Dataset Viewer
Auto-converted to Parquet Duplicate
image
imagewidth (px)
12
5.6k
text
stringlengths
1
10.9k
text_raw
stringlengths
1
10.9k
source
stringclasses
2 values
license
stringclasses
2 values
kind
stringclasses
3 values
orig_split
stringclasses
9 values
width
int32
12
5.6k
height
int32
16
2.43k
n \to \infty
n \to \infty
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
32
\begin{array} { r l } { \langle u | u | ^ { 2 } , b _ { j , 1 } \rangle } & { = \sum _ { k , l , m } \frac { A _ { k } A _ { l } A _ { m } } { L _ { k } L _ { l } L _ { m } } \left\langle e ^ { i \Gamma _ { k } + i \Gamma _ { l } - i \Gamma _ { m } } e ^ { - \frac { | y _ { k } | ^ { 2 } + | y _ { l } | ^ { 2 } + | y _...
\begin{array} { r l } { \langle u | u | ^ { 2 } , b _ { j , 1 } \rangle } & { = \sum _ { k , l , m } \frac { A _ { k } A _ { l } A _ { m } } { L _ { k } L _ { l } L _ { m } } \left\langle e ^ { i \Gamma _ { k } + i \Gamma _ { l } - i \Gamma _ { m } } e ^ { - \frac { | y _ { k } | ^ { 2 } + | y _ { l } | ^ { 2 } + | y _...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,152
128
\begin{array} { r l r } { V ( x , t ) } & { = } & { \frac { 1 } { | \sqrt { 2 \lambda } \alpha _ { 0 } | ^ { 2 q } } \sum _ { m = 0 } ^ { q } ( - 1 ) ^ { m + q } \frac { ( 2 ^ { m } q ! ) ^ { 2 } } { ( 2 m ) ! ( q - m ) ! } \lambda ^ { q - m } x ^ { 2 m } } \\ & { } & { - \frac { ( - i ) ^ { q } } { | \sqrt { 2 \lambda...
\begin{array} { r l r } { V ( x , t ) } & { = } & { \frac { 1 } { | \sqrt { 2 \lambda } \alpha _ { 0 } | ^ { 2 q } } \sum _ { m = 0 } ^ { q } ( - 1 ) ^ { m + q } \frac { ( 2 ^ { m } q ! ) ^ { 2 } } { ( 2 m ) ! ( q - m ) ! } \lambda ^ { q - m } x ^ { 2 m } } \\ & { } & { - \frac { ( - i ) ^ { q } } { | \sqrt { 2 \lambda...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,056
224
S = \textsf { S t a b s } ( \Pi ) = \{ P _ { i } \}
S = \textsf { S t a b s } ( \Pi ) = \{ P _ { i } \}
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
160
32
s _ { \mathrm { i m } } ^ { 2 } \left( a _ { 1 1 1 0 } + a _ { 2 0 0 1 } - s _ { \mathrm { i m } } a _ { 2 1 0 0 } \right)
s _ { \mathrm { i m } } ^ { 2 } \left( a _ { 1 1 1 0 } + a _ { 2 0 0 1 } - s _ { \mathrm { i m } } a _ { 2 1 0 0 } \right)
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
480
64
R
R
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
_ { D D }
_ { D D }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
c = ( m ^ { 2 } + n ^ { 2 } ) ^ { 2 } ,
c = ( m ^ { 2 } + n ^ { 2 } ) ^ { 2 } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
224
64
\begin{array} { r l r } & { } & { \left. \frac { \partial } { \partial b } \left\{ I ( u , v ) \right\} \right| _ { b = 0 } } \\ & { = } & { - j \left[ \int _ { 0 } ^ { 2 \pi } \int _ { 0 } ^ { 1 } e ^ { j [ a Z _ { 5 } ( r , \theta ) - 2 \pi \rho r c o s ( \phi - \theta ) } r d r d \theta \right. } \\ & { } & { \times...
\begin{array} { r l r } & { } & { \left. \frac { \partial } { \partial b } \left\{ I ( u , v ) \right\} \right| _ { b = 0 } } \\ & { = } & { - j \left[ \int _ { 0 } ^ { 2 \pi } \int _ { 0 } ^ { 1 } e ^ { j [ a Z _ { 5 } ( r , \theta ) - 2 \pi \rho r c o s ( \phi - \theta ) } r d r d \theta \right. } \\ & { } & { \times...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
832
320
p
p
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
N \to \infty
N \to \infty
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
160
32
a _ { 1 }
a _ { 1 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\ddot { \mathrm { ~ o ~ } }
\ddot { \mathrm { ~ o ~ } }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
j
j
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
( x , y ) \in \left[ - 1 0 , 1 0 \right]
( x , y ) \in \left[ - 1 0 , 1 0 \right]
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
288
64
\begin{array} { r l } { ( \pi _ { C | D } - \pi _ { D | D } ) q _ { D | D } = } & { \left[ - c + w _ { I } b + \frac { 1 - w _ { I } } { k } ( k - 1 ) q _ { C | C } b - \frac { 1 - w _ { I } } { k } ( k - 1 ) q _ { C | D } b \right] \left( 1 - \frac { k - 2 } { k - 1 } p _ { C } \right) } \\ { = } & { \left( - c + w _ ...
\begin{array} { r l } { ( \pi _ { C | D } - \pi _ { D | D } ) q _ { D | D } = } & { \left[ - c + w _ { I } b + \frac { 1 - w _ { I } } { k } ( k - 1 ) q _ { C | C } b - \frac { 1 - w _ { I } } { k } ( k - 1 ) q _ { C | D } b \right] \left( 1 - \frac { k - 2 } { k - 1 } p _ { C } \right) } \\ { = } & { \left( - c + w _ ...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,376
192
x
x
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
{ \cal Z } _ { B } = T r \; e ^ { - \beta \varepsilon _ { \kappa } ( \overline { { { \Phi } } } _ { 1 , \kappa } \Phi _ { 1 , \kappa } + \overline { { { \Phi } } } _ { 2 , \kappa } \Phi _ { 2 , \kappa } ) } e ^ { - \beta \mu ( N _ { 1 , \kappa } + N _ { 2 , \kappa } ) } ,
{ \cal Z } _ { B } = T r \; e ^ { - \beta \varepsilon _ { \kappa } ( \overline { { { \Phi } } } _ { 1 , \kappa } \Phi _ { 1 , \kappa } + \overline { { { \Phi } } } _ { 2 , \kappa } \Phi _ { 2 , \kappa } ) } e ^ { - \beta \mu ( N _ { 1 , \kappa } + N _ { 2 , \kappa } ) } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
544
64
d _ { n }
d _ { n }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
1 5 . 5 H \times 4 . 5 H \times 4 H = 6 . 2 \times 1 . 8 \times 1 . 6 m ^ { 3 }
1 5 . 5 H \times 4 . 5 H \times 4 H = 6 . 2 \times 1 . 8 \times 1 . 6 m ^ { 3 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
576
32
\begin{array} { r } { 4 \pi + \frac { \gamma \varepsilon ^ { 3 } } { 2 } ( 1 + \varepsilon + \varepsilon ^ { 2 } ) - \left( \frac { 2 \pi { h ^ { \ast } } ^ { 4 } } { \varepsilon ^ { 2 } } - \frac { \gamma { h ^ { \ast } } ^ { 2 } \varepsilon ^ { 2 } } { 2 } \right) \log { ( h ^ { \ast } ) } + \ensuremath { \operatorna...
\begin{array} { r } { 4 \pi + \frac { \gamma \varepsilon ^ { 3 } } { 2 } ( 1 + \varepsilon + \varepsilon ^ { 2 } ) - \left( \frac { 2 \pi { h ^ { \ast } } ^ { 4 } } { \varepsilon ^ { 2 } } - \frac { \gamma { h ^ { \ast } } ^ { 2 } \varepsilon ^ { 2 } } { 2 } \right) \log { ( h ^ { \ast } ) } + \ensuremath { \operatorna...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
992
160
F _ { 3 }
F _ { 3 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
r < R _ { \mathrm { p h o t } }
r < R _ { \mathrm { p h o t } }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
32
G _ { F }
G _ { F }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
9 6 \%
9 6 \%
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
32
x _ { 0 } = x ^ { u }
x _ { 0 } = x ^ { u }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
160
64
\begin{array} { r l } { \phi _ { i } ( \lambda _ { i } ) } & { = \mathrm { v e c } \left( { \uppercase { \mathbf { P } } } _ { i } \right) ^ { H } \mathrm { v e c } \left( { \uppercase { \mathbf { P } } } _ { i } \right) } \\ & { = { \lowercase { \mathbf { a } } } _ { i } ^ { H } { \uppercase { \mathbf { D } } } _ { i ...
\begin{array} { r l } { \phi _ { i } ( \lambda _ { i } ) } & { = \mathrm { v e c } \left( { \uppercase { \mathbf { P } } } _ { i } \right) ^ { H } \mathrm { v e c } \left( { \uppercase { \mathbf { P } } } _ { i } \right) } \\ & { = { \lowercase { \mathbf { a } } } _ { i } ^ { H } { \uppercase { \mathbf { D } } } _ { i ...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,760
192
V ( t )
V ( t )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\sigma ( r ) = \frac { 2 E } { 1 + \nu } \frac { h \kappa } { 4 \pi r ^ { 3 } }
\sigma ( r ) = \frac { 2 E } { 1 + \nu } \frac { h \kappa } { 4 \pi r ^ { 3 } }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
352
96
\begin{array} { r l r } { E _ { 1 2 } ( a , c , W \rightarrow 0 ) } & { { } = } & { \left\{ \begin{array} { l c l } { - \frac { 1 } { 2 } \cos 2 \theta } & { , } & { d = 0 } \\ { \frac { \pi } { 4 } \sin 2 \theta \cos 2 \theta - \cos 2 \theta + \ln [ | \tan \theta | ^ { \sin ^ { 2 } 2 \theta / 2 } ] } & { , } & { d = 2...
\begin{array} { r l r } { E _ { 1 2 } ( a , c , W \rightarrow 0 ) } & { { } = } & { \left\{ \begin{array} { l c l } { - \frac { 1 } { 2 } \cos 2 \theta } & { , } & { d = 0 } \\ { \frac { \pi } { 4 } \sin 2 \theta \cos 2 \theta - \cos 2 \theta + \ln [ | \tan \theta | ^ { \sin ^ { 2 } 2 \theta / 2 } ] } & { , } & { d = 2...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
896
192
( { \hat { T } } ( \mathbf { x } ) ) ^ { - 1 } \mathbf { \hat { r } } { \hat { T } } ( \mathbf { x } ) = \mathbf { \hat { r } } + \mathbf { x } { \hat { \mathbb { I } } }
( { \hat { T } } ( \mathbf { x } ) ) ^ { - 1 } \mathbf { \hat { r } } { \hat { T } } ( \mathbf { x } ) = \mathbf { \hat { r } } + \mathbf { x } { \hat { \mathbb { I } } }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
384
64
\rho
\rho
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\overrightarrow { \xi }
\overrightarrow { \xi }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
p _ { L }
p _ { L }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
{ \frac { \partial a _ { k } ^ { s } } { \partial t } } = \frac { i \epsilon } { 8 k _ { \perp } } \int _ { \mathbb { R } ^ { 6 } } \sum _ { s _ { p } s _ { q } } \left[ s _ { p } s _ { q } k _ { \perp } ^ { 2 } + \frac { k _ { \parallel } } { p q } \left( p _ { \parallel } q ^ { 2 } + q _ { \parallel } p ^ { 2 } \righ...
{ \frac { \partial a _ { k } ^ { s } } { \partial t } } = \frac { i \epsilon } { 8 k _ { \perp } } \int _ { \mathbb { R } ^ { 6 } } \sum _ { s _ { p } s _ { q } } \left[ s _ { p } s _ { q } k _ { \perp } ^ { 2 } + \frac { k _ { \parallel } } { p q } \left( p _ { \parallel } q ^ { 2 } + q _ { \parallel } p ^ { 2 } \righ...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
480
96
\phi ( t )
\phi ( t )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
64
\frac { d \rho } { d t } = - [ \overline { { F } } , \rho ] \neq 0 .
\frac { d \rho } { d t } = - [ \overline { { F } } , \rho ] \neq 0 .
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
32
k ^ { 2 } = k _ { x } ^ { 2 } + k _ { y } ^ { 2 }
k ^ { 2 } = k _ { x } ^ { 2 } + k _ { y } ^ { 2 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
32
\begin{array} { r } { \xi \cos \theta _ { \mathrm { L } } + \Delta \sin \theta _ { \mathrm { L } } = \sqrt { \xi ^ { 2 } + \Delta ^ { 2 } } . } \end{array}
\begin{array} { r } { \xi \cos \theta _ { \mathrm { L } } + \Delta \sin \theta _ { \mathrm { L } } = \sqrt { \xi ^ { 2 } + \Delta ^ { 2 } } . } \end{array}
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
544
64
C _ { \mathrm { G } } \approx 1 6 \: \mathrm { \ m u F / c m ^ { 2 } }
C _ { \mathrm { G } } \approx 1 6 \: \mathrm { \ m u F / c m ^ { 2 } }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
192
32
0
0
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
{ \begin{array} { r l } { { F ^ { \alpha \beta } } _ { ; \beta } } & { = 0 } \\ { F _ { [ \alpha \beta ; \gamma ] } } & { = { \frac { 1 } { 3 } } \left( F _ { \alpha \beta ; \gamma } + F _ { \beta \gamma ; \alpha } + F _ { \gamma \alpha ; \beta } \right) = { \frac { 1 } { 3 } } \left( F _ { \alpha \beta , \gamma } + F ...
{ \begin{array} { r l } { { F ^ { \alpha \beta } } _ { ; \beta } } & { = 0 } \\ { F _ { [ \alpha \beta ; \gamma ] } } & { = { \frac { 1 } { 3 } } \left( F _ { \alpha \beta ; \gamma } + F _ { \beta \gamma ; \alpha } + F _ { \gamma \alpha ; \beta } \right) = { \frac { 1 } { 3 } } \left( F _ { \alpha \beta , \gamma } + F ...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,280
160
\leq
\leq
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
F _ { r { \tilde { \theta } } } ^ { 1 } = - \, \frac { 1 } { 2 \lambda } \, { \dot { a } } ~ ~ , ~ ~ F _ { r { \tilde { \varphi } } } ^ { 2 } = \frac { 1 } { 2 \lambda } \, { \sin { \widetilde { \theta } } \, { \dot { a } } } ~ ~ , ~ ~ F _ { { \tilde { \theta } } { \tilde { \varphi } } } ^ { 3 } = \frac { 1 } { 2 \lamb...
F _ { r { \tilde { \theta } } } ^ { 1 } = - \, \frac { 1 } { 2 \lambda } \, { \dot { a } } ~ ~ , ~ ~ F _ { r { \tilde { \varphi } } } ^ { 2 } = \frac { 1 } { 2 \lambda } \, { \sin { \widetilde { \theta } } \, { \dot { a } } } ~ ~ , ~ ~ F _ { { \tilde { \theta } } { \tilde { \varphi } } } ^ { 3 } = \frac { 1 } { 2 \lamb...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
448
64
i
i
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
G _ { x x } ( 0 , 0 ) = g ( \rho R ^ { 2 } , \infty ) / N
G _ { x x } ( 0 , 0 ) = g ( \rho R ^ { 2 } , \infty ) / N
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
352
32
( \mathfrak { n } ^ { + } \otimes \mathbb C [ t ] ) v _ { \pmb { \xi } } = 0 , \quad \, ( h _ { \alpha _ { i } } \otimes t ^ { s } ) v _ { \pmb { \xi } } = \lambda ( h _ { \alpha _ { i } } ) \delta _ { s , 0 } v _ { \pmb { \xi } } , \quad \, ( x _ { \alpha _ { i } } ^ { - } \otimes 1 ) ^ { \lambda ( h _ { \alpha _ { i ...
( \mathfrak { n } ^ { + } \otimes \mathbb C [ t ] ) v _ { \pmb { \xi } } = 0 , \quad \, ( h _ { \alpha _ { i } } \otimes t ^ { s } ) v _ { \pmb { \xi } } = \lambda ( h _ { \alpha _ { i } } ) \delta _ { s , 0 } v _ { \pmb { \xi } } , \quad \, ( x _ { \alpha _ { i } } ^ { - } \otimes 1 ) ^ { \lambda ( h _ { \alpha _ { i ...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,088
64
^ *
^ *
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\begin{array} { r l r } { \sum _ { i } P _ { i , \nu } d Q _ { i , \nu } } & { { } = } & { \frac { 1 } { 2 } \left[ ( P _ { k } - P _ { - k } ) d ( Q _ { k } - Q _ { - k } ) - ( Q _ { k } + Q _ { - k } ) d ( P _ { k } + P _ { - k } ) \right] = } \end{array}
\begin{array} { r l r } { \sum _ { i } P _ { i , \nu } d Q _ { i , \nu } } & { { } = } & { \frac { 1 } { 2 } \left[ ( P _ { k } - P _ { - k } ) d ( Q _ { k } - Q _ { - k } ) - ( Q _ { k } + Q _ { - k } ) d ( P _ { k } + P _ { - k } ) \right] = } \end{array}
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,472
64
I - V
I - V
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
a _ { P } a _ { Q } ^ { \dagger } = S _ { P Q } - a _ { Q } ^ { \dagger } a _ { P } ,
a _ { P } a _ { Q } ^ { \dagger } = S _ { P Q } - a _ { Q } ^ { \dagger } a _ { P } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
160
32
\overline { { \v x } } _ { k } ^ { f } = \frac { 1 } { N _ { e } } \sum _ { n = 1 } ^ { N _ { e } } \v x _ { k } ^ { f , ( n ) }
\overline { { \v x } } _ { k } ^ { f } = \frac { 1 } { N _ { e } } \sum _ { n = 1 } ^ { N _ { e } } \v x _ { k } ^ { f , ( n ) }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
480
160
\mathrm { m w n v } = \int _ { - 1 } ^ { 1 } u _ { 2 } ( 0 , x _ { 2 } ) d x _ { 2 }
\mathrm { m w n v } = \int _ { - 1 } ^ { 1 } u _ { 2 } ( 0 , x _ { 2 } ) d x _ { 2 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
192
64
\mu \mathrm { m }
\mu \mathrm { m }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
a _ { \mu } ^ { \mathrm { S U S Y } } = { \frac { \tan \beta } { 4 8 \pi } } { \frac { m _ { \mu } ^ { 2 } } { M _ { \mathrm { S U S Y } } ^ { 2 } } } ( 5 \alpha _ { 2 } + \alpha _ { 1 } )
a _ { \mu } ^ { \mathrm { S U S Y } } = { \frac { \tan \beta } { 4 8 \pi } } { \frac { m _ { \mu } ^ { 2 } } { M _ { \mathrm { S U S Y } } ^ { 2 } } } ( 5 \alpha _ { 2 } + \alpha _ { 1 } )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
416
96
\delta
\delta
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\operatorname { a r c h a v e r c o s } ( y ) = 2 \operatorname { a r c c o s } \left( { \sqrt { y } } \right) = \operatorname { a r c c o s } \left( 2 y - 1 \right)
\operatorname { a r c h a v e r c o s } ( y ) = 2 \operatorname { a r c c o s } \left( { \sqrt { y } } \right) = \operatorname { a r c c o s } \left( 2 y - 1 \right)
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
864
64
x = { \frac { - 2 t ^ { 2 } - 1 } { - t ^ { 2 } - 1 } } \qquad \ d x = { \frac { 2 t } { ( - t ^ { 2 } - 1 ) ^ { 2 } } } \, \ d t ,
x = { \frac { - 2 t ^ { 2 } - 1 } { - t ^ { 2 } - 1 } } \qquad \ d x = { \frac { 2 t } { ( - t ^ { 2 } - 1 ) ^ { 2 } } } \, \ d t ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
576
96
\leftarrow c _ { n - 1 } \cdot
\leftarrow c _ { n - 1 } \cdot
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
32
F = \frac { v _ { x } ^ { 2 } + v _ { y } ^ { 2 } } { 2 g } .
F = \frac { v _ { x } ^ { 2 } + v _ { y } ^ { 2 } } { 2 g } .
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
\varphi _ { 1 }
\varphi _ { 1 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
64
\pm 2 \%
\pm 2 \%
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
R _ { 1 }
R _ { 1 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
64
64
\Phi = { \left\{ \begin{array} { l l } { \tan { \frac { \pi \alpha } { 2 } } } & { { \mathrm { i f ~ } } \alpha \neq 1 } \\ { - { \frac { 2 } { \pi } } \log | c \, t | } & { { \mathrm { i f ~ } } \alpha = 1 } \end{array} \right. }
\Phi = { \left\{ \begin{array} { l l } { \tan { \frac { \pi \alpha } { 2 } } } & { { \mathrm { i f ~ } } \alpha \neq 1 } \\ { - { \frac { 2 } { \pi } } \log | c \, t | } & { { \mathrm { i f ~ } } \alpha = 1 } \end{array} \right. }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
576
128
s \gg \lambda
s \gg \lambda
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
32
\sin ( \theta ) = - X _ { 3 }
\sin ( \theta ) = - X _ { 3 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
224
64
\epsilon _ { \mu \nu } p ^ { \mu } \Gamma ^ { \nu \lambda ( \beta ) } ( p ^ { 0 } , p ^ { 1 } ) = 0 = \epsilon _ { \mu \lambda } p ^ { \mu } \Gamma ^ { \nu \lambda ( \beta ) } ( p ^ { 0 } , p ^ { 1 } )
\epsilon _ { \mu \nu } p ^ { \mu } \Gamma ^ { \nu \lambda ( \beta ) } ( p ^ { 0 } , p ^ { 1 } ) = 0 = \epsilon _ { \mu \lambda } p ^ { \mu } \Gamma ^ { \nu \lambda ( \beta ) } ( p ^ { 0 } , p ^ { 1 } )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
864
64
g ( { \mathsf { d } } ) = 0
g ( { \mathsf { d } } ) = 0
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
32
\frac { \partial \langle \hat { \mathcal { L } } \rangle } { \partial \mu } = ( \langle \hat { A } \rangle - a ) \equiv 0 ,
\frac { \partial \langle \hat { \mathcal { L } } \rangle } { \partial \mu } = ( \langle \hat { A } \rangle - a ) \equiv 0 ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
320
96
\sigma ^ { \perp } = 6 1 . 2 \mathrm { ~ m ~ }
\sigma ^ { \perp } = 6 1 . 2 \mathrm { ~ m ~ }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
T ^ { 2 }
T ^ { 2 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
+ { \frac { 1 } { 2 m ^ { 2 } } } \Pi _ { \psi } ^ { 2 } + A _ { 0 } \Pi _ { \psi } + { \frac { m ^ { 2 } } { 2 } } ( ( \partial _ { i } \psi ) ^ { 2 } + 2 A _ { i } \partial _ { i } \psi ) .
+ { \frac { 1 } { 2 m ^ { 2 } } } \Pi _ { \psi } ^ { 2 } + A _ { 0 } \Pi _ { \psi } + { \frac { m ^ { 2 } } { 2 } } ( ( \partial _ { i } \psi ) ^ { 2 } + 2 A _ { i } \partial _ { i } \psi ) .
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
256
32
C _ { 1 } = a ( I - P ^ { T } ) ^ { \dag }
C _ { 1 } = a ( I - P ^ { T } ) ^ { \dag }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
32
H ^ { 2 } ( \mathrm { ~ \boldmath ~ q ~ } ^ { ( 1 ) } , \mathrm { ~ \boldmath ~ q ~ } ^ { ( 2 ) } ) = \frac { 1 } { 2 } \sum _ { j } \left( \sqrt { q _ { j } ^ { ( 1 ) } } - \sqrt { q _ { j } ^ { ( 2 ) } } \right) ^ { 2 } ,
H ^ { 2 } ( \mathrm { ~ \boldmath ~ q ~ } ^ { ( 1 ) } , \mathrm { ~ \boldmath ~ q ~ } ^ { ( 2 ) } ) = \frac { 1 } { 2 } \sum _ { j } \left( \sqrt { q _ { j } ^ { ( 1 ) } } - \sqrt { q _ { j } ^ { ( 2 ) } } \right) ^ { 2 } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
700
104
2 n _ { \mathrm { i n k } } h \cos { \theta _ { i } } = m \lambda ,
2 n _ { \mathrm { i n k } } h \cos { \theta _ { i } } = m \lambda ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
192
32
>
>
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
{ \mathcal { K } } ( \hat { M } _ { f } , \hat { M } _ { b } ) = \mathrm { c o n s t } \; \int { \mathcal { D } } { \mathcal { F } } { \mathcal { D } } { \mathcal { B } } \exp \left\{ - { \mathcal { W _ { F } } } - { \mathcal { W _ { B } } } - { \mathcal { W _ { F B } } } \right\}
{ \mathcal { K } } ( \hat { M } _ { f } , \hat { M } _ { b } ) = \mathrm { c o n s t } \; \int { \mathcal { D } } { \mathcal { F } } { \mathcal { D } } { \mathcal { B } } \exp \left\{ - { \mathcal { W _ { F } } } - { \mathcal { W _ { B } } } - { \mathcal { W _ { F B } } } \right\}
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
928
96
^ 1
^ 1
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\mathrm { N u } ^ { * , ( d r o p ) } < - 2
\mathrm { N u } ^ { * , ( d r o p ) } < - 2
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
288
64
y _ { 0 } = | H ( \mathrm { j } \omega ) |
y _ { 0 } = | H ( \mathrm { j } \omega ) |
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
\Gamma ( p ) \rightarrow \Gamma ( p ^ { \prime } ) = \Lambda \Gamma ( p ) \Lambda ^ { - 1 }
\Gamma ( p ) \rightarrow \Gamma ( p ^ { \prime } ) = \Lambda \Gamma ( p ) \Lambda ^ { - 1 }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
416
64
\forall ^ { p } L : = \left\{ x \in \{ 0 , 1 \} ^ { * } \ \left| \ \left( \forall w \in \{ 0 , 1 \} ^ { \leq p ( | x | ) } \right) \langle x , w \rangle \in L \right. \right\}
\forall ^ { p } L : = \left\{ x \in \{ 0 , 1 \} ^ { * } \ \left| \ \left( \forall w \in \{ 0 , 1 \} ^ { \leq p ( | x | ) } \right) \langle x , w \rangle \in L \right. \right\}
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
1,184
96
s
s
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
v
v
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
B W = 1 5
B W = 1 5
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
\tilde { \Omega }
\tilde { \Omega }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
64
E _ { c }
E _ { c }
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
\frac { d n _ { i } } { d t } = \frac { \partial n _ { i } } { \partial t } + \triangledown \cdot ( D \triangledown n _ { i } ) = \frac { \delta n _ { i } } { \delta t } ,
\frac { d n _ { i } } { d t } = \frac { \partial n _ { i } } { \partial t } + \triangledown \cdot ( D \triangledown n _ { i } ) = \frac { \delta n _ { i } } { \delta t } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
224
64
\begin{array} { r l } { h _ { 0 0 } ^ { \mathrm { P D } } \big | _ { \vec { x } = ( x , 0 , 0 ) } } & { = \frac { 1 - \cos ( x \omega _ { g } c _ { \vartheta } ) } { \omega _ { g } ^ { 2 } c _ { \vartheta } ^ { 2 } } \, ( \partial _ { 0 } ) ^ { 2 } h _ { 1 1 } ^ { \mathrm { T T } } \big | _ { \vec { x } = ( 0 , 0 , 0 )...
\begin{array} { r l } { h _ { 0 0 } ^ { \mathrm { P D } } \big | _ { \vec { x } = ( x , 0 , 0 ) } } & { = \frac { 1 - \cos ( x \omega _ { g } c _ { \vartheta } ) } { \omega _ { g } ^ { 2 } c _ { \vartheta } ^ { 2 } } \, ( \partial _ { 0 } ) ^ { 2 } h _ { 1 1 } ^ { \mathrm { T T } } \big | _ { \vec { x } = ( 0 , 0 , 0 )...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
320
64
\phi = 0
\phi = 0
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
( l + 1 ) ( l + 2 ) \cdots ( l + d - 1 )
( l + 1 ) ( l + 2 ) \cdots ( l + d - 1 )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
544
64
\neg ( A \land B )
\neg ( A \land B )
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
160
32
\lambda = L / H
\lambda = L / H
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
{ \mathcal { M } } = { \frac { 1 } { \sqrt { Z } } } { \Big ( } \operatorname* { l i m } _ { x _ { 1 } ^ { 0 } \rightarrow - \infty } - \operatorname* { l i m } _ { x _ { 1 } ^ { 0 } \rightarrow + \infty } { \Big ) } \int \! \mathrm { d } ^ { 3 } x _ { 1 } \; \mathrm { e } ^ { i p _ { 1 } \cdot x _ { 1 } } \langle \bet...
{ \mathcal { M } } = { \frac { 1 } { \sqrt { Z } } } { \Big ( } \operatorname* { l i m } _ { x _ { 1 } ^ { 0 } \rightarrow - \infty } - \operatorname* { l i m } _ { x _ { 1 } ^ { 0 } \rightarrow + \infty } { \Big ) } \int \! \mathrm { d } ^ { 3 } x _ { 1 } \; \mathrm { e } ^ { i p _ { 1 } \cdot x _ { 1 } } \langle \bet...
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
768
64
\mathcal { H } = - J \sum _ { \langle i j \rangle } \pmb { \sigma } _ { i } \pmb { \sigma } _ { j } ,
\mathcal { H } = - J \sum _ { \langle i j \rangle } \pmb { \sigma } _ { i } \pmb { \sigma } _ { j } ,
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
128
64
n _ { g } = v _ { g } / c
n _ { g } = v _ { g } / c
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
\hat { D } ^ { 2 } = D ^ { 2 } + i \gamma ^ { 5 } ( \partial _ { 5 } A \! \! \! / ) + i \{ \varphi , \partial _ { 5 } \} - \partial _ { 5 } { } ^ { 2 } .
\hat { D } ^ { 2 } = D ^ { 2 } + i \gamma ^ { 5 } ( \partial _ { 5 } A \! \! \! / ) + i \{ \varphi , \partial _ { 5 } \} - \partial _ { 5 } { } ^ { 2 } .
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
736
64
U
U
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
32
32
| \mathscr { W } | = - 2
| \mathscr { W } | = - 2
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
f = 1 , 2
f = 1 , 2
unimer-1m
apache-2.0 (card); underlying sources vary
mixed
train
96
32
End of preview. Expand in Data Studio

LaTeX-OCR Unified

A cleaned, deduplicated and test-decontaminated compilation of public image-to-LaTeX datasets, with one label convention and per-row provenance. About 1.66M formula images in total.

This is a compilation, not new source data. The images and labels belong to their original sources and are redistributed under those sources' terms (see Sources and licenses). What this repo adds is the curation: normalization, deduplication, decontamination and provenance columns.

Configs

Config Rows (train / validation) Sources License terms
permissive (default) 1,106,360 / 5,572 UniMER-1M, linxy/LaTeX_OCR Apache-2.0 per source cards (see caveats)
oleehyo-openrail 547,683 / 2,729 OleehyO/latex-formulas (cleaned) OpenRAIL: use restrictions apply, check before use

The two are kept apart on purpose so the OpenRAIL data never gets mixed in unnoticed.

from datasets import load_dataset
ds = load_dataset("sandysanta/latex-ocr-unified", "permissive", split="train", streaming=True)
row = next(iter(ds))   # row["image"] is a PIL image, row["text"] its LaTeX

Composition of permissive

source kind Rows
unimer-1m mixed (printed, screen-capture, handwritten; not labeled per sample) 940,684
linxy/LaTeX_OCR printed 84,119
linxy/LaTeX_OCR handwritten (human and synthetic handwriting configs) 87,129

Columns

Column Meaning
image PNG/JPEG bytes (decoded as an image by datasets)
text LaTeX label in UniMER's space-tokenized convention (the training target)
text_raw label exactly as in the source (identical to text except for oleehyo-openrail)
source, license where the row came from, and that source's stated license
kind printed, handwritten, or mixed
orig_split split/config in the source dataset (e.g. full/train)
width, height image size in pixels

How it was built

  1. Fingerprint every image: 16x16 grayscale MD5, robust to re-encoding.
  2. Decontaminate: any image whose fingerprint matches an image in UniMER-Test (23,757 formulas) or a linxy/LaTeX_OCR test split is dropped. This is deliberately conservative (coarse fingerprint), so a few legitimate images may be lost. It is not a guarantee of zero leakage.
  3. Deduplicate: rows with the same fingerprint and the same normalized label are removed, in priority order UniMER-1M, then linxy, then OleehyO. This removed 45,437 UniMER-1M duplicates, 924 from linxy and 1,928 from OleehyO.
  4. Normalize labels: UniMER-1M and linxy already use the space-tokenized convention and are unchanged. OleehyO labels (compact LaTeX) are tokenized to match, keeping \begin{...}/\end{...} glued as UniMER does. The original stays in text_raw.
  5. Split: a deterministic 0.5% validation split (by image fingerprint). There is no test split: evaluate on UniMER-Test, CROHME or similar. Do not add those to training.

Drop counts: UniMER-1M 986,122 in, 940,684 kept; linxy 173,160 in, 171,486 kept (750 test-matched); OleehyO 552,340 in, 550,412 kept.

Sources and licenses

Source License (per its card) Notes
UniMER-1M (UniMERNet authors) Apache-2.0 Itself a compilation; see its card for the underlying datasets, which may carry their own terms
linxy/LaTeX_OCR Apache-2.0 im2latex-style printed formulas plus handwritten sets
OleehyO/latex-formulas OpenRAIL Formula images scraped from arXiv; RAIL use restrictions flow to derivatives

The license column carries these statements per row. A permissive tag on a mirror does not guarantee the underlying images are free of third-party rights (arXiv formula images and some component datasets). You are responsible for checking each source's terms before commercial use. The license: other header here means "mixed, see the table above", not a new license on the content.

Known limitations

  • Label convention matters. Labels follow UniMER's convention (\infty, \,, \cal, \left, glued \begin{...}). Some Hub mirrors of UniMER rewrite labels (e.g. \infty to \infin, dropping \,); a model trained on those will score lower on UniMER-Test. This repo uses the official release.
  • Labels are noisy in the way real-world LaTeX is: many different strings render identically, so exact-match is a harsh metric. Prefer render-based metrics such as CDM.
  • UniMER-1M rows are not tagged printed/handwritten/screenshot (kind = mixed).
  • Decontamination covers UniMER-Test and the linxy test splits only. CROHME/HME100K test sets were not checked directly.

Citation

If you use this data, please cite the original sources: UniMERNet (arXiv:2404.15254, "UniMERNet: A Universal Network for Real-World Mathematical Expression Recognition"), the linxy/LaTeX_OCR and OleehyO/latex-formulas dataset pages above, and, for the CDM metric, arXiv:2409.03643.

Build script: build/build_dataset.py.

Downloads last month
314

Papers for sandysanta/latex-ocr-unified