ipex-llm 2.2.0b20250107__py3-none-win_amd64.whl → 2.2.0b20250109__py3-none-win_amd64.whl

{ipex_llm-2.2.0b20250107.dist-info → ipex_llm-2.2.0b20250109.dist-info}/RECORD

@@ -41,35 +41,35 @@ ipex_llm/langchain/llms/transformerspipelinellm.py,sha256=vm522YPPwWxxAPVvQBtxRf
 ipex_llm/langchain/vllm/__init__.py,sha256=T-EbRT6GJ_8RCu-iLmSzcftOimXSPQf2d5X72AUAy2Y,874
 ipex_llm/langchain/vllm/vllm.py,sha256=6dxc-ZISZQrJilEa_HA827l75Dv9rcHpY_G6FdJ8BVs,7793
 ipex_llm/libs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ipex_llm/libs/bloom-api.dll,sha256=aa3qzp0VqG1RzqCWC8uPfyzCsf3xC0qqgyj4WxDoOsM,36352
-ipex_llm/libs/bloom.dll,sha256=kVhdgqqPvusN-aFfVCBnSZSGUWRw9dv9ICv-klmF0UE,506880
-ipex_llm/libs/gptneox-api.dll,sha256=bjp6ZiVvmT6aWrDiqgV43pYeifLnGDlNnRwi3r1rTFo,24576
-ipex_llm/libs/gptneox.dll,sha256=xKfJ9oNMoUXzHH43vXrap9wbm-KgKY-AhARrVvU_iLg,567296
-ipex_llm/libs/libbloom_avx.dll,sha256=VGPb6z7Jh4pgStkfnzqEBW-mnzLTETVUsc20ym5c1eI,535040
-ipex_llm/libs/libbloom_vnni.dll,sha256=_ibtsNyO3G9KV30mKJky49eKMm_8ZVgKPA0FdjjIvB0,506880
-ipex_llm/libs/libgptneox_avx.dll,sha256=onI5sXdEsCAPL3JAIGH5r4jwPbevYi6bw15CmPyZL5s,595456
-ipex_llm/libs/libgptneox_vnni.dll,sha256=6BvdfvGnikm78KlPcdGq6IiAecPkPwm6i38PotgB9FQ,567808
-ipex_llm/libs/libllama_avx.dll,sha256=TRqBjWOmR7dCQtEq9c5l0wshBPM2e7klfca8eV-f79Q,589824
-ipex_llm/libs/libllama_vnni.dll,sha256=F9va1YfNVe9XGfilbio3mEdbbGtmpfAR-9T8vWEqulI,561664
-ipex_llm/libs/libstarcoder_avx.dll,sha256=IvM2V18nf_g9tsaLz8W30qgchg496V89PpCoKMHnQPY,626688
-ipex_llm/libs/libstarcoder_vnni.dll,sha256=X9H9Tyy3DdQg1LIvc8ILP33ilj4amEUfwF_E9KPVWvE,598528
-ipex_llm/libs/llama-api.dll,sha256=-XYElLfNrWEP63PStFxXX9wM0kP6PGYQJH6oePg5u-I,25600
-ipex_llm/libs/llama.dll,sha256=zLW5de-ASaVBoPQbRYUMK6L6FoPwmt6II4bGFVeHE-M,561152
-ipex_llm/libs/main-bloom.exe,sha256=cNaMadt2EpUgCu55XjQW0Vi1UUBQ5eA9CnfXZV6OGjc,103424
-ipex_llm/libs/main-gptneox.exe,sha256=54va_bkWTOw0Gf_EMEF2MaqNsKEugEsxTxUszTAxY7Q,98816
-ipex_llm/libs/main-llama.exe,sha256=banfh_rJMhBR4FERotQ6FBOKDEhz-KF49Zu6E86Wlso,99840
-ipex_llm/libs/main-starcoder.exe,sha256=L-SIpoGn0vA8CD37MMm44i56CTWKXQjnC6WWZE1okPw,157696
-ipex_llm/libs/pipeline.dll,sha256=rOzOT0lY3RMQlIDHCvaAx_bjAvQD8BBZNyqi3BssG2U,72704
-ipex_llm/libs/quantize-bloom.exe,sha256=GvY5jFYGrK_wFa1RyAurX0vD_UAg7FCD9Ls1zAbYYKc,126464
-ipex_llm/libs/quantize-bloom_vnni.exe,sha256=j3i2Q_ymqlnrPBUCskU3gqrwV198NkfLxmlz0KOsqoI,127488
-ipex_llm/libs/quantize-gptneox.exe,sha256=he-NK5wy8u_t3S26Qb4Euenf0E6mw8O6lsNOcX-m2ts,104448
-ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=Oz7iHHL01QZYJpCBsPeYXhBCRBJ9kFklYZ2UhEw77rk,104960
-ipex_llm/libs/quantize-llama.exe,sha256=Ns-98bA07AeTPaN7v0zs3cOb3PAvUmj7R_Xvsn-1bTM,109568
-ipex_llm/libs/quantize-llama_vnni.exe,sha256=t4wts7FmUT4n5_Ii6wqtrVMv73pf3pOMMfrziRZAI5U,110592
-ipex_llm/libs/quantize-starcoder.exe,sha256=xYLEIQ4gtU23ae7cN_1hZrxqDrSiQdCNu3EZRbffNLs,127488
-ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=5gmwDZTTqM9fFuj9UihVyA91TmcoBq95xxAofaBi4sM,128512
-ipex_llm/libs/starcoder-api.dll,sha256=UhlNrAN44aZwkbjdmVFr9E2tV3iuQ8jR1OvEgiQzVXk,21504
-ipex_llm/libs/starcoder.dll,sha256=1jGXZ-nQo3m1kMXeOVwd-fc68p7kjPGfzAu4ifngrlE,598016
+ipex_llm/libs/bloom-api.dll,sha256=vLC0Hy_vsNY1cTKoFMJ7g9iPzKV5X_YL8U9N3k8FHFk,36352
+ipex_llm/libs/bloom.dll,sha256=hVQn4GiRqyD8P-_XbuQD_inkB74jxR8MutecrzrIQWQ,506880
+ipex_llm/libs/gptneox-api.dll,sha256=R7YnJz6n4TCdPq5NYWuSaanj-hGrYnzZfJ9P4z033Ws,24576
+ipex_llm/libs/gptneox.dll,sha256=Iyi18Z9buCMvglEGOPHYlpL4kSa4CADaXUMXJF4Bhqs,567296
+ipex_llm/libs/libbloom_avx.dll,sha256=8NZmJGuvsGISZDOO1wp6jtfh6xWGmLfFhS9RfGl8WEk,535040
+ipex_llm/libs/libbloom_vnni.dll,sha256=mTGMJzQiM7Mj0Qdrz7Xs7YWWypXijWum_LmSBR5Q2uE,506880
+ipex_llm/libs/libgptneox_avx.dll,sha256=617wPDx7BAimc3oOSF2X0jh9h6j4mdBFRBt-1Qkmz1A,595456
+ipex_llm/libs/libgptneox_vnni.dll,sha256=DmCL8HfUA8DbNQiZHDB77fpadSqnHSaxHpWRCw_FTDA,567808
+ipex_llm/libs/libllama_avx.dll,sha256=JHRfoFLp_B-3J5B1oa2FnKIIjKK6ZJaJrTfJXLgrGIA,589824
+ipex_llm/libs/libllama_vnni.dll,sha256=WxH8QfmfpuZ9rZJP2lcoXX5Zp9lonun-hhw26O40w78,561664
+ipex_llm/libs/libstarcoder_avx.dll,sha256=9Dqf2uWGA8XlGAjOqgMQiC-iNJYBhOhNiFk0GZTfwiQ,626688
+ipex_llm/libs/libstarcoder_vnni.dll,sha256=ob3olLuRXCGvpNuBYImrt92aejvhDsZpi-m_r75lIUY,598528
+ipex_llm/libs/llama-api.dll,sha256=1OMZVvZun0kQBd1EEQZUvSRJXHfaJgo9WYka4tAamGA,25600
+ipex_llm/libs/llama.dll,sha256=ZLuhZQAjPc76oJ-PO_yUlZi24qlzIBs8OXnR9-2LTGU,561152
+ipex_llm/libs/main-bloom.exe,sha256=T17qQytgapUQAmXSkH38RO21Ynb9NrOSiD3erxr4kJI,103424
+ipex_llm/libs/main-gptneox.exe,sha256=qskuY8ijrPRcBzTtGb8gB1PAoSvE4wkJe2GgiAAWS-4,98816
+ipex_llm/libs/main-llama.exe,sha256=t-mQ7luxLky6boib8wSRecL81DAdKF3HsLZ-oN4Agm4,99840
+ipex_llm/libs/main-starcoder.exe,sha256=WQ2L7L2XdmHxzt6L3uaHuWwTJVL9xg683y4_7avVVgk,157696
+ipex_llm/libs/pipeline.dll,sha256=_Kx68vo1DLJc57qF8aGEhdiCFbcAsm8xYQWeJHogVuM,72704
+ipex_llm/libs/quantize-bloom.exe,sha256=CTJhn-jYm4V99Fcv2R2mHkcYMfzQOU8IiqMKfqRK59A,126464
+ipex_llm/libs/quantize-bloom_vnni.exe,sha256=aauMlSe4e05rKB61u4PQdckBNWQzrISe5_HXkRCTCfI,127488
+ipex_llm/libs/quantize-gptneox.exe,sha256=QP-fDnaF6FElKal_kRh_N5nB-WUQKah8vG33lJQ3g9c,104448
+ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=PUgZZTq1a36tbtsgu-jnK0MHlsFQ-zPtAsAQ7PLczJU,104960
+ipex_llm/libs/quantize-llama.exe,sha256=uagQGbAgP8ripvyS0CREOQH0JhwHRi5dWeo_Q3criNc,109568
+ipex_llm/libs/quantize-llama_vnni.exe,sha256=fB7-E0PGSliaZDfu9bCl1hdcoD7xIMoRHiE6fFTJ75g,110592
+ipex_llm/libs/quantize-starcoder.exe,sha256=6OMrGkHL87jw18EVzlEj9NJpjTl7b3_7YOF_YRC24sI,127488
+ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=i6SaJyFmS_S2jjl9Cpa_5S-RyHo1_ABD0gx2maxTdfc,128512
+ipex_llm/libs/starcoder-api.dll,sha256=CgnbCueyY_M3QN-lskMQBPPVrfuKgy2Qt0aVV6wXPgQ,21504
+ipex_llm/libs/starcoder.dll,sha256=bc_dnjXNNVaXQlVdGp-a6IMZO3J2Ux9hwEQH7xoEaek,598016
 ipex_llm/llamaindex/__init__.py,sha256=T-EbRT6GJ_8RCu-iLmSzcftOimXSPQf2d5X72AUAy2Y,874
 ipex_llm/llamaindex/llms/__init__.py,sha256=KP1lEdGqDuxPoxL1ZSH25Pm2kKMPJBWUTLR0ckSLMIU,1139
 ipex_llm/llamaindex/llms/bigdlllm.py,sha256=FQBzq1KOjfc6uofTXAha3O7TqpJkNfOFepXQmOVlbnI,26314
@@ -87,27 +87,27 @@ ipex_llm/serving/fastchat/tgi_api_protocol.py,sha256=brT3k3-V0NJrU4fRqUwWjC0O3iO
 ipex_llm/serving/fastchat/tgi_api_server.py,sha256=agNTAEiZPSuj3dEdIdYKwkoY0cXOUDX06DiM9VP2knQ,24418
 ipex_llm/serving/fastchat/vllm_worker.py,sha256=ZLz2Q9GxJO6r_LOiP6epgCRjBGk-K4EB1SNEWSJp5DA,11091
 ipex_llm/transformers/__init__.py,sha256=l4KkMkLe-pRC7b_kj6LCfeifgE-Uo33_Av_FwN9HnFA,1074
-ipex_llm/transformers/convert.py,sha256=pFm6VlU84u_Llr2sp6-gRrEYDeNgIk2QPukolq4IE1s,99947
+ipex_llm/transformers/convert.py,sha256=umI137wqV2d4itS0AJQoZcygeWBATpSJSDJ805cZ-SY,98499
 ipex_llm/transformers/convert_ipex.py,sha256=iKXo0n8fVFTOA2fNYYrByMFK0dovL-kLd2sVDk88AlQ,14334
 ipex_llm/transformers/embedding.py,sha256=bdgk59DvD4ZZyxRzewXOR7g56nThgO6uhIwk8QL7f-s,9299
 ipex_llm/transformers/kv.py,sha256=k4TU18LlA-Sbq9WNNQnfuzu3RSFBwFhmaV3BcGN5bAo,19191
 ipex_llm/transformers/lisa.py,sha256=F5WxbtXQ7RdKulj83h_2DnEIgKiKGZf7zvOmg6QBl2s,3289
-ipex_llm/transformers/loader.py,sha256=cOgX93xOC-4dt01GTJ5wyd7PjZ8S43r4mctkR2YxVuw,6893
+ipex_llm/transformers/loader.py,sha256=AwjV5RpI2t2bedlv7ZhLm8cfd-QJZm5hny-XyjIvdnk,6876
 ipex_llm/transformers/lookup.py,sha256=b6OlZ9OV10R9qeWw8mVryVpDxszkjwLkldvi7GPMJY8,19614
-ipex_llm/transformers/low_bit_linear.py,sha256=nKraUvZJ7UdXP29HSE4CJPIVxmN-TvG8dpT4gpleuyQ,41688
-ipex_llm/transformers/model.py,sha256=KcRjkauGg48BYrUBoUZaVMpg7Piuz5JrfIpVZd3EIjs,41105
+ipex_llm/transformers/low_bit_linear.py,sha256=Obdd08D9dvuroS_6XWo4DXO_DrNRsbAqjz-mQAHmfxY,40845
+ipex_llm/transformers/model.py,sha256=fj7LBjrWtWwDJJYXnWiXsLGS4ayqqHfnh0p51dSDssE,40908
 ipex_llm/transformers/modelling_bigdl.py,sha256=7JpNVMuyq_OmtNUaMFMXdxPWZp2q0QHC02QeA-VTPOw,6709
 ipex_llm/transformers/npu_model.py,sha256=YW02GeVz-9ZGqxAeSz0AOvciS-17bo9eK5ZOBrICwSQ,39508
 ipex_llm/transformers/patches.py,sha256=halPWm__ORh2fRFSIFPiCNg3LQBfrRkTPtmtRpBJCZQ,1286
 ipex_llm/transformers/pipeline_parallel.py,sha256=uNZpOXljNmdoEYnP8U-VFiN4dRZb2piQbIf2bG9LQnE,49051
 ipex_llm/transformers/qlora.py,sha256=jtPGsvWFjbTUGzDBCdfftnCis_0nJQNRpACSwXUbbGU,14943
 ipex_llm/transformers/relora.py,sha256=-dYzUV0P-IhO2jFdnzN9-v_sFzJpRj3ZwN9eCJzOoCw,16567
-ipex_llm/transformers/speculative.py,sha256=Zf1nQb5GXpJQrUHBTL-H4RUBfdv3lGhfehzudHimhYk,64109
+ipex_llm/transformers/speculative.py,sha256=0XNLgc9dGswJHVPrXo4iM7pPxkWwfFfJMECcivJSnIc,63368
 ipex_llm/transformers/streamer.py,sha256=RrVlLblzCOtABRUpaMXAyaMnCGgLUtAi_YesLumRbww,4842
 ipex_llm/transformers/training_patch.py,sha256=oxMkUtqyvqJiprw6dE3skkYfD1HOmUlH9N0hBkbn0G0,10799
-ipex_llm/transformers/utils.py,sha256=fXLIlr9hoBr27p3w3xzczZGPk2cCTIRbUKBkiVCGYbc,16889
+ipex_llm/transformers/utils.py,sha256=9IRSqfDokf8QFW9T47R--i3RL1E-_O31HO7IJf7H6pg,16748
 ipex_llm/transformers/xpu_customize_fwd.py,sha256=wFpIhs5F6tkNs8gBOrLxWdhLzO3EDHovVkERPIAoAvg,7611
-ipex_llm/transformers/xpu_ops.py,sha256=H46-69pMRQhekbAEoDfNacCInLWycMHDqrgMGLvFYfI,4362
+ipex_llm/transformers/xpu_ops.py,sha256=vw4cUwvqUqDr45d-WMIkCpM2oiHfjN-VjF0bjMSF4kY,4830
 ipex_llm/transformers/awq/__init__.py,sha256=Du5gu3-eeAkeDO_dEMBTzrDBA66DSN3uL3-rn8WGXQw,875
 ipex_llm/transformers/awq/act.py,sha256=YwomJzOOKwkKtzGrm4L4kwBstBLO1Z8SK4CKi8PSYVQ,2172
 ipex_llm/transformers/awq/awq.py,sha256=cGyRQJWwAEJtOtdSbsBoQ33KX_Ie0pv5OJHC0ACEELE,8861
@@ -137,46 +137,45 @@ ipex_llm/transformers/gguf/models/model_implement/yuan2/configuration_yuan.py,sh
 ipex_llm/transformers/gguf/models/model_implement/yuan2/yuan_hf_model.py,sha256=_AOGMV65XHxgTxIib7lgs49InopcecTzRwgtYR8NTUg,51084
 ipex_llm/transformers/models/__init__.py,sha256=tp2DcVkKg1-QvdYk7DY7rZvQWCDQ4ZjU8NAQ7Fclrpg,584
 ipex_llm/transformers/models/aquila.py,sha256=VZb5Drpo_fTxwcExZ397LygnsNPX2sVbie9_JeFudZI,5252
-ipex_llm/transformers/models/baichuan.py,sha256=oJCAEENSG8oQhJ-QPN2SiapARjAGdOM6nEbyCcYOMCo,19334
-ipex_llm/transformers/models/bert.py,sha256=bJNic2pt1kph0kBwdK5MRGyWupFfx2Ts0V3D1L-5kWo,6085
+ipex_llm/transformers/models/baichuan.py,sha256=cAQLmVG-3R8CSTGTcDy2JOOzVe-Ej8AXjIEIjvZBGlo,18376
+ipex_llm/transformers/models/bert.py,sha256=0Mm9jkvkzBxtc_z_GE1TcZoPz-HOg2Z2973ZEWgSwJk,5601
 ipex_llm/transformers/models/bloom.py,sha256=PxfzyYT-nFn3K5rZhTQjmcEjUUzAhUFzxIN4kzRlCuc,8103
 ipex_llm/transformers/models/chatglm.py,sha256=UHai1t2AUtGmF765_eHF8LUMVQzp_oCBx8TJB21WrHk,12597
-ipex_llm/transformers/models/chatglm2.py,sha256=SGCABJdYQLW0zDarEoWrEQLuWlbq9iQhYU8ZeR1-ptQ,15957
-ipex_llm/transformers/models/chatglm4.py,sha256=AAhAFFDDas5DBQPfh2Mwl7a2v7taKf6xphoeeNNFaBI,16593
-ipex_llm/transformers/models/chatglm4v.py,sha256=YRfuf9g1E0MQ_7wbHAOMvadFnO-j3LqI_k1SaRkDs0M,14055
-ipex_llm/transformers/models/common.py,sha256=4obQMGF02FCiXrHnFle9Fsx7C33b1FDt37qJJ4YgxRc,11578
+ipex_llm/transformers/models/chatglm2.py,sha256=KyAIX7zGVQDQuwwM3QMBNWZbTeMHEzKUIgAryT0voHc,14933
+ipex_llm/transformers/models/chatglm4.py,sha256=QvUehdaCePB3MNHyWg3dneDxmjtBdxYeKUyQUVcsgfM,16886
+ipex_llm/transformers/models/chatglm4v.py,sha256=L6y45M_wjS2_HqchmCUxRlQZUNuSNCGOiynAQrGh918,14124
+ipex_llm/transformers/models/common.py,sha256=Q3IEfGqvxoHyfIIF5s8qHmOJBBP3b2jyVAXk8C3b1Pg,11636
 ipex_llm/transformers/models/decilm.py,sha256=P-PBuDPf07GvKggLwJx_wPwIn6esN3rX8ai2JxRuZmE,5246
 ipex_llm/transformers/models/gemma.py,sha256=_E3Yw8Y45xyNVeLqyVKcpr8kjuICtETeL82cJ-bWJuU,9424
 ipex_llm/transformers/models/gemma2.py,sha256=2WZuv-FLzJyTJFaYxOuzJt47QE64M0lHnzAiO5T6ozI,8049
-ipex_llm/transformers/models/glm.py,sha256=gHYgfn20jPRL-ElXy-rUqMh6_LQcc5x7DEXSZuRA4E0,7094
+ipex_llm/transformers/models/glm.py,sha256=lmeEWd_W2O638VzVW4Gm6cJre5XZcg_QBmPs8NWqXsM,7202
 ipex_llm/transformers/models/gpt2.py,sha256=YSaNgK1uLCFDuIFqnKO0Mi-AsOZsYav-7pNf_NpKGdM,3445
 ipex_llm/transformers/models/gptbigcode.py,sha256=cP1_qGWoa43R2WacAMblShjku4QupcCZiLaPPAoOUs4,9101
-ipex_llm/transformers/models/gptj.py,sha256=TTIx461X2nOcIkrAcZhEf7d7mjJ3yvEC9KLVc1-hrpc,17973
 ipex_llm/transformers/models/gptneox.py,sha256=loRh1x_5S6BCeOr_s5xr-N_1SQHL3Y5IiUBAEyoMUqQ,6172
-ipex_llm/transformers/models/internlm.py,sha256=ZbIUMDwNRcrCeduXfbA_uq1AUEWawEt6CJRvQl3LkAg,17832
+ipex_llm/transformers/models/internlm.py,sha256=OifyiobRligleyZLpLBSe44A6Sq0uMG-8-NOcRCcT4Q,18080
 ipex_llm/transformers/models/internvl.py,sha256=Vx0vENIEQLX2M6P398mw5TOhpks0U8xf8rtRQvy94go,8154
-ipex_llm/transformers/models/llama.py,sha256=ozwtdQ0MbanJEtW4LBFGxqs_QAq82EonhL2dL6tGyw0,8567
-ipex_llm/transformers/models/minicpm.py,sha256=ib2rJTN7Tf7znBCtVrtXsF-_Uuk2aA7KVg02xzatLiI,10103
-ipex_llm/transformers/models/minicpm3.py,sha256=FhNS6mi2rg7dSdF_QQGrao3g9EC6XLn1MTKd-kd0wF0,9191
-ipex_llm/transformers/models/minicpmv.py,sha256=ZV4s48WNIyRoEkvENnlmopnx3ojZANBer0LI6bRtxrY,9826
-ipex_llm/transformers/models/mistral.py,sha256=rE1GWQxXvF6aG-buPHDR13zeynDZEDIubPF4PiVhZbM,7451
-ipex_llm/transformers/models/mllama.py,sha256=ogpLmmN_OwcFUyjYB-oDC-l3uw8urFvUEc5edkjWHAk,10939
+ipex_llm/transformers/models/llama.py,sha256=NzpyQve_RC9ez1W-jWPLGZ80k_S1I5Rx5saAzCsDIoI,8558
+ipex_llm/transformers/models/minicpm.py,sha256=eaPNVNrep0_xGoELhZd886ff0ceoKqB6cusdAhd52eE,10145
+ipex_llm/transformers/models/minicpm3.py,sha256=11cYl8KM2hoIJNMAOZMxiwCu6dMhup9ric_OEn8-VrQ,9363
+ipex_llm/transformers/models/minicpmv.py,sha256=PP05b5iTnrMpiseCn8iJcxKJDnfq7WqXp9Mrch0kKZ0,9876
+ipex_llm/transformers/models/mistral.py,sha256=uVhkdXaq15v1P3QY0emVsA7SxUbAWChHEEXYN-drjpQ,7449
+ipex_llm/transformers/models/mllama.py,sha256=ZyRq9DTKsvk1AlRbr-z6ngjS3Sr_7YuGZ6-Yr1MBBAM,10937
 ipex_llm/transformers/models/mpt.py,sha256=z02NwHogJZVh-Mk4sYoIzR90SFIKhoNN_-ifsD907TQ,9540
 ipex_llm/transformers/models/phi.py,sha256=E6qz4EEuHIVGvaPo-wtLC5lz3iyMqTbAE_cRlcjQRKI,6670
-ipex_llm/transformers/models/phi3.py,sha256=jkiadJ85ToHpymY5GOM6orWlnx6LKN8_-v1MUcfGWPg,15159
+ipex_llm/transformers/models/phi3.py,sha256=Fo6PlZ24Gdm7eeeZOTMm1Bfh3U6P4rvq7-_2FHvp0vE,15503
 ipex_llm/transformers/models/phixtral.py,sha256=MDTMghcu7qAmZmRcUGqXXDXhSU3y_N59HRIXmlcjp5g,4890
-ipex_llm/transformers/models/qwen.py,sha256=XIJ_bLzediBURWU-OOS3H6WBIGXQue6jDdUHJsAabwY,19391
-ipex_llm/transformers/models/qwen2.py,sha256=b49HO4GSudwGJ3n6uHVno1oo3DgRt3jOjtQnLOB3cdY,25530
-ipex_llm/transformers/models/qwen2_moe.py,sha256=EA_OYxYAEgrvi7VpDW192AJXG9Fwe2aBtOAZPkOAJk4,19350
-ipex_llm/transformers/models/qwen2_vl.py,sha256=jIm4yZSd751BkRqgj3wR1QBkDIh-TMCLAMM8SZ8n6Qo,13419
+ipex_llm/transformers/models/qwen.py,sha256=A3WiVCzA7NLkcjp4zhFkZvKZzZWZlg0WFuVV_556TAI,19543
+ipex_llm/transformers/models/qwen2.py,sha256=JLaY9ZT7A22oO0G8K-nvjvKQDaIrKA5o-jEHvk_y3eI,25604
+ipex_llm/transformers/models/qwen2_moe.py,sha256=a0gYo-ngf8SxaEnBdZUJDnPS6Mkn_poDd8xqhx50icI,19516
+ipex_llm/transformers/models/qwen2_vl.py,sha256=NrhxlaPj7W-HUBmKc3CSTwZy1lkoZ9qDaxM4GvE0kHs,13583
 ipex_llm/transformers/models/qwen_vl.py,sha256=j7Nzzz2Qvynu9yrCXmoEfERjw43hXof5TbXIs7Ms-oY,17105
 ipex_llm/transformers/models/rwkv4.py,sha256=H4KMtxN0JA2ZTXnonHpsUUJ5xULemo-D1Jzl0ri_UY8,6123
 ipex_llm/transformers/models/rwkv5.py,sha256=OkRNj1pCAZg1z2Fw-I0DEnxLEdZyPeRSQ6msrkxLOCs,10710
 ipex_llm/transformers/models/sd.py,sha256=VvHV5u-0k2MgHu3NL9113hPj7DgfxqctuKzEEeNfRDU,5981
-ipex_llm/transformers/models/stablelm.py,sha256=RGQCYuQhYqtZ1j3RZkYi0_QvCRnUgUIPYxfBcLnElzg,6885
-ipex_llm/transformers/models/starcoder2.py,sha256=4P3mhRYf2Kreb1ESjrQGfy1puLMmZXgV35zf-Tksvao,6462
-ipex_llm/transformers/models/utils.py,sha256=Qbz7UkYSbsM5bodH2445O0-JF50Mu3UEwW0j2ZNxHSU,15997
-ipex_llm/transformers/models/yuan.py,sha256=1jRPebwAK2ENbyYokOmb4LSVo-szucWiygz9zTv-scs,7656
+ipex_llm/transformers/models/stablelm.py,sha256=fj-XtOnR6kggnFUQTMPCOOzolkPztN06WAv8QW-XRnI,7054
+ipex_llm/transformers/models/starcoder2.py,sha256=ONKvD7JCkRM0DI-R56x28QFBJ7CjD5hOZBQ_3WfOcNk,6626
+ipex_llm/transformers/models/utils.py,sha256=ihbWS5kQK2KHDVPkMhgjik3nM8B2fWf-E-z4BWNUstk,15568
+ipex_llm/transformers/models/yuan.py,sha256=JYAn_ZaSGK0NBJLEIxCACfAq084a66GFJkdd5NbpmMA,7732
 ipex_llm/transformers/npu_models/__init__.py,sha256=ulEUGLjaP48LCrVeury3UxLjXxKzRi0UpSG4bYu-7f8,585
 ipex_llm/transformers/npu_models/baichuan.py,sha256=fJtd7fBrttySghRUgfZTAdxLjsSNC-XL08HISsXigLE,4685
 ipex_llm/transformers/npu_models/baichuan_mp.py,sha256=tHhO-0v5z6IhxsfzAPYWXVbLrV_4z89DIb4JjE3207M,45026
@@ -244,11 +243,11 @@ ipex_llm/vllm/xpu/engine/__init__.py,sha256=pY_CpyuZd72fr6s32ejeKHKFW0K4vUU2rzZj
 ipex_llm/vllm/xpu/engine/engine.py,sha256=k4-D27WS_Gk3mA--w3HWAjPjb4Aiu043MVPi0ZoAUBc,5984
 ipex_llm/vllm/xpu/entrypoints/openai/api_server.py,sha256=GshTZFB8e4PWvqckfbmTOU6b0oLkNn7A-vzLuG9--j8,21544
 ipex_llm/vllm/xpu/entrypoints/openai/cli_args.py,sha256=2rENA2ucynMaIjiZBEh2ez1o5vR32GaP514t39CD7KM,8676
-ipex_llm-2.2.0b20250107.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
-ipex_llm-2.2.0b20250107.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
-ipex_llm-2.2.0b20250107.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
-ipex_llm-2.2.0b20250107.dist-info/METADATA,sha256=rPJCuVvUndZ0XZBTZzlQEPi1y_W0fpmQTEGmogyRzRw,12705
-ipex_llm-2.2.0b20250107.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
-ipex_llm-2.2.0b20250107.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
-ipex_llm-2.2.0b20250107.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
-ipex_llm-2.2.0b20250107.dist-info/RECORD,,
+ipex_llm-2.2.0b20250109.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
+ipex_llm-2.2.0b20250109.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
+ipex_llm-2.2.0b20250109.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
+ipex_llm-2.2.0b20250109.dist-info/METADATA,sha256=gPslIWSw_X5E5ULhQa8rOHeRo_UeBDXCAyPjBSPB-nU,12705
+ipex_llm-2.2.0b20250109.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
+ipex_llm-2.2.0b20250109.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
+ipex_llm-2.2.0b20250109.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
+ipex_llm-2.2.0b20250109.dist-info/RECORD,,

ipex_llm/transformers/models/gptj.py

@@ -1,441 +0,0 @@
-#
-# Copyright 2016 The BigDL Authors.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-#
-# This file is adapted from
-# https://github.com/huggingface/transformers/blob/main/src/transformers/models/gptj/modeling_gptj.py
-#
-
-import torch
-from typing import Optional, Tuple, Union
-from ipex_llm.transformers.models.utils import init_kv_cache, extend_kv_cache, \
-    apply_rotary_pos_emb, append_kv_cache, apply_ipex_rotate_every_two
-from transformers.utils.import_utils import is_torch_fx_proxy
-from transformers.modeling_outputs import BaseModelOutputWithPast
-from transformers.models.gptj.modeling_gptj import GPTJModel
-from ipex_llm.utils.common import invalidInputError
-
-import os
-
-KV_CACHE_ALLOC_BLOCK_LENGTH = int(os.environ.get("KV_CACHE_ALLOC_BLOCK_LENGTH", 256))
-
-
-def _get_embed_positions(self, position_ids):
-    embed_positions = self.embed_positions
-    if embed_positions.device != position_ids.device:
-        embed_positions = embed_positions.to(position_ids.device)
-        self.embed_positions = embed_positions
-    return embed_positions.repeat(position_ids.shape[0], 1, 1)
-
-
-def _attn(
-    self,
-    query,
-    key,
-    value,
-    attention_mask=None,
-    head_mask=None,
-):
-    # compute causal mask from causal mask buffer
-    query_length, key_length = query.size(-2), key.size(-2)
-    causal_mask = self.bias[:, :, key_length - query_length: key_length, :key_length]
-
-    # Keep the attention weights computation in fp32 to avoid overflow issues
-    query = query.to(torch.float32)
-    key = key.to(torch.float32)
-
-    attn_weights = torch.matmul(query, key.transpose(-1, -2))
-
-    mask_value = torch.finfo(attn_weights.dtype).min
-    # Need to be a tensor, otherwise we get error:
-    # `RuntimeError: expected scalar type float but found double`.
-    # Need to be on the same device, otherwise `RuntimeError: ..., x and y to be on the same device`
-    mask_value = torch.tensor(mask_value, dtype=attn_weights.dtype).to(attn_weights.device)
-    attn_weights = torch.where(causal_mask, attn_weights, mask_value)
-
-    attn_weights = attn_weights / self.scale_attn
-
-    if attention_mask is not None:
-        # Apply the attention mask
-        attn_weights = attn_weights + attention_mask
-
-    attn_weights = nn.functional.softmax(attn_weights, dim=-1)
-    attn_weights = attn_weights.to(value.dtype)
-    attn_weights = self.attn_dropout(attn_weights)
-
-    # Mask heads if we want to
-    if head_mask is not None:
-        attn_weights = attn_weights * head_mask
-
-    attn_output = torch.matmul(attn_weights, value)
-
-    return attn_output, attn_weights
-
-
-def gptj_attention_forward(
-    self,
-    hidden_states: torch.FloatTensor,
-    layer_past: Optional[Tuple[torch.Tensor]] = None,
-    attention_mask: Optional[torch.FloatTensor] = None,
-    position_ids: Optional[torch.LongTensor] = None,
-    head_mask: Optional[torch.FloatTensor] = None,
-    use_cache: Optional[bool] = False,
-    rotary_emb: Optional[Tuple]=None,
-    output_attentions: Optional[bool] = False,
-) -> Union[
-    Tuple[torch.Tensor, Tuple[torch.Tensor]],
-    Optional[Tuple[torch.Tensor, Tuple[torch.Tensor], Tuple[torch.Tensor, ...]]],
-]:
-    query = self.q_proj(hidden_states)
-    key = self.k_proj(hidden_states)
-    value = self.v_proj(hidden_states)
-
-    query = self._split_heads(query, self.num_attention_heads, self.head_dim, True)
-    key = self._split_heads(key, self.num_attention_heads, self.head_dim, True)
-    value = self._split_heads(value, self.num_attention_heads, self.head_dim, False)
-
-    sin, cos = rotary_emb
-    use_fuse_rope = hidden_states.device.type == "xpu" and not self.training
-
-    if self.rotary_dim is not None:
-        k_rot = key[:, :, :, : self.rotary_dim]
-        q_rot = query[:, :, :, : self.rotary_dim]
-
-        if use_fuse_rope:
-            apply_ipex_rotate_every_two(q_rot, k_rot, cos, sin)
-        else:
-            k_pass = key[:, :, :, self.rotary_dim:]
-            q_pass = query[:, :, :, self.rotary_dim:]
-            q_rot, k_rot = apply_rotary_pos_emb(q_rot, k_rot, cos, sin, position_ids, "gptj")
-            key = torch.cat([k_rot, k_pass], dim=-1)
-            query = torch.cat([q_rot, q_pass], dim=-1)
-    else:
-        if use_fuse_rope:
-            apply_ipex_rotate_every_two(query, key, cos, sin)
-        else:
-            query, key = apply_rotary_pos_emb(query, key, cos, sin, position_ids, "gptj")
-
-    batch_size, q_len, _ = hidden_states.size()
-
-    key = key.permute(0, 2, 1, 3).contiguous()
-    query = query.permute(0, 2, 1, 3).contiguous()
-
-    kv_seq_len = key.size(-2)
-    device = hidden_states.device
-
-    if layer_past is not None:
-        kv_seq_len += layer_past[0].size(2)
-
-    if layer_past is not None:
-        cache_k = layer_past[0]
-        cache_v = layer_past[1]
-        past_length = cache_k.size(2)
-        if cache_k.stride()[1] < kv_seq_len * cache_k.size(3):
-            new_cache_k, new_cache_v = extend_kv_cache(batch_size,
-                                                       self.num_attention_heads,
-                                                       self.head_dim,
-                                                       past_length,
-                                                       kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH,
-                                                       dtype=cache_v.dtype,
-                                                       device=device)
-            new_cache_k[:] = cache_k
-            new_cache_v[:] = cache_v
-            cache_k = new_cache_k
-            cache_v = new_cache_v
-        key, value = append_kv_cache(cache_k, cache_v, key, value)
-
-    elif use_cache:
-        key_cache, value_cache = init_kv_cache(batch_size,
-                                               self.num_attention_heads,
-                                               self.head_dim,
-                                               kv_seq_len,
-                                               kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH,
-                                               dtype=value.dtype,
-                                               device=device)
-        key_cache[:] = key
-        value_cache[:] = value
-        key = key_cache
-        value = value_cache
-
-    if use_cache is True:
-        present = (key, value)
-    else:
-        present = None
-
-    # compute self-attention: V x Softmax(QK^T)
-    attn_output, attn_weights = self._attn(query, key, value, attention_mask, head_mask)
-
-    attn_output = self._merge_heads(attn_output, self.num_attention_heads, self.head_dim)
-    attn_output = self.out_proj(attn_output)
-    attn_output = self.resid_dropout(attn_output)
-
-    outputs = (attn_output, present)
-    if output_attentions:
-        outputs += (attn_weights,)
-
-    return outputs  # a, present, (attentions)
-
-
-def gptj_block_forward(
-    self,
-    hidden_states: Optional[torch.FloatTensor],
-    layer_past: Optional[Tuple[torch.Tensor]] = None,
-    attention_mask: Optional[torch.FloatTensor] = None,
-    position_ids: Optional[torch.LongTensor] = None,
-    head_mask: Optional[torch.FloatTensor] = None,
-    use_cache: Optional[bool] = False,
-    rotary_emb: Optional[Tuple]=None,
-    output_attentions: Optional[bool] = False,
-) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
-    residual = hidden_states
-    hidden_states = self.ln_1(hidden_states)
-    attn_outputs = self.attn(
-        hidden_states=hidden_states,
-        layer_past=layer_past,
-        attention_mask=attention_mask,
-        position_ids=position_ids,
-        head_mask=head_mask,
-        use_cache=use_cache,
-        rotary_emb=rotary_emb,
-        output_attentions=output_attentions,
-    )
-    attn_output = attn_outputs[0]  # output_attn: a, present, (attentions)
-    outputs = attn_outputs[1:]
-
-    feed_forward_hidden_states = self.mlp(hidden_states)
-    hidden_states = attn_output + feed_forward_hidden_states + residual
-
-    if use_cache:
-        outputs = (hidden_states,) + outputs
-    else:
-        outputs = (hidden_states,) + outputs[1:]
-
-    return outputs  # hidden_states, present, (attentions)
-
-
-def create_sinusoidal_positions(num_pos: int, dim: int) -> torch.Tensor:
-    inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2) / dim))
-    sinusoid_inp = torch.einsum("i , j -> i j",
-                                torch.arange(num_pos, dtype=torch.float), inv_freq).float()
-    return torch.cat((torch.sin(sinusoid_inp), torch.cos(sinusoid_inp)), dim=1)
-
-
-old_init = GPTJModel.__init__
-
-
-def gptj_model_new_init(self, config):
-    old_init(self, config)
-    embed_dim = config.hidden_size
-    rotary_dim = config.rotary_dim
-    pos_embd_dim = rotary_dim or embed_dim
-    max_positions = config.max_position_embeddings
-    self.embed_positions = create_sinusoidal_positions(max_positions, pos_embd_dim)
-
-
-def get_new_embed_positions(position_ids, prev_embed_positions):
-    embed_positions = prev_embed_positions
-    if embed_positions.device != position_ids.device:
-        embed_positions = embed_positions.to(position_ids.device)
-        prev_embed_positions = embed_positions
-    return embed_positions.repeat(position_ids.shape[0], 1, 1), prev_embed_positions
-
-
-def gptj_model_forward(
-    self,
-    input_ids: Optional[torch.LongTensor] = None,
-    past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
-    attention_mask: Optional[torch.FloatTensor] = None,
-    token_type_ids: Optional[torch.LongTensor] = None,
-    position_ids: Optional[torch.LongTensor] = None,
-    head_mask: Optional[torch.FloatTensor] = None,
-    inputs_embeds: Optional[torch.FloatTensor] = None,
-    use_cache: Optional[bool] = None,
-    output_attentions: Optional[bool] = None,
-    output_hidden_states: Optional[bool] = None,
-    return_dict: Optional[bool] = None,
-) -> Union[Tuple, BaseModelOutputWithPast]:
-    output_attentions = output_attentions if output_attentions is not None \
-        else self.config.output_attentions
-    output_hidden_states = (
-        output_hidden_states if output_hidden_states is not None
-        else self.config.output_hidden_states
-    )
-    use_cache = use_cache if use_cache is not None else self.config.use_cache
-    return_dict = return_dict if return_dict is not None else self.config.use_return_dict
-
-    if input_ids is not None and inputs_embeds is not None:
-        invalidInputError(False,
-                          "You cannot specify both input_ids and inputs_embeds at the same time")
-    elif input_ids is not None:
-        self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
-        input_shape = input_ids.size()
-        input_ids = input_ids.view(-1, input_shape[-1])
-        batch_size = input_ids.shape[0]
-    elif inputs_embeds is not None:
-        input_shape = inputs_embeds.size()[:-1]
-        batch_size = inputs_embeds.shape[0]
-    else:
-        invalidInputError(False, "You have to specify either input_ids or inputs_embeds")
-
-    device = input_ids.device if input_ids is not None else inputs_embeds.device
-
-    if token_type_ids is not None:
-        token_type_ids = token_type_ids.view(-1, input_shape[-1])
-
-    if past_key_values is None:
-        past_length = 0
-        past_key_values = tuple([None] * len(self.h))
-    else:
-        past_length = past_key_values[0][0].size(-2)
-
-    if position_ids is None:
-        position_ids = torch.arange(past_length, input_shape[-1] + past_length,
-                                    dtype=torch.long, device=device)
-        position_ids = position_ids.unsqueeze(0)
-
-    # Attention mask.
-    if attention_mask is not None:
-        if batch_size <= 0:
-            invalidInputError(False, "batch_size has to be defined and > 0")
-        attention_mask = attention_mask.view(batch_size, -1)
-        # We create a 3D attention mask from a 2D tensor mask.
-        # Sizes are [batch_size, 1, 1, to_seq_length]
-        # So we can broadcast to [batch_size, num_heads, from_seq_length, to_seq_length]
-        # this attention mask is more simple than the triangular masking of causal attention
-        # used in OpenAI GPT, we just need to prepare the broadcast dimension here.
-        attention_mask = attention_mask[:, None, None, :]
-
-        # Since attention_mask is 1.0 for positions we want to attend and 0.0 for
-        # masked positions, this operation will create a tensor which is 0.0 for
-        # positions we want to attend and the dtype's smallest value for masked positions.
-        # Since we are adding it to the raw scores before the softmax, this is
-        # effectively the same as removing these entirely.
-        attention_mask = attention_mask.to(dtype=self.dtype)  # fp16 compatibility
-        attention_mask = (1.0 - attention_mask) * torch.finfo(self.dtype).min
-
-    # Prepare head mask if needed
-    # 1.0 in head_mask indicate we keep the head
-    # attention_probs has shape bsz x num_attention_heads x N x N
-    # head_mask has shape n_layer x batch x num_attention_heads x N x N
-    head_mask = self.get_head_mask(head_mask, self.config.n_layer)
-
-    if inputs_embeds is None:
-        inputs_embeds = self.wte(input_ids)
-
-    hidden_states = inputs_embeds
-
-    if token_type_ids is not None:
-        token_type_embeds = self.wte(token_type_ids)
-        hidden_states = hidden_states + token_type_embeds
-
-    hidden_states = self.drop(hidden_states)
-
-    output_shape = (-1,) + input_shape[1:] + (hidden_states.size(-1),)
-
-    if self.gradient_checkpointing and self.training:
-        if use_cache:
-            logger.warning_once(
-                "`use_cache=True` is incompatible with gradient checkpointing."
-                "Setting `use_cache=False`..."
-            )
-            use_cache = False
-
-    presents = () if use_cache else None
-    all_self_attentions = () if output_attentions else None
-    all_hidden_states = () if output_hidden_states else None
-
-    # Repeat cos sin here, call only once for each token.
-    # If put this to attension forward, it will generate too many times.
-    if is_torch_fx_proxy(position_ids) or torch.jit.is_tracing():
-        # The logic to conditionally copy to GPU could not be traced, so we do this
-        # every time in the torch.fx case
-        embed_positions = get_embed_positions(self.embed_positions, position_ids)
-    else:
-        embed_positions, self.embed_positions = get_new_embed_positions(position_ids,
-                                                                        self.embed_positions)
-
-    repeated_position_ids = position_ids.unsqueeze(-1).repeat(1, 1, embed_positions.shape[-1])
-    sincos = torch.gather(embed_positions, 1, repeated_position_ids)
-    sin, cos = torch.split(sincos, sincos.shape[-1] // 2, dim=-1)
-    sin = torch.repeat_interleave(sin[:, :, None, :], 2, 3)
-    cos = torch.repeat_interleave(cos[:, :, None, :], 2, 3)
-
-    for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
-        # Model parallel
-        if self.model_parallel:
-            torch.cuda.set_device(hidden_states.device)
-            # Ensure layer_past is on same device as hidden_states (might not be correct)
-            if layer_past is not None:
-                layer_past = tuple(past_state.to(hidden_states.device) for past_state in layer_past)
-            # Ensure that attention_mask is always on the same device as hidden_states
-            if attention_mask is not None:
-                attention_mask = attention_mask.to(hidden_states.device)
-            if isinstance(head_mask, torch.Tensor):
-                head_mask = head_mask.to(hidden_states.device)
-        if output_hidden_states:
-            all_hidden_states = all_hidden_states + (hidden_states,)
-
-        if self.gradient_checkpointing and self.training:
-            outputs = self._gradient_checkpointing_func(
-                block.__call__,
-                hidden_states,
-                None,
-                attention_mask,
-                position_ids,
-                head_mask[i],
-                use_cache,
-                output_attentions,
-            )
-        else:
-            outputs = block(
-                hidden_states=hidden_states,
-                layer_past=layer_past,
-                attention_mask=attention_mask,
-                position_ids=position_ids,
-                head_mask=head_mask[i],
-                use_cache=use_cache,
-                rotary_emb=(sin, cos),
-                output_attentions=output_attentions,
-            )
-
-        hidden_states = outputs[0]
-        if use_cache is True:
-            presents = presents + (outputs[1],)
-
-        if output_attentions:
-            all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],)
-
-        # Model Parallel: If it's the last layer for that device, put things on the next device
-        if self.model_parallel:
-            for k, v in self.device_map.items():
-                if i == v[-1] and "cuda:" + str(k) != self.last_device:
-                    hidden_states = hidden_states.to("cuda:" + str(k + 1))
-
-    hidden_states = self.ln_f(hidden_states)
-
-    hidden_states = hidden_states.view(output_shape)
-    # Add last hidden state
-    if output_hidden_states:
-        all_hidden_states = all_hidden_states + (hidden_states,)
-
-    if not return_dict:
-        return tuple(v for v in [hidden_states, presents, all_hidden_states, all_self_attentions]
-                     if v is not None)
-
-    return BaseModelOutputWithPast(
-        last_hidden_state=hidden_states,
-        past_key_values=presents,
-        hidden_states=all_hidden_states,
-        attentions=all_self_attentions,
-    )