ZjTbSSKrSSKJr SSKJr SSKrSSKJr SSKJ r SSK J r SSK JrJr SSKJr SS KJrJr SS KJr SS KJr SS KJrJr SS KJrJr SSKJ r J!r! SSK"J#r# SSK$J%r%J&r& SSK'J(r(J)r)J*r* SSK+J,r, SSK-J.r. "SS\RR^5r0"SS\R^5r1\"S5S6Sj5r2S\RfS\4S\Rf4Sjr5S7S\R^S \RfS!\RfS"\RfS#\RfS-S$\6S%\6S&\#\%4S'jjr7S(r8\"\25"S)S*\R^55r9"S+S,\R^5r:"S-S.\5r;\&"S/S0\!55r<\&"S1S2\<55r=\&"S3S4\<\55r>/S5Qr?g)8N)Callable)Optional)initialization)ACT2FN)Cache DynamicCache)GenerationMixin)use_kernel_func_from_hubuse_kernelized_func)create_causal_mask)GradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstring)can_return_tuplemaybe_autocastmerge_with_config_defaults)capture_outputs)NanoChatConfigcF^\rSrSrSS\4U4SjjjrSrSrSrSr U=r $) NanoChatRMSNorm-epsc.>[TU]5 XlgN)super__init__r!)selfr! __class__s /root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/nanochat/modeling_nanochat.pyr%NanoChatRMSNorm.__init__.s cU[R"URS5RSSS9UR-5-$)NT)keepdim)torchrsqrtpowmeanr!r&xs r(_normNanoChatRMSNorm._norm2s45;;quuQx}}R}>IJJJr*c^URUR55RU5$r#)r5floattype_asr3s r(forwardNanoChatRMSNorm.forward5s"zz!'')$,,Q//r*c SUR3$)Nzeps=r!)r&s r( extra_reprNanoChatRMSNorm.extra_repr8sdhhZ  r*r=)gư>) __name__ __module__ __qualname____firstlineno__r8r%r5r:r>__static_attributes__ __classcell__r's@r(rr-s)EK0!!r*rc ^\rSrSr%\R \S'SS\4U4Sjjjr\ SS\S-S\ SS\ S-S \ S \ 44S jj5r\R"5\S 55rS rU=r$)NanoChatRotaryEmbedding<inv_freqNconfigc>[TU]5 URUlURUlXlUR R SUlURnURS:wa[URnU"UR U5uo@l URSUSS9 URSUR5SS9 g)N rope_typedefaultrJF) persistentoriginal_inv_freq) r$r%max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenrKrope_parametersrMcompute_default_rope_parametersrattention_scalingregister_bufferclone)r&rKdevice rope_init_fnrJr's r(r% NanoChatRotaryEmbedding.__init__?s "("@"@$*$B$B! 44[A!%!E!E >>Y &.t~~>L+7 V+L(( ZeD 0(..2BuUr*rYz torch.deviceseq_lenreturnz torch.Tensorc URSn[USS5=(d URUR-nSnSU[R "SUS[R S9RU[RS9U- -- nXe4$) aH Computes the inverse frequencies according to the original RoPE implementation Args: config ([`~transformers.PreTrainedConfig`]): The model configuration. device (`torch.device`): The device to use for initialization of the inverse frequencies. seq_len (`int`, *optional*): The current sequence length. Unused for this type of RoPE. Returns: Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE). rope_thetahead_dimNg?rr,dtype)rYrb) rTgetattr hidden_sizenum_attention_headsr/arangeint64tor8)rKrYr\basedimattention_factorrJs r(rU7NanoChatRotaryEmbedding.compute_default_rope_parametersOs&%%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjj k ))r*cLURSSS2S4R5RURSSS5R UR 5nUSS2SSS24R5n[ UR R[5(a0UR RS:waUR ROSn[USS9 UR5UR5-RSS5n[R"Xf4SS 9nUR5UR-nUR5UR-n SSS5 WR UR S 9W R UR S 94$!,(df  N@=f) Nrr-rmpscpuF) device_typeenabledr,rjra)rJr8expandshaperhrY isinstancetypestrr transposer/catcosrVsinrb) r&r4 position_idsinv_freq_expandedposition_ids_expandedrpfreqsembrzr{s r(r:NanoChatRotaryEmbedding.forwardmsN!MM$4-8>>@GG HZHZ[\H]_acdehhijiqiqr ,QaZ 8 > > @'1!((--'E'E!((--[`J`ahhmmfk  U C&,,.1F1L1L1NNYYZ[]^_E))UN3C'')d444C'')d444C D vvAGGv$cff177f&;;; D Cs BF F#)rVrKrRrSrMr#NNN)r@rArBrCr/Tensor__annotations__rr% staticmethodrinttupler8rUno_gradrr:rDrErFs@r(rHrH<sllV~VV (,+/"*%*(*t* ~u$ % **: ]]_ < C/0Gw;q>C/0G  r* hidden_statesn_repr]cURup#pEUS:XaU$USS2SS2SSS2SS24RX#XU5nURX#U-XE5$)z This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim) rN)rtrsreshape)rrbatchnum_key_value_headsslenr`s r( repeat_kvrs_ 2?1D1D.E z!!Qa"23::5W\dlmM  e(CT TTr*modulequerykeyvalueattention_maskscalingdropoutkwargsc [X R5n[X0R5n [R"XR SS55U-n UbX-n [ R RU S[RS9RUR5n [ R RXURS9n [R"X5n U R SS5R5n X4$)Nr,rr-)rjrb)ptrainingr)rnum_key_value_groupsr/matmulrxnn functionalsoftmaxfloat32rhrbrr contiguous) rrrrrrrr key_states value_states attn_weights attn_outputs r(eager_attention_forwardrs3 ; ;