qi^ddlmZddlmZddlZddlmZddlmZddlm Z m Z ddl m Z dd l mZmZmZdd lmZmZdd lmZdd lmZmZmZmZdd lmZmZddlmZm Z ddl!m"Z"m#Z#ddl$m%Z%ddl&m'Z'm(Z(m)Z)ddl*m+Z+m,Z,ddl-m.Z.ddl/m0Z0edGddejbZ2GddejbZ3GddejbZ4dZ5edd>dZ6d ejnd!e8d"ejnfd#Z9 d?d$ejbd%ejnd&ejnd'ejnd(ejndzd)e:d*e:d+e%e'fd,Z;ee6Gd-d.ejbZ<Gd/d0eZ=e(Gd1d2e#Z>e(Gd3d4e>Z?e(Gd5d6e>e Z@Gd7d8ee>ZAGd9d:ee>ZBGd;dZCgd=ZDy)@)Callable)OptionalN)nn)ACT2FN)Cache DynamicCache)GenerationMixin)use_kernel_forward_from_hubuse_kernel_func_from_hubuse_kernelized_func)create_causal_mask!create_sliding_window_causal_mask)FlashAttentionKwargs)GenericForQuestionAnswering GenericForSequenceClassificationGenericForTokenClassificationGradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringcan_return_tuple)maybe_autocastmerge_with_config_defaults)capture_outputs) Qwen3ConfigRMSNormcheZdZddeddffd Zdej dej fdZdZxZ S) Qwen3RMSNormepsreturnNct|tjt j ||_||_y)z; Qwen3RMSNorm is equivalent to T5LayerNorm N)super__init__r Parametertorchonesweightvariance_epsilon)self hidden_sizer' __class__s Z/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/qwen3/modeling_qwen3.pyr+zQwen3RMSNorm.__init__3s1 ll5::k#:; # hidden_statesc"|j}|jtj}|j dj dd}|tj ||jzz}|j|j|zS)NT)keepdim) dtypetor-float32powmeanrsqrtr0r/)r1r6 input_dtypevariances r4forwardzQwen3RMSNorm.forward;sy#)) %((7  $$Q',,R,>% Ht?T?T4T(UU {{]--k:::r5c^t|jjd|jS)Nz, eps=)tupler/shaper0)r1s r4 extra_reprzQwen3RMSNorm.extra_reprBs* ))*+6$2G2G1HIIr5)gư>) __name__ __module__ __qualname__floatr+r-TensorrCrG __classcell__r3s@r4r&r&1s7$$$$;U\\;ell;Jr5r&c$eZdZfdZdZxZS)Qwen3MLPct|||_|j|_|j|_t j |j|jd|_t j |j|jd|_t j |j|jd|_ t|j|_ yNFbias) r*r+configr2intermediate_sizerLinear gate_projup_proj down_projr hidden_actact_fnr1rUr3s r4r+zQwen3MLP.__init__Gs  !--!'!9!94#3#3T5K5KRWXyy!1!143I3IPUV 4#9#94;K;KRWXV../ r5c|j|j|j||j|z}|SN)rZr\rXrY)r1xrZs r4rCzQwen3MLP.forwardQs6NN4;;t~~a/@#ADLLQRO#ST r5)rHrIrJr+rCrMrNs@r4rPrPFs 0r5rPc eZdZUejed<d deffd Ze d dedzde dde dzde d e ffd Z ejed ZxZS)Qwen3RotaryEmbeddinginv_freqNrUct||j|_|j|_||_|j j d|_|j}|jdk7rt|j}||j |\}|_ |jd|d|jd|jdy)N rope_typedefaultrcF) persistentoriginal_inv_freq) r*r+max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenrUrope_parametersrecompute_default_rope_parametersrattention_scalingregister_bufferclone)r1rUdevice rope_init_fnrcr3s r4r+zQwen3RotaryEmbedding.__init__Ys "("@"@$*$B$B! 44[A!%!E!E >>Y &.t~~>L+7 V+L($( ZeD 0(..2BuUr5rqz torch.deviceseq_lenr(z torch.Tensorc|jd}t|ddxs|j|jz}d}d|t j d|dtj j|tj|z zz }||fS) a Computes the inverse frequencies according to the original RoPE implementation Args: config ([`~transformers.PreTrainedConfig`]): The model configuration. device (`torch.device`): The device to use for initialization of the inverse frequencies. seq_len (`int`, *optional*): The current sequence length. Unused for this type of RoPE. Returns: Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE). rope_thetahead_dimNg?rr8r;)rqr;) rlgetattrr2num_attention_headsr-arangeint64r<rK)rUrqrsbasedimattention_factorrcs r4rmz4Qwen3RotaryEmbedding.compute_default_rope_parametersis&%%l3fj$/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) Nrr9r"mpscpuF) device_typeenabledr8r}rw)rcrKexpandrFr<rq isinstancetypestrr transposer-catcosrnsinr;) r1r` position_idsinv_freq_expandedposition_ids_expandedrfreqsembrrs r4rCzQwen3RotaryEmbedding.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$r_)NNN)rHrIrJr-rL__annotations__r#r+ staticmethodrintrErKrmno_gradrrCrMrNs@r4rbrbVsllV{V %)+/"*d"*(*t* ~u$ % **:U]]_ <