#j>8,ddlZddlmZddlmZddlmZmZmZm Z m Z m Z m Z m Z mZmZmZmZmZmZmZmZmZmZmZmZmZmZmZmZddlmZddl m!Z!dd l"m#Z#dd l$m%Z%dd l&m'Z'dd l(m)Z)dd l*m+Z+ddl,m-Z-ddl.m/Z/ddl0m1Z1ddl2m3Z3ddl4m5Z5ddl6m7Z7ddl8m9Z9ddl:m;Z;ddlm?Z?ddl@mAZAddlBmCZCddlDmEZEddlFmGZGddlHmIZIddlJmKZKddlLmMZMide%d e)d!e+d"e9d#e!d$eCd%eEd&e5d'e7d(e/d)e;d*e=d+e-d,e1d-eKd.e'd/eMeIe3e#eAe?eGd0ZNided e d!e d(e d"ed#ed$ed%ed&ed'ed*ed+e d,e d)ed-ed.e d/eeeeeeed0ZOeeee e eeefZPejQeRZSGd1d2ZTGd3d4ZUd5eVfd6ZWd7eVfd8ZXd9ZYdS):N) AutoConfig)logging) AqlmConfigAutoRoundConfig AwqConfigBitNetQuantConfigBitsAndBytesConfigCompressedTensorsConfig EetqConfigFbgemmFp8ConfigFineGrainedFP8ConfigFourOverSixConfig FPQuantConfig GPTQConfig HiggsConfig HqqConfig MetalConfig Mxfp4ConfigQuantizationConfigMixinQuantizationMethod QuantoConfig QuarkConfig SinqConfig SpQRConfig TorchAoConfig VptqConfig) HfQuantizer)AqlmHfQuantizer)AutoRoundQuantizer) AwqQuantizer)BitNetHfQuantizer)Bnb4BitHfQuantizer)Bnb8BitHfQuantizer)CompressedTensorsHfQuantizer)EetqHfQuantizer)FbgemmFp8HfQuantizer)FineGrainedFP8HfQuantizer)FourOverSixHfQuantizer)FPQuantHfQuantizer)GptqHfQuantizer)HiggsHfQuantizer)HqqHfQuantizer)MetalHfQuantizer)Mxfp4HfQuantizer)QuantoHfQuantizer)QuarkHfQuantizer)SinqHfQuantizer)SpQRHfQuantizer)TorchAoHfQuantizer)VptqHfQuantizerawqbitsandbytes_4bitbitsandbytes_8bitgptqaqlmquantoquark fouroversixfp_quanteetqhiggshqqzcompressed-tensors fbgemm_fp8torchaobitnetvptq)spqrfp8z auto-roundmxfp4metalsinqcDeZdZdZedefdZedZdS)AutoQuantizationConfigz The Auto-HF quantization config class that takes care of automatically dispatching to the correct quantization config given a quantization config stored in a dictionary. quantization_config_dictc |d}|dds|ddr*|ddrdnd}tj|z}n|td|tvr9td|d t t t|}||S) N quant_method load_in_8bitF load_in_4bit_4bit_8bitThe model's quantization config from the arguments has no `quant_method` attribute. Make sure that the model has been correctly quantizedUnknown quantization type, got  - supported types are: ) getrBITS_AND_BYTES ValueError AUTO_QUANTIZATION_CONFIG_MAPPINGlistAUTO_QUANTIZER_MAPPINGkeys from_dict)clsrNrPsuffix target_clss f/lsinfo/ai/hellotax_ai/base_platform/venv/lib/python3.11/site-packages/transformers/quantizers/auto.pyr_z AutoQuantizationConfig.from_dicts/33NCC # ' ' > > BZB^B^_motBuBu  8 < <^U S S`WWY`F->>>[>( # #[ # # #rdrMceZdZdZedeezfdZedZedeezdedzfdZ e dZ dS) AutoHfQuantizerz The Auto-HF quantizer class that takes care of automatically instantiating to the correct `HfQuantizer` given the `QuantizationConfig`. rfc t|trt|}|j}|t jkr6t|tstd|j r|dz }n|dz }|tvr9td|dtt t|}||fi|S)NzZFound `quant_method=bitsandbytes` but `quantization_config` is not a `BitsAndBytesConfig`.rTrSrVrW) isinstancersrMr_rPrrYr TypeErrorrQr]rZr\r^)r`rfrlrPrbs rc from_configzAutoHfQuantizer.from_configs )4 0 0 X"8"B"BCV"W"W *7  -< < <13EFF p#/ (' ' 5 5 5:,::/446677::  ,L9 z-88888rdc Ptj|fi|}||S)N)rMrhry)r`rkrlrfs rcrhzAutoHfQuantizer.from_pretraineds04DEbmmflmm2333rdquantization_config_from_argsNc&|d}nd}t|trDt|trtj|}nt|}|D|jj|jjkr*td|jjd|jjdt|tr~t|tri| }| D]\}}t||||r(|dt| dz }|dkr7t|ttt fst#j|nt&||S) z handles situations where both quantization_config from args and quantization_config from model config are present. NzYou passed `quantization_config` or equivalent parameters to `from_pretrained` but the model you're loading already has a `quantization_config` attribute. The `quantization_config` from the model will be used.zThe model is quantized with z but you are passing a z| config. Please make sure to pass the same quantization config class to `from_pretrained` with different loading attributes.z"However, loading attributes (e.g. z]) will be overwritten with the one you passed to `from_pretrained`. The rest will be ignored.)rwrsrr_rM __class__rnrZLOADING_ATTRIBUTES_CONFIG_TYPESget_loading_attributesitemssetattrr\r^rrrwarningswarnloggerinfo)r`rfr{ warning_msgloading_attr_dictattrvals rcmerge_quantization_configsz*AutoHfQuantizer.merge_quantization_configss ) 4y K K )4 0 0 \7II \&5&?@S&T&T##&<&F&FGZ&[&[# * 5#-6:W:a:jjjF/B/L/UFFoLoVo_FFF  )+J K K BPZ )+JQ Q  B!> T T V V .4466 8 8 c+T37777  B BDIZI_I_IaIaDbDb B B BB "  Z0CkS^`tEu%v%v  M+ & & & & KK $ $ $""rdc |dd}|dds|ddr*|ddrdnd}tj|z}n|td|tvrGt d|d tt d dSd S) NrPrQFrRrSrTrUrVrWz~. Hence, we will skip the quantization. To remove the warning, you can delete the quantization_config attribute in config.jsonT) rXrrYrZr[rwarningr\r]r^)rNrPras rcsupports_quant_methodz%AutoHfQuantizer.supports_quant_method s/33NDII # ' ' > > BZB^B^_motBuBu  8 < <^U S S`WWY`F-.register_config_fn$s\ 5 5 5DDDDEE E#677 JHII I36(0 rdrg)rrs` rcregister_quantization_configr!s$ rdnamecfd}|S)zRegister a custom quantizer.ctvrtddt|tst d|t<|S)Nz Quantizer 'rz!Quantizer must extend HfQuantizer)r]rZrrrx)r`rs rcregister_quantizer_fnz1register_quantizer..register_quantizer_fn4s[ ) ) )E4EEEFF F#{++ A?@@ @'*t$ rdrg)rrs` rcregister_quantizerr1s$ ! rdc<t|d}|r!t|jsd}|s|Q|r&t|j||_n||_t|j|}nd}||||||}||}| |}t|jdds |jj }t|d||d<|||fS)NrfF) pre_quantized) device_map weights_only dequantizevaluequant) hasattrrurrfrryvalidate_environmentupdate_device_mapupdate_tp_planupdate_ep_planrirP)configrfrr user_agentr hf_quantizerrPs rcget_hf_quantizerrAsSF$9::M_BB6C]^^  +7  =)8)S)S*,?**F & &*=F &&22  &'3   ))!% *   "33J?? ,,V44,,V44|7uMM O';HL"),"N"NJw   ++rd)Zrmodels.auto.configuration_autorutilsrutils.quantization_configrrrr r r r r rrrrrrrrrrrrrrrrbaserquantizer_aqlmr quantizer_auto_roundr! quantizer_awqr"quantizer_bitnetr#quantizer_bnb_4bitr$quantizer_bnb_8bitr%quantizer_compressed_tensorsr&quantizer_eetqr'quantizer_fbgemm_fp8r(quantizer_finegrained_fp8r)quantizer_fouroversixr*quantizer_fp_quantr+quantizer_gptqr,quantizer_higgsr- quantizer_hqqr.quantizer_metalr/quantizer_mxfp4r0quantizer_quantor1quantizer_quarkr2quantizer_sinqr3quantizer_spqrr4quantizer_torchaor5quantizer_vptqr6r]r[r get_loggerrnrrMrustrrrrrgrdrcrs7777774++++++444444''''''//////222222222222FFFFFF++++++666666@@@@@@999999222222++++++------))))))------------//////------++++++++++++111111++++++ <++ O   O     )" O  >6&!  !" O#$  $$   /4$ 9$+$+$ J $  J $  J $ l$ [$$$ $ 9$1$/$ [$}$  !$" J#$$  !   /$$$ 6  #  H % %&#&#&#&#&#&#&#&#Rllllllll^      !S ! ! ! ! !,!,!,!,!,rd