+ ܳj ~^RIt^RIHt^RIHt!RR]4t!RR]4t!RR ]4t!R R ]4tR#) N) Optimizer)Optimizer1Statec6aa]tRt^ toRV3RlltRtVtV;t#)LARSc  `<V^8Xd \R4h\S V` RVVW43RVVVV V R7 R#)a Base LARS optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize. lr (`float`): The learning rate. momentum (`float`, defaults to 0): The momentum value speeds up the optimizer by taking bigger steps. dampening (`float`, defaults to 0): The dampening value reduces the momentum of the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. nesterov (`bool`, defaults to `False`): Whether to use Nesterov momentum. optim_bits (`int`, defaults to 32): The number of bits of the optimizer state. args (`object`, defaults to `None`): An object with additional arguments. min_8bit_size (`int`, defaults to 4096): The minimum number of elements of the parameter tensors for 8-bit optimization. max_unorm (`float`, defaults to 0.02): The maximum gradient norm. 'LARS without momentum is not supported!lars max_unormNNotImplementedErrorsuper__init__) selfparamslrmomentum dampening weight_decaynesterov optim_bitsargs min_8bit_sizer __class__s &&&&&&&&&&&X/home/luca/screen-checking/.venv/lib/python3.14/site-packages/bitsandbytes/optim/lars.pyr LARS.__init__ sLL q=%&OP P      !       )rrrF N{Gz?__name__ __module__ __qualname____firstlineno__r__static_attributes____classdictcell__ __classcell__r __classdict__s@@rrr s3 3 rrc6aa]tRt^BtoRV3RlltRtVtV;t#)LARS8bitc  `<V^8Xd \R4h\S V` RVVW43RV^VVV R7 R#)aS 8-bit LARS optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize. lr (`float`): The learning rate. momentum (`float`, defaults to 0): The momentum value speeds up the optimizer by taking bigger steps. dampening (`float`, defaults to 0): The dampening value reduces the momentum of the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. nesterov (`bool`, defaults to `False`): Whether to use Nesterov momentum. args (`object`, defaults to `None`): An object with additional arguments. min_8bit_size (`int`, defaults to 4096): The minimum number of elements of the parameter tensors for 8-bit optimization. max_unorm (`float`, defaults to 0.02): The maximum gradient norm. rr r r Nr rrrrrrrrrr rs &&&&&&&&&&rrLARS8bit.__init__CsLF q=%&OP P      !      rrrrrFNr!r"r#r+s@@rr.r.B0 0 rr.c6aa]tRt^vtoRV3RlltRtVtV;t#) LARS32bitc  `<V^8Xd \R4h\S V` RVVW43RV^ VVV R7 R#)aT 32-bit LARS optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize. lr (`float`): The learning rate. momentum (`float`, defaults to 0): The momentum value speeds up the optimizer by taking bigger steps. dampening (`float`, defaults to 0): The dampening value reduces the momentum of the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. nesterov (`bool`, defaults to `False`): Whether to use Nesterov momentum. args (`object`, defaults to `None`): An object with additional arguments. min_8bit_size (`int`, defaults to 4096): The minimum number of elements of the parameter tensors for 8-bit optimization. max_unorm (`float`, defaults to 0.02): The maximum gradient norm. rr r r Nr r0s &&&&&&&&&&rrLARS32bit.__init__wsLF q=%&OP P      !       rrr2r#r+s@@rr5r5vr3rr5ctaa]tRt^toRV3RlltV3Rlt]P!4RRl4tRt Vt V;t #) PytorchLARSc <VR8d\RV 24hVR8d\RV 24hVR8d\RV 24h\VVVVVVR7pV'dV^8:gV^8wd \R4h\S V` W4R#)r zInvalid learning rate: zInvalid momentum value: zInvalid weight_decay value: )rrrrrr z8Nesterov momentum requires a momentum and zero dampeningN) ValueErrordictrr) rrrrrrrr defaultsrs &&&&&&&& rrPytorchLARS.__init__s 86rd;< < c>7zBC C # ;L>JK K%   Q)q.WX X *rcp<\SV`V4VPFpVPRR4K R#)rFN)r __setstate__ param_groups setdefault)rstategrouprs&& rr@PytorchLARS.__setstate__s0 U#&&E   Z /'rc *RpVe.\P!4;_uu_4V!4pRRR4VPEFpVR,pVR,pVR,pVR,pVR,pVR,p VR,EFrp V PfKVPV ,p V Pp V^8wdV P WR 7p V^8wdV P R R4p V f*\P!V 4P4p WR &M)V PV4PV ^V, R 7V'dWV,,pMT pR pVR 8dzV P\P8XgQh\P!V P44p\P!X4pVVV,8dVV,V, pV PXV )V,R 7EKu EK V# +'giEL;i) zPerforms a single optimization step. Args: closure (callable, optional): A closure that reevaluates the model and returns the loss. Nrrrrr rr)alphamomentum_bufferg?r )torch enable_gradrAgradrCaddgetclonedetachmul_add_dtypefloat32norm)rclosurelossrDrrrrr rprCd_pbufupdate update_scalepnormunorms&& rstepPytorchLARS.steps  ""$$y%&&E 0LZ(Hk*IZ(Hk*ItB8__66> 1 ff1$''!'8Cq=))$5trcsE  !84 ?4 n1 1 h1 1 hY)Yr