnc v|jdd}g|d|j}|j|j|} |j |j |} |j |} | j|jdd} | j|jdd} | j|jdd} |\} } t| | | | \} } |'| | |d}|j| | |j|\} } tj|jjt }||| | | |f|j"sdn |j$|j&|j(d|\}}|j*g|dj-}|j/|}||fS)Nr>r)sincosrlrN)dropoutscalingr;)shapehead_dimq_normq_projk_normk_projv_projview transposerupdatergr get_interfacerf_attn_implementationrtrainingr9rsr;reshape contiguouso_proj)rFr$rkr%rrlrG input_shape hidden_shape query_states key_states value_statesrqrp cache_kwargsattention_interface attn_output attn_weightss rJforwardzOlmo3Attention.forwards$))#2. 88b8$--8 {{4;;}#=> [[]!;< {{=1 #((6@@AF __\2<.) t+ .)  .) ((4/ .)+,.) u||U\\D00 1.)r_rec eZdZy)Olmo3DecoderLayerNrbrBr_rJrrrcr_rc eZdZy)Olmo3RotaryEmbeddingNrbrBr_rJrrrcr_rc eZdZy)Olmo3PreTrainedModelNrbrBr_rJrr rcr_rceZdZdeffd Z ddej dzdejdzdej dzdedzdejdzd ej dzd e dzd e e d e fd ZxZS) Olmo3Modelrfc &t||t|j|j|_t jt|jDcgc]}t||c}|_ t||_ ycc}w)N)epsrf)rDrErar*r:r(nn ModuleListrCr,rr'r rotary_embrjs rJrEzOlmo3Model.__init__so   !3!39L9LM mmCHIaIaCb ci vy 1 c  /f= dsBNr"r% position_idsrr#rlr2rGrmc |du|duz r td||j|}|r|t|j}|E||j nd} t j |jd|j| z}||jd}t|x} ts*|j|||||d} td i| td i| d} |} |j| |} |jd|jj D](}|| f| |j"j$|||| d|} *|j'| } t)| | S) Nz:You must specify exactly one of input_ids or inputs_embedsrrr>)device)rfr#r%rlrr)rAr@)r%rrrlrk)last_hidden_staterrB) ValueErrorr&rrfget_seq_lengthrarangertr unsqueeze isinstancer[r r rr'r, self_attnrir(r )rFr"r%rrr#rlr2rGpast_seen_tokenscausal_mask_mapping mask_kwargsr$rk decoder_layers rJrzOlmo3Model.forwards  -t"; <YZ Z  *.*;*;I*FM 0*$++>O  !CRC^==?de  ]003M{>.2.204(,2626!%@ ##d*@  t+@ &&- @  @ ((4/ @ ((4/@ $;@ +,@  !@ r_rc eZdZy)Olmo3ForCausalLMNrbrBr_rJrr\rcr_r)rrrr),collections.abcrrtorch.nnrtransformers.utils.genericr cache_utilsrrconfiguration_utilsrr masking_utilsr r modeling_outputsr modeling_rope_utilsr modeling_utilsrprocessing_utilsrgemma2.modeling_gemma2rolmo2.modeling_olmo2rrrrrrrrrrarerrrrr__all__rBr_rJrs% 9.JR715&:   W#"W#t < 4)^4)n )  0  / I I X '  r_