
    wiVD                     v   d Z ddlmZmZmZmZmZ ddlZddlmZ ddl	m
Z
mZmZmZmZmZmZmZmZmZmZmZmZ ddgZ G d	 de          Zd
de de
 de de de dz   e_         dee         dee         dee         dee         dee         dedededededededefdZdee         dee         dee         dee         dee         dedededededededefdZ ee          	 	 	 	 	 d"dee         dee         dee         dee         dee         d ee         dedededededededefd!            ZdS )#z1Implementation for the Resilient backpropagation.    )castListOptionalTupleUnionN)Tensor   )_capturable_doc_default_to_fused_or_foreach_differentiable_doc_disable_dynamo_if_unsupported_foreach_doc!_get_capturable_supported_devices_get_scalar_dtype_maximize_doc_params_doc_use_grad_for_differentiable_view_as_real	OptimizerParamsTRproprpropc                        e Zd Z	 	 	 dddddddedeeef         d	eeef         d
eeef         dede	e         dedef fdZ
 fdZd Zedd            Z xZS )r   {Gz?g      ?g333333?gư>2   FN)
capturableforeachmaximizedifferentiableparamslretas
step_sizesr   r   r    r!   c          	         t          |t                    r'|                                dk    rt          d          d|k    st          d|           d|d         cxk     rdcxk     r|d         k     s#n t          d|d          d|d                    t	          |||||||	          }	t                                          ||	           d S )
Nr	   zTensor lr must be 1-elementg        zInvalid learning rate: r         ?zInvalid eta values: z, )r#   r$   r%   r   r    r!   r   )
isinstancer   numel
ValueErrordictsuper__init__)selfr"   r#   r$   r%   r   r   r    r!   defaults	__class__s             a/root/.openclaw/workspace/chatterbox_venv_py311/lib/python3.11/site-packages/torch/optim/rprop.pyr-   zRprop.__init__   s     b&!! 	<bhhjjAoo:;;;byy;r;;<<<T!W,,,,s,,,,T!W,,,,HDGHHtAwHHIII!)!
 
 
 	*****    c                    t                                          |           | j        D ]}|                    dd            |                    dd           |                    dd           |                    dd           |d         D ]}| j                            |g           }t          |          dk    rt          j        |d                   sjt          |d                   }|d         r(t          j
        |t                      |j        	          n!t          j
        |t                      
          |d<   d S )Nr   r    Fr!   r   r"   r   stepdtypedevicer6   )r,   __setstate__param_groups
setdefaultstategetlentorch	is_tensorfloattensorr   r7   )r.   r<   grouppp_statestep_valr0   s         r1   r9   zRprop.__setstate__<   sO   U###& 	 	EY---Z///-u555\51118_ 
 
*..B//w<<1$$U_WV_-M-M$$WV_55H
 !.O$,=,?,?    #\(:K:M:MNNN FO	
	 	r2   c           	      |   d}|d         D ]}|j         |t          j        |          z  }|                    |           |j         }	|	j        rt          d          |                    |	           | j        |         }
t          |
          dk    r|d         r(t          j        dt                      |j
                  n!t          j        dt                                |
d	<   t          j        |t          j        
          |
d<   |j        j        r3t          j        |	t          |d         |d                             |
d<   nt          j        |	|d                   |
d<   |                    |
d                    |                    |
d                    |                    |
d	                    |S )NFr"   z'Rprop does not support sparse gradientsr   r    r5   r8   r4   memory_formatprevr#   	step_size)gradr?   
is_complexappend	is_sparseRuntimeErrorr<   r>   zerosr   r7   
zeros_likepreserve_formatr6   	full_likecomplex)r.   rC   r"   gradsprevsr%   state_stepshas_complexrD   rM   r<   s              r1   _init_groupzRprop._init_groupO   s   x  	.  	.Av~5+A...KMM!6D~ N"#LMMMLLJqME 5zzQ \*DEK*;*=*=ahOOOOR/@/B/BCCC f !& 0%BW X X Xf7% L */geDk5;??* *E+&& */uT{)K)KE+&LLv'''eK0111uV}----r2   c                    |                                   d}|5t          j                    5   |            }ddd           n# 1 swxY w Y   | j        D ]u}g }g }g }g }g }|d         \  }	}
|d         \  }}|d         }|d         }|                     ||||||          }t          ||||||||	|
|||d         |d         |           v|S )	zPerform a single optimization step.

        Args:
            closure (Callable, optional): A closure that reevaluates the model
                and returns the loss.
        Nr$   r%   r   r    r!   r   )	step_size_minstep_size_maxetaminusetaplusr   r    r!   r   rZ   ) _cuda_graph_capture_health_checkr?   enable_gradr:   r[   r   )r.   closurelossrC   r"   rW   rX   r%   rY   r_   r`   r]   r^   r   r    rZ   s                   r1   r4   z
Rprop.stepu   sj    	--///"$$ ! !wyy! ! ! ! ! ! ! ! ! ! ! ! ! ! ! & 	 	E#%F"$E"$E')J(*K %fHg+0+>(M=I&GZ(H**vueZ K ++!!$%56 .'    " s   AA
A)r   r   r   N)__name__
__module____qualname__r   r   rA   r   r   boolr   r-   r9   r[   r   r4   __classcell__)r0   s   @r1   r   r      s        $($.*4+ !"&$+ + ++ %- + E5L!	+
 %,'+ + $+ + + + + + + +<    &$ $ $L "/ / / "!/ / / / /r2   a
  Implements the resilient backpropagation algorithm.

    .. math::
       \begin{aligned}
            &\rule{110mm}{0.4pt}                                                                 \\
            &\textbf{input}      : \theta_0 \in \mathbf{R}^d \text{ (params)},f(\theta)
                \text{ (objective)},                                                             \\
            &\hspace{13mm}      \eta_{+/-} \text{ (etaplus, etaminus)}, \Gamma_{max/min}
                \text{ (step sizes)}                                                             \\
            &\textbf{initialize} :   g^0_{prev} \leftarrow 0,
                \: \eta_0 \leftarrow \text{lr (learning rate)}                                   \\
            &\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} \textbf{for} \text{  } i = 0, 1, \ldots, d-1 \: \mathbf{do}            \\
            &\hspace{10mm}  \textbf{if} \:   g^i_{prev} g^i_t  > 0                               \\
            &\hspace{15mm}  \eta^i_t \leftarrow \mathrm{min}(\eta^i_{t-1} \eta_{+},
                \Gamma_{max})                                                                    \\
            &\hspace{10mm}  \textbf{else if}  \:  g^i_{prev} g^i_t < 0                           \\
            &\hspace{15mm}  \eta^i_t \leftarrow \mathrm{max}(\eta^i_{t-1} \eta_{-},
                \Gamma_{min})                                                                    \\
            &\hspace{15mm}  g^i_t \leftarrow 0                                                   \\
            &\hspace{10mm}  \textbf{else}  \:                                                    \\
            &\hspace{15mm}  \eta^i_t \leftarrow \eta^i_{t-1}                                     \\
            &\hspace{5mm}\theta_t \leftarrow \theta_{t-1}- \eta_t \mathrm{sign}(g_t)             \\
            &\hspace{5mm}g_{prev} \leftarrow  g_t                                                \\
            &\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 the paper
    `A Direct Adaptive Method for Faster Backpropagation Learning: The RPROP Algorithm
    <http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.21.1417>`_.
    z
    Args:
        a{  
        lr (float, optional): learning rate (default: 1e-2)
        etas (Tuple[float, float], optional): pair of (etaminus, etaplus), that
            are multiplicative increase and decrease factors
            (default: (0.5, 1.2))
        step_sizes (Tuple[float, float], optional): a pair of minimal and
            maximal allowed step sizes (default: (1e-6, 50))
        z	
        z

    r"   rW   rX   r%   rY   r]   r^   r_   r`   r    r   r!   rZ   c                   t          |           D ];\  }}||         }|	s|n| }||         }||         }||         }t          j                                        sF|
rDt	                      }|j        j        |j        j        k    r|j        j        |v sJ d| d            |dz  }t          j        |          rPt          j        |          }t          j        |          }t          j        |          }t          j        |          }|r:|	                    |
                                                                          }n'|	                    |                                          }|
r|                    t          j        |                    d          ||                     |                    t          j        |                    d          ||                     |                    t          j        |                    d          d|                     nH|||                    d          <   |||                    d          <   d||                    d          <   |                    |                              ||           |
                    t          j                  }|
r=|                    t          j        |                    |          d|                     nd||                    |          <   |                    |                                |d           |                    |           =d S )NIIf capturable=True, params and state_steps must be on supported devices: .r	   r   rI   value)	enumerater?   compileris_compilingr   r7   typerN   view_as_realmulclonesigncopy_wheregtlteqmul_clamp_rT   addcmul_)r"   rW   rX   r%   rY   r]   r^   r_   r`   r    r   r!   rZ   iparamrM   rK   rL   r4   capturable_supported_devicesrx   s                        r1   _single_tensor_rpropr      s     f%% 1 15Qx#.tt$QxqM	1~ ~**,, 	{ 	{+L+N+N(!T[%555L%)EEEEz[wzzz FEF 		E"" 	6%d++D%d++D&u--E*955I 	)88DJJLL))..00DD88D>>&&((D 	!JJu{4771::w==>>>JJu{4771::x>>???JJu{4771::q$778888&D'D D 	t##M=AAA zz(=z>> 	(JJu{4778#4#4a>>????&'D""# 	tyy{{IR888

4c1 1r2   c          
      V   t          |           dk    rd S |r
J d            t          j                                        sG|
rEt	                      t          fdt          | |          D                       sJ d d            t          j        | ||||g          }|	                                D ]r\  \  }}}}}}t          t          t                   |          }t          t          t                   |          }t          t          t                   |          }t          t          t                   |          }t          t          t                   |          }t          j                                        s9|d         j        r,t          j        |t          j        dd          d	           nt          j        |d
           |rt!          ||||           t          j        ||          }|	rt          j        |           t          j        ||           |	rt          j        |           |}t          j        |           |
r|D ]}|                    t          j        |                    d          ||                     |                    t          j        |                    d          ||                     |                    t          j        |                    d          d
|                     nM|D ]J}|||                    d          <   |||                    d          <   d
||                    d          <   Kt          j        ||           |D ]}|                    ||           t9          |          }t;          t          |                    D ]P}||                             t          j        ||                             |          d||                              Q~d |D             }t          j        |||d           td S )Nr   z#_foreach ops don't support autogradc              3   n   K   | ]/\  }}|j         j        |j         j        k    o|j         j        v V  0d S re   )r7   rt   ).0rD   r4   r   s      r1   	<genexpr>z&_multi_tensor_rprop.<locals>.<genexpr>:  s]       
 
 4 HMT[-- >!==
 
 
 
 
 
r2   rl   rm   r'   cpu)r7   )alphar	   c                 6    g | ]}|                                 S rH   )rx   )r   rM   s     r1   
<listcomp>z'_multi_tensor_rprop.<locals>.<listcomp>  s     <<<ddiikk<<<r2   rn   ro   )r>   r?   rr   rs   r   allzipr   "_group_tensors_by_device_and_dtypevaluesr   r   r   is_cpu_foreach_add_rB   r   _foreach_mul_foreach_neg__foreach_copy__foreach_sign_ry   rz   r{   r|   r}   _foreach_mul_r   listrange_foreach_addcmul_)r"   rW   rX   r%   rY   r]   r^   r_   r`   r    r   r!   rZ   grouped_tensorsgrouped_params_grouped_grads_grouped_prevs_grouped_step_sizes_grouped_state_steps__grouped_paramsgrouped_gradsgrouped_prevsgrouped_step_sizesgrouped_state_stepssignsrx   rL   r   
grad_signsr   s                                 @r1   _multi_tensor_rpropr   "  s     6{{aDDDDD >&&(( wZ w'H'J'J$ 
 
 
 
 v{33
 
 
 
 
 	w 	w wWsvvv		w 	w 
  B	z;7 O ""$$J
 J
 		 	d6lO<<T&\>::T&\>::!$v,0CDD"4<1EFF ~**,, 	81DQ1G1N 	8#U\#e%D%D%DC      3Q777  	}>P   "=-@@ 	'&&&
 	]M::: 	/...%U### 		% = =

5;twwqzz7DAABBB

5;twwqzz8TBBCCC

5;twwqzz1d;;<<<<=
  % %#*TWWQZZ #+TWWQZZ #$TWWQZZ   	.666+ 	; 	;I]M:::: ]++s=))** 	 	A!""E!HKK111mA6FGG   
  =<m<<<
J(:"	
 	
 	
 	
 	
QJ
 J
r2   )single_tensor_fnFr   c
                   t           j                                        s(t          d |D                       st	          d          |t          | |d          \  }}|r-t           j                                        rt	          d          |r&t           j                                        st          }nt          } || |||||
|||||||	           dS )zpFunctional API that performs rprop algorithm computation.

    See :class:`~torch.optim.Rprop` for details.
    c              3   J   K   | ]}t          |t          j                  V  d S re   )r(   r?   r   )r   ts     r1   r   zrprop.<locals>.<genexpr>  s?       5 5()
1el##5 5 5 5 5 5r2   zPAPI has changed, `state_steps` argument must contain a list of singleton tensorsNF)	use_fusedz6torch.jit.script not supported with foreach optimizers)r]   r^   r_   r`   r   r    r!   rZ   )
r?   rr   rs   r   rQ   r   jitis_scriptingr   r   )r"   rW   rX   r%   rY   r   r   r    r!   rZ   r]   r^   r_   r`   r   funcs                   r1   r   r     s'   4 >&&(( 
 5 5-85 5 5 2 2 
 ^
 
 	
 1Ne
 
 

7  U59))++ USTTT $uy--// $"#D##%     r2   )NFFFF)__doc__typingr   r   r   r   r   r?   r   	optimizerr
   r   r   r   r   r   r   r   r   r   r   r   r   __all__r   rA   ri   r   r   r   rH   r2   r1   <module>r      s   9 8 5 5 5 5 5 5 5 5 5 5 5 5 5 5                                    " G
H H H H HI H H HX"F	  
  
  
  
  G1 lALA<A <A V	A
 fA A A A A A A A A A A AHk
Lk
<k
 <k
 V	k

 fk
 k
 k
 k
 k
 k
 k
 k
 k
 k
 k
 k
d  1EFFF # ; ;L;<; <; V	;
 f; d^; ; ; ; ; ; ;  !;" #; ; ; GF; ; ;r2   