$hR7ddlZddlmZddlmZmZmZmZmZmZm Z m Z ddl m Z m Z ddgZGddeZd d e d e d ed ze_ dde ede ede ede edede edededededededefdZdZde ede ede ede ededededededededefdZde ede ede ede ededededededededefdZdS) N)Tensor) Optimizer_use_grad_for_differentiable _get_value _view_as_real_default_to_fused_or_foreach_differentiable_doc _foreach_doc _maximize_doc)ListOptionalAdagradadagradc|eZdZ dddddeeded effd Zfd Zd Zd Ze ddZ xZ S)r{Gz?r绽|=NF)maximizedifferentiableforeachrrc d|kstd|d|kstd|d|kstd|d|kstd|d|kstd|t|||||||| } t|| |jD]} | dD]|} |j| } t jdt j | d <t j | rt||n|}t j | |t j | d <}dS) NgzInvalid learning rate: zInvalid lr_decay value: zInvalid weight_decay value: z)Invalid initial_accumulator_value value: zInvalid epsilon value: )lrlr_decayeps weight_decayinitial_accumulator_valuerrrparamsdtypestep) memory_formatsum) ValueErrordictsuper__init__ param_groupsstatetorchtensorfloat32 is_complexcomplex full_likepreserve_format)selfrrrrrrrrrdefaultsgrouppr( init_value __class__s S/var/www/html/auto_sub_bot/venv/lib/python3.11/site-packages/torch/optim/adagrad.pyr&zAdagrad.__init__ sbyy;r;;<< <hBBBCC Cl""JLJJKK K///W2z*%L /i(z*$%56'       s /33)rrrrrNN) __name__ __module__ __qualname__rboolr&r9rErPrr __classcell__)r5s@r6rr s "# "&1$111$11111111f P P P P P--- , , ,"%%%"!%%%%%r7a[Implements Adagrad algorithm. .. math:: \begin{aligned} &\rule{110mm}{0.4pt} \\ &\textbf{input} : \gamma \text{ (lr)}, \: \theta_0 \text{ (params)}, \: f(\theta) \text{ (objective)}, \: \lambda \text{ (weight decay)}, \\ &\hspace{12mm} \tau \text{ (initial accumulator value)}, \: \eta\text{ (lr decay)}\\ &\textbf{initialize} : state\_sum_0 \leftarrow 0 \\[-1.ex] &\rule{110mm}{0.4pt} \\ &\textbf{for} \: t=1 \: \textbf{to} \: \ldots \: \textbf{do} \\ &\hspace{5mm}g_t \leftarrow \nabla_{\theta} f_t (\theta_{t-1}) \\ &\hspace{5mm} \tilde{\gamma} \leftarrow \gamma / (1 +(t-1) \eta) \\ &\hspace{5mm} \textbf{if} \: \lambda \neq 0 \\ &\hspace{10mm} g_t \leftarrow g_t + \lambda \theta_{t-1} \\ &\hspace{5mm}state\_sum_t \leftarrow state\_sum_{t-1} + g^2_t \\ &\hspace{5mm}\theta_t \leftarrow \theta_{t-1}- \tilde{\gamma} \frac{g_t}{\sqrt{state\_sum_t}+\epsilon} \\ &\rule{110mm}{0.4pt} \\[-1.ex] &\bf{return} \: \theta_t \\[-1.ex] &\rule{110mm}{0.4pt} \\[-1.ex] \end{aligned} For further details regarding the algorithm we refer to `Adaptive Subgradient Methods for Online Learning and Stochastic Optimization`_. a Args: params (iterable): iterable of parameters to optimize or dicts defining parameter groups lr (float, optional): learning rate (default: 1e-2) lr_decay (float, optional): learning rate decay (default: 0) weight_decay (float, optional): weight decay (L2 penalty) (default: 0) eps (float, optional): term added to the denominator to improve numerical stability (default: 1e-10) z z .. _Adaptive Subgradient Methods for Online Learning and Stochastic Optimization: http://jmlr.org/papers/v12/duchi11a.html FrrKrLrMrNrrrOrrrrrcntd|Dstd|t||d\} }|r-tjrtd|r&tjst }nt}||||||| | | || || dS)ztFunctional API that performs Adagrad algorithm computation. See :class:`~torch.optim.Adagrad` for details. c3JK|]}t|tjVdSrU) isinstancer)r).0ts r6 zadagrad..s.@@qz!U\**@@@@@@r7zPAPI has changed, `state_steps` argument must contain a list of singleton tensorsNF) use_fusedz6torch.jit.script not supported with foreach optimizersrrrrrNrrrO)all RuntimeErrorr r)jit is_scripting_multi_tensor_adagrad_single_tensor_adagrad)rrKrLrMrNrrrOrrrrr_funcs r6rrs, @@K@@@ @ @  ^   1&.TYZZZ 7U59))++USTTT&uy--//&$%D  ! '%      r7c|}|dks|dkrtj|Stj|||S)Nr)sizenumelr) empty_likesparse_coo_tensor)rG grad_indicesr<rls r6 _make_sparserqs^ 99;;Dq  FLLNNa$7$7%%%  "< > >>r7c t||||D]F\} } }}|dz }t|}| s| n| } |dkr-| jrtd| | |} |d|dz |zzz }| jr| } | }| }|t| || d| | }| |}| t| |||z | Xtj| }|r ===??D==??L,,..K NN<lKOOA.Gs$8a8aD8a8a8a8a8a8ar7TFrbg?cpu)devicersrcHg|]} dt|dz zzz S)r)r)r^r rrs r6 z)_multi_tensor_adagrad..qs7bbbtbSAD!1!1A!5 AABbbbr7rv)r=r"_group_tensors_by_device_and_dtyper<anyrhr) _foreach_negris_cpu _foreach_add_r* _foreach_add_foreach_addcmul_ _foreach_sqrt _foreach_mul_ _foreach_mul_foreach_addcdiv_)rrKrLrMrrrrrNrrrOgrouped_tensorlists device_params device_gradsdevice_state_sumsdevice_state_stepsridevice_has_sparse_grad minus_clrr numerators ` ` r6rgrg/sq DDDDD  6{{a#FPUWacnGoppUhUoUoUqUq9?9?Q M-'8:Lq!0!aS8a8aT`8a8a8a5a5a !  "!")! $-'       < -l;;L  J -7H I I I a ' 7   2ELU4S4S4S[^ _ _ _ _ _   2A 6 6 6 1   c#L-|TTTTT$1, Uabbb bbbbbOabbb   1<UVWWWW!"344 C%%% 1      i 8 8 8$II*<CCI  y#>>>>s9?9?r7)NNFF)r)r optimizerrrrrr r r r typingr r__all__r__doc__rYr?rrqrhrgr7r6rsI hhhhhhhhhhhhhhhhhhhh!!!!!!!! i }}}}}i}}}@2           3)f!" 33 L3 <3V 3f 33d^333 333 3 !3333l???8= L8= <8=V 8=f 8=  8=8=8= 8=8=8=8=8=8=8=8=vP? LP? <P?V P?f P?  P?P?P? P?P?P?P?P?P?P?P?P?P?r7