+ ܳj0^RIHt!RR]4t!RR]4t!RR]4t!RR ]4t!R R ]4t!R R ]4tR#))Optimizer2Statec6aa]tRt^ toRV3RlltRtVtV;t#)AdamWc  :<\S V`RVVVVVVVV V R7 R#)ak Base AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. 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. is_paged (`bool`, defaults to `False`): Whether the optimizer is a paged optimizer or not. adamis_pagedNsuper__init__ selfparamslrbetaseps weight_decayamsgrad optim_bitsargs min_8bit_sizer __class__s &&&&&&&&&&&Y/home/luca/screen-checking/.venv/lib/python3.14/site-packages/bitsandbytes/optim/adamw.pyr AdamW.__init__ s6L            MbP?g?g+?:0yE>{Gz?F NF__name__ __module__ __qualname____firstlineno__r __static_attributes____classdictcell__ __classcell__r __classdict__s@@rrr 1 1 rrc6aa]tRt^>toRV3RlltRtVtV;t#) AdamW8bitc  <V'd \R4hV^ 8wd \R4h\S V` RVVVVV^VV V R7 R#)a 8-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. Note: This parameter is not supported in AdamW8bit and must be False. optim_bits (`int`, defaults to 32): The number of bits of the optimizer state. Note: This parameter is not used in AdamW8bit as it always uses 8-bit optimization. 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. is_paged (`bool`, defaults to `False`): Whether the optimizer is a paged optimizer or not. z'AdamW8bit does not support amsgrad=TruezGAdamW8bit only supports optim_bits=32 (default value for compatibility)rrN ValueErrorr r r s &&&&&&&&&&&rr AdamW8bit.__init__?s[R FG G  fg g           rrrr$r,s@@rr0r0>s< < rr0c6aa]tRt^~toRV3RlltRtVtV;t#) AdamW32bitc  :<\S V`RVVVVV^ VV V R7 R#)am 32-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. optim_bits (`int`, defaults to 32): The number of bits of the optimizer state. args (`object`, defaults to `None`): An object with additional arguments. min_8bit_size (`int`, defaults to 4096): The minimum number of elements of the parameter tensors for 8-bit optimization. is_paged (`bool`, defaults to `False`): Whether the optimizer is a paged optimizer or not. rrNr r s &&&&&&&&&&&rr AdamW32bit.__init__s6L            rrrr$r,s@@rr6r6~r.rr6c6aa]tRt^toRV3RlltRtVtV;t#) PagedAdamWc  :<\S V`RVVVVVVVV RR7 R#)a Paged AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. 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. rTrNr rrrrrrrrrrrs &&&&&&&&&&rr PagedAdamW.__init__s6F            rrrrr r!Fr"Nr#r$r,s@@rr:r:. . rr:c6aa]tRt^toRV3RlltRtVtV;t#)PagedAdamW8bitc  <V'd \R4hV^ 8wd \R4h\S V` RVVVVV^VV RR7 R#)a Paged 8-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. Note: This parameter is not supported in PagedAdamW8bit and must be False. optim_bits (`int`, defaults to 32): The number of bits of the optimizer state. Note: This parameter is not used in PagedAdamW8bit as it always uses 8-bit optimization. args (`object`, defaults to `None`): An object with additional arguments. min_8bit_size (`int`, defaults to 4096): The minimum number of elements of the parameter tensors for 8-bit optimization. z,PagedAdamW8bit does not support amsgrad=TruezLPagedAdamW8bit only supports optim_bits=32 (default value for compatibility)rTrNr2r<s &&&&&&&&&&rr PagedAdamW8bit.__init__s[L KL L  kl l           rrr>r$r,s@@rrArAs9 9 rrAc6aa]tRtRtoRV3RlltRtVtV;t#)PagedAdamW32biti"c  :<\S V`RVVVVV^ VV RR7 R#)a  Paged 32-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize. lr (`float`, defaults to 1e-3): The learning rate. betas (`tuple(float, float)`, defaults to (0.9, 0.999)): The beta values are the decay rates of the first and second-order moment of the optimizer. eps (`float`, defaults to 1e-8): The epsilon value prevents division by zero in the optimizer. weight_decay (`float`, defaults to 1e-2): The weight decay value for the optimizer. amsgrad (`bool`, defaults to `False`): Whether to use the [AMSGrad](https://hf.co/papers/1904.09237) variant of Adam that uses the maximum of past squared gradients instead. 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. rTrNr r<s &&&&&&&&&&rr PagedAdamW32bit.__init__#s6F            rrr>r$r,s@@rrErE"r?rrEN)bitsandbytes.optim.optimizerrrr0r6r:rArErrrrIsW 92 O2 j= = @2 2 j/ / d: _: z/ o/ r