iWddlmZddlmZddlZddlmZddlmZddlm Z ddl m Z m Z dd l mZdd lmZmZmZdd lmZdd lmZmZmZmZdd lmZmZddlmZmZddl m!Z!m"Z"ddl#m$Z$ddl%m&Z&m'Z'ddl(m)Z)m*Z*ddl+m,Z,GddejZZ.edGddejZZ/GddejZZ0dZ1edd?dZ2dejfd e4d!ejffd"Z5 d@d#ejZd$ejfd%ejfd&ejfd'ejfdzd(e6d)e6d*e$e&fd+Z7ee2Gd,d-ejZZ8Gd.d/eZ9eGd0d1e"Z:eGd2d3e:Z;ed45Gd6d7e:eZ<ed45Gd8d9ee:Z=ed45Gd:d;ee:Z>ed45Gd<d=ee:Z?gd>Z@y)A)Callable)OptionalN)nn)auto_docstring)ACT2FN)Cache DynamicCache)GenerationMixin)use_kernel_forward_from_hubuse_kernel_func_from_hubuse_kernelized_func)create_causal_mask)GenericForQuestionAnswering GenericForSequenceClassificationGenericForTokenClassificationGradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargscan_return_tuple)check_model_inputsmaybe_autocast) ArceeConfigc$eZdZfdZdZxZS)ArceeMLPct|||_|j|_|j|_t j |j|j|j|_t j |j|j|j|_ t|j|_ y)Nbias) super__init__config hidden_sizeintermediate_sizerLinearmlp_biasup_proj down_projr hidden_actact_fnselfr( __class__s r/home/obispo/Crisostomo_bridge/mision_env/lib/python3.12/site-packages/transformers/models/arcee/modeling_arcee.pyr'zArceeMLP.__init__2s  !--!'!9!9yy!1!143I3IPVP_P_` 4#9#94;K;KRXRaRabV../ c`|j|j|j|SN)r.r0r-)r2xs r4forwardzArceeMLP.forward;s"~~dkk$,,q/:;;r5)__name__ __module__ __qualname__r'r9 __classcell__r3s@r4r"r"1s 0% Ht?T?T4T(UU {{]--k:::r5c^t|jjd|jS)Nz, eps=)tuplerFshaperG)r2s r4 extra_reprzArceeRMSNorm.extra_reprPs* ))*+6$2G2G1HIIr5)gư>)r:r;r<r'r9rYr=r>s@r4rArA?s$;Jr5rAc eZdZUejed<d deffd Ze d dedzde dde dzde d e ffd Z ejed ZxZS)ArceeRotaryEmbeddinginv_freqNr(ct||j|_|j|_||_|j j d|_|j}|jdk7rt|j}||j |\}|_ |jd|d|jd|jdy)N rope_typedefaultr\F) persistentoriginal_inv_freq) r&r'max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr(rope_parametersr^compute_default_rope_parametersrattention_scalingregister_bufferclone)r2r(device rope_init_fnr\r3s r4r'zArceeRotaryEmbedding.__init__Ws "("@"@$*$B$B! 44[A!%!E!E >>Y &.t~~>L+7 V+L($( ZeD 0(..2BuUr5rjz torch.deviceseq_lenreturnz torch.Tensorc|jd}t|ddxs|j|jz}d}d|t j d|dtj j|tj|z zz }||fS) a Computes the inverse frequencies according to the original RoPE implementation Args: config ([`~transformers.PreTrainedConfig`]): The model configuration. device (`torch.device`): The device to use for initialization of the inverse frequencies. seq_len (`int`, *optional*): The current sequence length. Unused for this type of RoPE. Returns: Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE). rope_thetahead_dimNg?rrJrM)rjrM) regetattrr)num_attention_headsrDarangeint64rNfloat)r(rjrlbasedimattention_factorr\s r4rfz4ArceeRotaryEmbedding.compute_default_rope_parametersgs&%%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjj k )))r5cN|jddddfjj|jdddj |j }|dddddfj}t |j jtr/|j jdk7r|j jnd}t|d5|j|jzjdd}tj||fd }|j|jz}|j|jz} dddj |j  j |j fS#1swYAxYw) NrrKrmpscpuF) device_typeenabledrJrxrq)r\rvexpandrXrNrj isinstancetypestrr transposerDcatcosrgsinrM) r2r8 position_idsinv_freq_expandedposition_ids_expandedr}freqsembrrs r4r9zArceeRotaryEmbedding.forwardsR!MM$4-8>>@GG HZHZ[\H]_acdehhijiqiqr ,QaZ 8 > > @'1!((--'E!((--[`J`ahhmmfk  U C 5&,,.1F1L1L1NNYYZ[]^_E))UEN3C'')d444C'')d444C  5 vvAGGv$cff177f&;;;  5 5s BFF$r7)NNN)r:r;r<rDTensor__annotations__r r' staticmethodrintrWrvrfno_gradrr9r=r>s@r4r[r[TsllV{V %)+/"*d"*(*t* ~u$ % **:U]]_ <