[Tex/LaTex] How to draw an LSTM cell in Tikz


I am a newcomer to Tikz and have been trying to draw an recurrent neural network Long-Short Term Memory (LSTM) cell in Tikz, but have trouble correctly aligning the boxes I need inside the cell. The LSTM cell looks as follows

LSTM sketch

I have the following attempt, but clearly it's a far way from being done.

LSTM tikz

The code is

\usetikzlibrary{positioning, fit, arrows.meta, shapes}

    elementwiseoperation/.style={circle, draw, inner sep=0pt},
    elementwisefunction/.style={ellipse, draw, inner sep=1pt},
    ct/.style={circle, draw, minimum width=1cm, inner sep=1pt},
    gt/.style={rectangle, draw, minimum width=4mm, minimum height=3mm, inner sep=1pt},
    filter/.style={circle, draw, minimum width=8mm, inner sep=1pt, path picture={\draw[thick, rounded corners] (path picture bounding box.center)--++(65:2mm)--++(0:1mm);
    \draw[thick, rounded corners] (path picture bounding box.center)--++(245:2mm)--++(180:1mm);}},

    % Input cell
    \node[ct, label={[mylabel]Cell}] (ct1) {$c^{t-1}$};
    % Input hidden
    \node[ct, below=3cm of ct1.south, label={[mylabel]Hidden}] (ht1) {$h^{t-1}$};3
    % Input x
    \node[ct, below right=1cm and 1 cm of ht1, label={[mylabel]left:Input}] (xt1) {$x^{t}$};
    % Elementwise operations on cell
    \node[elementwiseoperation, right=1.5cm of ct1] (mul1) {$\times$};
    \node[elementwiseoperation, right=of mul1] (add1) {$+$};
    \coordinate[left of=mul1] (celllinesplit0);
    \coordinate[right of=add1] (celllinesplit1);
    \coordinate[right of=celllinesplit1] (celllinesplit2);
    \coordinate[above=of xt1, right=of ht1] (h and x join);
    % New cell 
    \node[elementwisefunction, below=of celllinesplit1] (tanh) {tanh};
    \node[elementwiseoperation, below of=add1] (mul2) {$\times$};
    \node[ct, right of=celllinesplit1, label={[mylabel]Cell}] (ct2) {$c^{t}$};
    \node[gt, below of=mul2] (cellbox) {tanh};
    \node[gt, left=2mm of cellbox] (inputbox) {$\sigma$};
    \node[gt, left=2mm of inputbox, below=of mul1] (forgetbox) {$\sigma$};
    \node[gt, right=2mm of cellbox] (outputbox) {$\sigma

    \draw[->] (ct1) to (mul1);
    \draw[->] (mul1) to (add1);
    \draw[->] (mul2) to (add1);
    \draw[->] (add1) to (ct2);
    \draw[->] (add1) to[out=0,in=90] (tanh);
    \draw[->] (forgetbox) to (mul1);

    \draw[-] (xt1) to (h and x join)[in=0];
    \draw[-] (ht1) to (h and x join)[in=0];

Thanks in advance for any attempt at this, it is much appreciated.

Best Answer

Just for fun, and to prove that the arrow paths with corners can be rounded. An option using absolute positioning, and labeled coordinates, with intersectiĆ³ns (A|-B), and displacements ++(a,b).

RESULT: enter image description here


% By J. Leon, Beerware licence is acceptable...
\usetikzlibrary{positioning, fit, arrows.meta, shapes}

% used to avoid putting the same thing several times...
% Command \empt{var1}{var2}
\newcommand{\empt}[2]{$#1^{\langle #2 \rangle}$}


    font=\sf \scriptsize,
    % Styles
    cell/.style={% For the main box
        rounded corners=5mm, 
        very thick,
    operator/.style={%For operators like +  and  x
        inner sep=-0.5pt,
        minimum height =.2cm,
    function/.style={%For functions
        inner sep=1pt
    ct/.style={% For external inputs and outputs
        line width = .75pt,
        minimum width=1cm,
        inner sep=1pt,
    gt/.style={% For internal inputs
        minimum width=4mm,
        minimum height=3mm,
        inner sep=1pt
    mylabel/.style={% something new that I have learned
    ArrowC1/.style={% Arrows with rounded corners
        rounded corners=.25cm,
    ArrowC2/.style={% Arrows with big rounded corners
        rounded corners=.5cm,

%Start drawing the thing...    
    % Draw the cell: 
    \node [cell, minimum height =4cm, minimum width=6cm] at (0,0){} ;

    % Draw inputs named ibox#
    \node [gt] (ibox1) at (-2,-0.75) {$\sigma$};
    \node [gt] (ibox2) at (-1.5,-0.75) {$\sigma$};
    \node [gt, minimum width=1cm] (ibox3) at (-0.5,-0.75) {Tanh};
    \node [gt] (ibox4) at (0.5,-0.75) {$\sigma$};

   % Draw opƩrators   named mux# , add# and func#
    \node [operator] (mux1) at (-2,1.5) {$\times$};
    \node [operator] (add1) at (-0.5,1.5) {+};
    \node [operator] (mux2) at (-0.5,0) {$\times$};
    \node [operator] (mux3) at (1.5,0) {$\times$};
    \node [function] (func1) at (1.5,0.75) {Tanh};

    % Draw External inputs? named as basis c,h,x
    \node[ct, label={[mylabel]Cell}] (c) at (-4,1.5) {\empt{c}{t-1}};
    \node[ct, label={[mylabel]Hidden}] (h) at (-4,-1.5) {\empt{h}{t-1}};
    \node[ct, label={[mylabel]left:Input}] (x) at (-2.5,-3) {\empt{x}{t}};

    % Draw External outputs? named as basis c2,h2,x2
    \node[ct, label={[mylabel]Label1}] (c2) at (4,1.5) {\empt{c}{t}};
    \node[ct, label={[mylabel]Label2}] (h2) at (4,-1.5) {\empt{h}{t}};
    \node[ct, label={[mylabel]left:Label3}] (x2) at (2.5,3) {\empt{h}{t}};

% Start connecting all.
    %Intersections and displacements are used. 
    % Drawing arrows    
    \draw [ArrowC1] (c) -- (mux1) -- (add1) -- (c2);

    % Inputs
    \draw [ArrowC2] (h) -| (ibox4);
    \draw [ArrowC1] (h -| ibox1)++(-0.5,0) -| (ibox1); 
    \draw [ArrowC1] (h -| ibox2)++(-0.5,0) -| (ibox2);
    \draw [ArrowC1] (h -| ibox3)++(-0.5,0) -| (ibox3);
    \draw [ArrowC1] (x) -- (x |- h)-| (ibox3);

    % Internal
    \draw [->, ArrowC2] (ibox1) -- (mux1);
    \draw [->, ArrowC2] (ibox2) |- (mux2);
    \draw [->, ArrowC2] (ibox3) -- (mux2);
    \draw [->, ArrowC2] (ibox4) |- (mux3);
    \draw [->, ArrowC2] (mux2) -- (add1);
    \draw [->, ArrowC1] (add1 -| func1)++(-0.5,0) -| (func1);
    \draw [->, ArrowC2] (func1) -- (mux3);

    \draw [-, ArrowC2] (mux3) |- (h2);
    \draw (c2 -| x2) ++(0,-0.1) coordinate (i1);
    \draw [-, ArrowC2] (h2 -| x2)++(-0.5,0) -| (i1);
    \draw [-, ArrowC2] (i1)++(0,0.2) -- (x2);