iX&ddlmZddlmZddlmZddlmZddlm Z ddl m Z ddl m Z dd lmZmZdd lmZmZdd lmZmZdd lmZdd lmZmZmZddlmZmZm Z ddl!m"Z"m#Z#ddl$m%Z%m&Z&er ddl'Z'ddl'm(Z(Gdde(jRZ*dZ+de'jXde-de'jXfdZ. d7de(jRde'jXde'jXde'jXd e'jXdzd!e/d"e/d#eefd$Z0d8d%Z1e e1Gd&d'e(jRZ2Gd(d)e(jRZ3Gd*d+e Z4eGd,d-eZ5Gd.d/e5Z6Gd0d1e(jRZ7ed23Gd4d5e5e Z8gd6Z9y)9)Callable)Optional)ACT2FN)Cache)GenerationMixin)use_kernelized_func)GradientCheckpointingLayer)BaseModelOutputWithPoolingCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringis_torch_available)can_return_tuplecheck_model_inputsmaybe_autocast) AutoModelAutoModelForCausalLM) GlmAsrConfigGlmAsrEncoderConfigN)nnc eZdZUejed<d deffd Ze d dedzde dde dzde d e ffd Z ejed ZxZS)GlmAsrRotaryEmbeddinginv_freqNconfigct||j|_|j|_||_|j j d|_|j}|jdk7rt|j}||j |\}|_ |jd|d|jd|jdy)N rope_typedefaultr!F) persistentoriginal_inv_freq) super__init__max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr"rope_parametersr$compute_default_rope_parametersr attention_scalingregister_bufferclone)selfr"device rope_init_fnr! __class__s t/home/obispo/Crisostomo_bridge/mision_env/lib/python3.12/site-packages/transformers/models/glmasr/modeling_glmasr.pyr)zGlmAsrRotaryEmbedding.__init__/s "("@"@$*$B$B! 44[A!%!E!E >>Y &.t~~>L+7 V+L($( ZeD 0(..2BuUr3z torch.deviceseq_lenreturnz torch.Tensorcn|jd}|jjdd}t|ddxs|j|jz}t ||z}d}d|t jd|dt jj|t j |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_thetapartial_rotary_factorg?head_dimNrrdtype)r3r?) r-getgetattr hidden_sizenum_attention_headsinttorcharangeint64tofloat) r"r3r8baser<r=dimattention_factorr!s r6r.z5GlmAsrRotaryEmbedding.compute_default_rope_parameters?s&%%l3 & 6 6 : :;RTW X6:t4h8J8JfNhNh8h(223 U\\!S!5;;?BB&X]XcXcBdgjj k )))r7cN|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) NrrmpscpuF) device_typeenabledrrKr>)r!rIexpandshaperHr3 isinstancetypestrr transposerEcatcosr/sinr?) r2x position_idsinv_freq_expandedposition_ids_expandedrQfreqsembr[r\s r6forwardzGlmAsrRotaryEmbedding.forward_sR!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$N)NNN)__name__ __module__ __qualname__rETensor__annotations__rr) staticmethodrrDtuplerIr.no_gradrrc __classcell__r5s@r6r r ,sllV|V &*+/"*t#*(*t* ~u$ % **>U]]_ <