להלן תיאור סמנטי של פעולות המוגדרות בממשק XlaBuilder
. בדרך כלל, פעולות אלה xla_data.proto
אחד לאחד לפעולות המוגדרות בממשק RPC ב- xla_data.proto
.
הערה בנושא המינוח: סוג הנתונים הכללי ש- XLA עוסק בו הוא מערך N ממדי המחזיק אלמנטים מסוג אחיד כלשהו (כגון צף של 32 סיביות). לאורך התיעוד, מערך משמש לציון מערך שרירותי-ממדי. מטעמי נוחות, למקרים מיוחדים שמות ספציפיים ומוכרים יותר; למשל וקטור הוא מערך 1-ממדי ומטריקס הוא מערך דו-ממדי.
אחרי הכל
ראה גם XlaBuilder::AfterAll
.
AfterAll לוקח מספר וריאציות של אסימונים ומייצר אסימון יחיד. אסימונים הם סוגים פרימיטיביים אשר ניתנים להשחלה בין פעולות משפיעות לוואי לאכיפת הזמנה. AfterAll
יכול לשמש AfterAll
של אסימונים להזמנת פעולה לאחר פעולות מוגדרות.
AfterAll(operands)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operands | XlaOp | מספר וריאציות של אסימונים |
AllGather
ראה גם XlaBuilder::AllGather
.
מבצע שרשור בין עותקים משוכפלים.
AllGather(operand, all_gather_dim, shard_count, replica_group_ids, channel_id)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך לשרשור בין עותקים משוכפלים. |
all_gather_dim | int64 | ממד שרשור. |
replica_groups | וקטור וקטורים של int64 | קבוצות שביניהן מתבצעת השרשור. |
channel_id | int64 אופציונלי | מזהה ערוץ אופציונלי לתקשורת חוצה-מודולים. |
-
replica_groups
היא רשימה של קבוצות העתקים שביניהן מתבצעת השרשור (ניתן לאחזר מזהה העתק עבור ההעתק הנוכחי באמצעותReplicaId
). סדר ההעתקים בכל קבוצה קובע את הסדר שבו תשומותיהם ממוקמות בתוצאה.replica_groups
חייבים להיות ריקים (במקרה כזה כל העתקים שייכים לקבוצה אחת, מסודרים מ-0
ל-N - 1
), או להכיל מספר זהה של אלמנטים כמספר העתקים. לדוגמה,replica_groups = {0, 2}, {1, 3}
מבצע שרשור בין העתקים0
ו-2
ו-1
ו-3
. -
shard_count
הוא הגודל של כל קבוצת העתקים. אנו זקוקים לכך במקרים שבהםreplica_groups
ריקים. -
channel_id
משמש לתקשורת בין-מודול: רקall-gather
פעולות עם אותוchannel_id
יכולים לתקשר אחד עם השני.
צורת הפלט היא צורת הקלט עם all_gather_dim
את all_gather_dim
shard_count
. לדוגמא, אם יש שתי עותקים משוכפלים, ואופראנד הוא בעל הערך [1.0, 2.5]
ו- [3.0, 5.25]
בהתאמה בשני ההעתקים, אז ערך הפלט מ- OP זה שבו all_gather_dim
הוא 0
יהיה [1.0, 2.5, 3.0, 5.25]
בשני ההעתקים.
AllReduce
ראה גם XlaBuilder::AllReduce
.
מבצע חישוב מותאם אישית בין עותקים משוכפלים.
AllReduce(operand, computation, replica_group_ids, channel_id)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך או כפל מערכים שאינו ריק כדי לצמצם את כל ההעתקים. |
computation | XlaComputation | חישוב הפחתה |
replica_groups | וקטור וקטורים של int64 | קבוצות שביניהן מבוצעות ההפחתות |
channel_id | int64 אופציונלי | מזהה ערוץ אופציונלי לתקשורת חוצה-מודולים |
- כאשר
operand
הוא גוון מערכים, כל ההפחתה מתבצעת על כל אלמנט של הגובה. -
replica_groups
היא רשימה של קבוצות העתקים שביניהן מתבצעת ההפחתה (ניתן לאחזר מזהה העתק עבור ההעתק הנוכחי באמצעותReplicaId
).replica_groups
חייבים להיות ריקים (במקרה כזה כל העתקים שייכים לקבוצה אחת), או להכיל מספר זהה של אלמנטים כמו מספר העתקים. לדוגמה,replica_groups = {0, 2}, {1, 3}
מבצעת הפחתה בין העתקים0
ו-2
ו-1
ו-3
. -
channel_id
משמש לתקשורת חוצה-מודולים: רק פעולותchannel_id
all-reduce
עם אותוchannel_id
יכולות לתקשר זו עם זו.
צורת הפלט זהה לצורת הקלט. לדוגמא, אם יש שתי עותקים משוכפלים ולאופראנד יש את הערך [1.0, 2.5]
ו- [3.0, 5.25]
בהתאמה בשני ההעתקים, אז ערך הפלט מחישוב אופ וסיכום זה יהיה [4.0, 7.75]
בשניהם העתקים. אם הקלט הוא כפול, הפלט הוא גם כפל.
חישוב התוצאה של AllReduce
מחייב קלט אחד מכל עותק משוכפל, כך שאם עותק מסונכרן אחד מבצע צומת AllReduce
יותר פעמים AllReduce
, הרי העתק הקודם ימתין לנצח. מכיוון שההעתקים פועלים באותה תוכנית, אין הרבה דרכים שזה יקרה, אך יתכן כאשר מצב של לולאת זמן תלוי בנתונים מההזנה והנתונים המוזנים גורמים לולאת ה- while לחזור פעמים רבות יותר. על העתק אחד מאשר אחר.
AllToAll
ראה גם XlaBuilder::AllToAll
.
AllToAll היא פעולה קולקטיבית השולחת נתונים מכל הליבות לכל הליבות. יש לו שני שלבים:
- שלב הפיזור. על כל ליבה, את האופרנד מחולק
split_count
מספר הבלוקים לאורךsplit_dimensions
, ואת הגושים מפוזרים לכל הליבות, למשל, גוש ה- i הוא לשלוח אל הליבה ה- i. - שלב האיסוף. כל ליבה משרשרת את הבלוקים שהתקבלו לאורך
concat_dimension
.
ניתן להגדיר את הליבות המשתתפות על ידי:
-
replica_groups
: כל ReplicaGroup מכיל רשימה של מזהה העתק המשתתף בחישוב (ניתן לאחזר את מזהה העתק עבור העתק הנוכחי באמצעותReplicaId
). AllToAll יוחל בתוך קבוצות משנה בסדר שצוין. לדוגמה,replica_groups = { {1,2,3}, {4,5,0} }
פירושו ש- AllToAll יוחל בתוך עותקים משוכפלים{1, 2, 3}
ובשלב האיסוף, והחסימות שהתקבלו יש לשרשר באותו סדר של 1, 2, 3. לאחר מכן, AllToAll אחר יוחל בתוך העתקים 4, 5, 0, וסדר השרשור הוא גם 4, 5, 0. אםreplica_groups
ריק, כל העתקים שייכים לאחד הקבוצה, בסדר השרשור של הופעתם.
תנאים מוקדמים:
- גודל המימד של
split_dimension
מתחלק לפיsplit_count
. - צורתו של האופרנד אינה כפולה.
AllToAll(operand, split_dimension, concat_dimension, split_count, replica_groups)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך קלט ממדי |
split_dimension | int64 | ערך במרווח [0, n) שמונה את הממד שלאורכו מפוצל האופראנד |
concat_dimension | int64 | ערך במרווח [0, n) שמונה את הממד שלאורכו משורשרות הגושים המפוצלים |
split_count | int64 | מספר הליבות שמשתתפות בפעולה זו. אם replica_groups ריק, זה צריך להיות מספר העתקים; אחרת, זה צריך להיות שווה למספר העתקים בכל קבוצה. |
replica_groups | וקטור ReplicaGroup | כל קבוצה מכילה רשימה של מזהה העתק. |
להלן דוגמה של Alltoall.
XlaBuilder b("alltoall");
auto x = Parameter(&b, 0, ShapeUtil::MakeShape(F32, {4, 16}), "x");
AllToAll(x, /*split_dimension=*/1, /*concat_dimension=*/0, /*split_count=*/4);

בדוגמה זו, ישנם 4 ליבות המשתתפות ב- Alltoall. בכל ליבת, האופרנד מחולק לארבעה חלקים לאורך הממד 0, כך שלכל חלק יש צורה f32 [4,4]. 4 החלקים מפוזרים לכל ליבות. ואז כל ליבה משרשרת את החלקים שהתקבלו לאורך מימד 1, בסדר או בליבה 0-4. כך שלפלט על כל ליבה יש צורה f32 [16,4].
BatchNormGrad
ראה גם XlaBuilder::BatchNormGrad
ונייר הנורמליזציה של האצווה המקורית לתיאור מפורט של האלגוריתם.
מחשבת שיפועים של נורמת אצווה.
BatchNormGrad(operand, scale, mean, variance, grad_output, epsilon, feature_index)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך ממדי n שיש לנורמל (x) |
scale | XlaOp | מערך ממדי אחד (\(\gamma\)) |
mean | XlaOp | מערך ממדי אחד (\(\mu\)) |
variance | XlaOp | מערך ממדי אחד (\(\sigma^2\)) |
grad_output | XlaOp | BatchNormTraining עברו ל- BatchNormTraining (\( \nabla y\)) |
epsilon | float | ערך אפסילון (\(\epsilon\)) |
feature_index | int64 | אינדקס לממד תכונות operand |
עבור כל תכונה בממד התכונה ( feature_index
הוא המדד בגין ממד תכונת operand
), המבצע מחשב את הדרגתיים ביחס operand
, offset
ו scale
בכל הממדים האחרים. feature_index
חייב להיות אינדקס חוקי עבור מאפיין התכונות operand
.
שלושת הדרגתיות מוגדרים על ידי הנוסחאות הבאות (בהנחה שמערך 4-ממדי הוא operand
ועם אינדקס מימד התכונות l
, גודל אצווה m
וגדלים מרחביים w
ו- h
):
\[ \begin{split} c_l&= \frac{1}{mwh}\sum_{i=1}^m\sum_{j=1}^w\sum_{k=1}^h \left( \nabla y_{ijkl} \frac{x_{ijkl} - \mu_l}{\sigma^2_l+\epsilon} \right) \\\\ \nabla x_{ijkl} &= \frac{\gamma_{l} }{\sqrt{\sigma^2_{l}+\epsilon} } \left( \nabla y_{ijkl} - \mathrm{mean}(\nabla y) - c_l (x_{ijkl} - \mu_{l}) \right) \\\\ \nabla \gamma_l &= \sum_{i=1}^m\sum_{j=1}^w\sum_{k=1}^h \left( \nabla y_{ijkl} \frac{x_{ijkl} - \mu_l}{\sqrt{\sigma^2_{l}+\epsilon} } \right) \\\\\ \nabla \beta_l &= \sum_{i=1}^m\sum_{j=1}^w\sum_{k=1}^h \nabla y_{ijkl} \end{split} \]
mean
הקלט variance
מייצגים ערך רגעים על פני מידות אצווה ומרחבי.
סוג הפלט הוא כפולה של שלוש ידיות:
תפוקות | סוּג | סֵמַנטִיקָה |
---|---|---|
grad_operand | XlaOp | שיפוע ביחס operand הקלט (\( \nabla x\)) |
grad_scale | XlaOp | שיפוע ביחס scale הקלט (\( \nabla \gamma\)) |
grad_offset | XlaOp | שיפוע ביחס offset הקלט (\( \nabla \beta\)) |
BatchNormInference
ראה גם XlaBuilder::BatchNormInference
ונייר הנורמליזציה של האצווה המקורית לתיאור מפורט של האלגוריתם.
מנרמל מערך על פני מידות אצווה ומרחביות.
BatchNormInference(operand, scale, offset, mean, variance, epsilon, feature_index)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך ממדי כדי לנרמל |
scale | XlaOp | מערך ממדי אחד |
offset | XlaOp | מערך ממדי אחד |
mean | XlaOp | מערך ממדי אחד |
variance | XlaOp | מערך ממדי אחד |
epsilon | float | ערך אפסילון |
feature_index | int64 | אינדקס לממד תכונות operand |
עבור כל תכונה במימד תכונה ( feature_index
הוא המדד בגין המימד תכונה operand
), המבצע מחשבת ובשונות בכל הממדים האחרים ומשתמש ובשונות לנרמל כל אלמנט operand
. feature_index
חייב להיות אינדקס חוקי עבור מאפיין התכונות operand
.
BatchNormInference
שווה ערך BatchNormInference
ל- BatchNormTraining
ללא mean
variance
מחשוב עבור כל אצווה. הוא משתמש mean
הקלט variance
במקום כערכים משוערים. מטרת פעולה זו היא להפחית את חביון ההסקה ומכאן השם BatchNormInference
.
הפלט הוא מערך מנורמלי וממודל בעל צורה זהה operand
קלט.
BatchNormTraining
ראה גם XlaBuilder::BatchNormTraining
the original batch normalization paper
לתיאור מפורט של האלגוריתם.
מנרמל מערך על פני מידות אצווה ומרחביות.
BatchNormTraining(operand, scale, offset, epsilon, feature_index)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך ממדי n לנורמליזציה (x) |
scale | XlaOp | מערך ממדי אחד (\(\gamma\)) |
offset | XlaOp | מערך ממדי אחד (\(\beta\)) |
epsilon | float | ערך אפסילון (\(\epsilon\)) |
feature_index | int64 | אינדקס לממד תכונות operand |
עבור כל תכונה בממד התכונה ( feature_index
הוא המדד בגין ממד תכונת operand
), מבצע מחשבת ובשונות בכל הממדים האחרים ומשתמש ובשונות לנרמל כל אלמנט operand
. feature_index
חייב להיות אינדקס חוקי עבור מאפיין התכונות operand
.
האלגוריתם הולך כדלקמן לכל אצווה ב operand
\(x\) המכיל m
אלמנטים עם w
ו h
כגודל של ממדים מרחביים (בהנחה operand
הוא 4 מערך מימדי):
מחשבת ממוצע אצווה \(\mu_l\) עבור כל תכונה
l
בממד התכונה: \(\mu_l=\frac{1}{mwh}\sum_{i=1}^m\sum_{j=1}^w\sum_{k=1}^h x_{ijkl}\)מחשבת שונות אצווה \(\sigma^2_l\): \(\sigma^2_l=\frac{1}{mwh}\sum_{i=1}^m\sum_{j=1}^w\sum_{k=1}^h (x_{ijkl} - \mu_l)^2\)
מנרמל, מאזניים ומשמרות: \(y_{ijkl}=\frac{\gamma_l(x_{ijkl}-\mu_l)}{\sqrt[2]{\sigma^2_l+\epsilon} }+\beta_l\)
ערך האפסון, בדרך כלל מספר קטן, מתווסף כדי למנוע שגיאות מחולקות על ידי אפס.
סוג הפלט הוא XlaOp
של שלושה XlaOp
:
תפוקות | סוּג | סֵמַנטִיקָה |
---|---|---|
output | XlaOp | מערך ממדי n עם אותה צורה כמו operand קלט (y) |
batch_mean | XlaOp | מערך ממדי אחד (\(\mu\)) |
batch_var | XlaOp | מערך ממדי אחד (\(\sigma^2\)) |
ה- batch_mean
ו- batch_var
הם רגעים המחושבים על פני מידות האצווה והמרחב באמצעות הנוסחאות לעיל.
BitcastConvertType
ראה גם XlaBuilder::BitcastConvertType
.
בדומה ל- tf.bitcast
ב- TensorFlow, מבצע פעולת bitcast מבחינה אלמנט מצורת נתונים לצורת יעד. המימדים חייבים להתאים, והמרה היא אלמנטית; למשל s32
אלמנטים הופכים f32
אלמנטים באמצעות שגרת bitcast. Bitcast מיושם כקאסט ברמה נמוכה, ולכן מכונות עם ייצוגי נקודה צפה שונה יתנו תוצאות שונות.
BitcastConvertType(operand, new_element_type)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך מסוג T עם דימומים D |
new_element_type | PrimitiveType | הקלד U |
מידות האופראנד וצורת היעד חייבות להתאים. רוחב הסיביות של סוגי המקור ורכיבי היעד חייב להיות שווה. אסור שסוגי אלמנט המקור והיעד יהיו כפולות.
מִשׁדָר
ראה גם XlaBuilder::Broadcast
.
מוסיף מימדים למערך על ידי שכפול הנתונים במערך.
Broadcast(operand, broadcast_sizes)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | המערך לשכפול |
broadcast_sizes | ArraySlice<int64> | גדלי המידות החדשות |
הממדים החדשים שנוספים מצד שמאל, כלומר אם broadcast_sizes
יש ערכים {a0, ..., aN}
ואת הצורה האופרנד יש ממדים {b0, ..., bM}
אז את הצורה של תפוקת יש ממדים {a0, ..., aN, b0, ..., bM}
.
המימדים החדשים מתווספים לעותקים של האופראנד, כלומר
output[i0, ..., iN, j0, ..., jM] = operand[j0, ..., jM]
לדוגמה, אם operand
הוא סקלר f32
עם ערך 2.0f
, ו broadcast_sizes
הוא {2, 3}
, אז התוצאה תהיה מערך עם הצורה f32[2, 3]
וכל הערכים התוצאה תהיה 2.0f
.
BroadcastInDim
ראה גם XlaBuilder::BroadcastInDim
.
מרחיב את הגודל והדרגה של מערך על ידי שכפול הנתונים במערך.
BroadcastInDim(operand, out_dim_size, broadcast_dimensions)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | המערך לשכפול |
out_dim_size | ArraySlice<int64> | גדלי המידות של צורת המטרה |
broadcast_dimensions | ArraySlice<int64> | לאיזה מימד בצורת היעד כל מימד של צורת האופרנד מתאים |
דומה לשידור, אך מאפשר להוסיף ממדים בכל מקום ולהרחיב מידות קיימות עם גודל 1.
operand
משודרת לצורה המתוארת על ידי out_dim_size
. broadcast_dimensions
ממפה את הממדים של operand
לממדים של צורת היעד, כלומר את ממד i'th של האופרנד ממופה אלי broadcast_dimension [i] "th ממד של צורת הפלט. הממדים של operand
חייבים להיות בגודל 1 או להיות בגודל זהה לממד בצורת הפלט אליהם ממופים. הממדים הנותרים מלאים במידות בגודל 1. שידור מימדי מנוון ואז משדר לאורך מימדים מנווונים אלה כדי להגיע לצורת הפלט. הסמנטיקה מתוארת בפירוט בדף השידור .
שִׂיחָה
ראה גם XlaBuilder::Call
.
מעורר חישוב עם הטיעונים הנתונים.
Call(computation, args...)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
computation | XlaComputation | חישוב מסוג T_0, T_1, ..., T_{N-1} -> S עם פרמטרים N מסוג שרירותי |
args | רצף של N XlaOp s | N טיעונים מסוג שרירותי |
הטריות וסוגי args
חייבים להתאים לפרמטרים של computation
. מותר שלא יהיה args
.
כולסקי
ראה גם XlaBuilder::Cholesky
.
מחשב את הפירוק הכלסקי של אצווה של מטריצות חיוביות סימטריות (הרמיטיות).
Cholesky(a, lower)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
a | XlaOp | מערך דרגה> 2 מסוג מורכב או נקודה צפה. |
lower | bool | אם להשתמש במשולש העליון או התחתון של a . |
אם lower
הוא true
, מחשב להנמיך-משולש מטריצות l
כזה
. אם lower
false
, מחשבים מטריצות עליונות משולשות u
כך
.
נתוני קלט נקראים רק מהמשולש התחתון / העליון של a
, בהתאם לערך lower
. מתעלמים מערכים מהמשולש האחר. נתוני הפלט מוחזרים באותו משולש; הערכים במשולש האחר מוגדרים ליישום ועשויים להיות כל דבר.
אם בדרגת עולה 2, היא כאל קבוצה של מטריצות, שבו כל פרט לקטין 2 ממדים ממדים אצווה. a
a
אם a
אינו חיובי סימטרי (הרמיטי), התוצאה מוגדרת ליישום.
מַהְדֵק
ראה גם XlaBuilder::Clamp
.
מהדק אופרנד בטווח שבין ערך מינימלי למקסימום.
Clamp(min, operand, max)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
min | XlaOp | מערך מסוג T |
operand | XlaOp | מערך מסוג T |
max | XlaOp | מערך מסוג T |
בהינתן אופרנד וערכי מינימום ומקסימום, מחזיר את האופרנד אם הוא נמצא בטווח שבין המינימום למקסימום, אחרת מחזיר את הערך המינימלי אם האופרנד נמצא מתחת לטווח זה או הערך המקסימלי אם האופרנד נמצא מעל לטווח זה. כלומר, clamp(a, x, b) = min(max(a, x), b)
.
כל שלושת המערכים חייבים להיות באותה צורה. לחלופין, כצורת שידור מוגבלת, min
ו / או max
יכולות להיות סקלר מסוג T
דוגמה עם סקלר min
max
:
let operand: s32[3] = {-1, 5, 9};
let min: s32 = 0;
let max: s32 = 6;
==>
Clamp(min, operand, max) = s32[3]{0, 5, 6};
הִתמוֹטְטוּת
ראה גם XlaBuilder::Collapse
tf.reshape
.
מכווץ ממדים של מערך לממד אחד.
Collapse(operand, dimensions)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך מסוג T |
dimensions | וקטור int64 | תת קבוצה רצופה לפי סדר ממדי T. |
קריסה מחליפה את תת המשנה הנתונה של ממדי האופרנד בממד יחיד. טיעוני הקלט הם מערך שרירותי מסוג T וקטור קבוע זמן קבוע של מדדי מימד. מדדי המימד חייבים להיות לפי סדר (מספרים של ממדים נמוכים עד גבוהים), תת קבוצה רצופה של הממדים של T. לפיכך, {0, 1, 2}, {0, 1} או {1, 2} הם כל קבוצות המאפיינים החוקיות, אך {1, 0} או {0, 2} אינן. הם מוחלפים על ידי מימד חדש יחיד, באותו מיקום ברצף המידות כמו אלה שהם מחליפים, כאשר גודל הממד החדש שווה לתוצר של גדלי המידות המקוריים. מספר המימדים הנמוך ביותר במימדים הוא dimensions
המשתנה האיטי ביותר (העיקרי ביותר) בקן הלולאה שממוטט מימד זה, ומספר המידות הגבוה ביותר הוא המשתנה המהיר ביותר (מינורי ביותר). עיין במפעיל tf.reshape
אם יש צורך בהזמנת קריסה כללית יותר.
לדוגמא, בואו להיות מערך של 24 אלמנטים:
let v = f32[4x2x3] { { {10, 11, 12}, {15, 16, 17} },
{ {20, 21, 22}, {25, 26, 27} },
{ {30, 31, 32}, {35, 36, 37} },
{ {40, 41, 42}, {45, 46, 47} } };
// Collapse to a single dimension, leaving one dimension.
let v012 = Collapse(v, {0,1,2});
then v012 == f32[24] {10, 11, 12, 15, 16, 17,
20, 21, 22, 25, 26, 27,
30, 31, 32, 35, 36, 37,
40, 41, 42, 45, 46, 47};
// Collapse the two lower dimensions, leaving two dimensions.
let v01 = Collapse(v, {0,1});
then v01 == f32[4x6] { {10, 11, 12, 15, 16, 17},
{20, 21, 22, 25, 26, 27},
{30, 31, 32, 35, 36, 37},
{40, 41, 42, 45, 46, 47} };
// Collapse the two higher dimensions, leaving two dimensions.
let v12 = Collapse(v, {1,2});
then v12 == f32[8x3] { {10, 11, 12},
{15, 16, 17},
{20, 21, 22},
{25, 26, 27},
{30, 31, 32},
{35, 36, 37},
{40, 41, 42},
{45, 46, 47} };
קולקטיב
ראה גם XlaBuilder::CollectivePermute
.
CollectivePermute היא פעולה קולקטיבית השולחת ומקבלת העתקי נתונים צולבים.
CollectivePermute(operand, source_target_pairs)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך קלט ממדי |
source_target_pairs | <int64, int64> | רשימה של (source_replica_id, target_replica_id) זוגות. עבור כל זוג, האופרנד נשלח מהעתק המקור לשכפול היעד. |
שים לב כי ישנן המגבלות הבאות לגבי source_target_pair
:
- לשני זוגות לא צריך להיות מזהה העתק יעד זהה, ולא צריך להיות אותו מזהה העתק מקור.
- אם מזהה העתק אינו מטרה בשום זוג, הפלט באותה העתק הוא טנסור מורכב מ- 0 (ים) עם אותה צורה כמו הקלט.
שרשור
ראה גם XlaBuilder::ConcatInDim
.
שרשור מחבר מערך ממספר אופנדות מערך. המערך הוא באותו דרגה של כל אחד ממפעילי מערך הקלט (שחייב להיות מאותו דרגה זה לזה) ומכיל את הארגומנטים לפי הסדר שצוינו.
Concatenate(operands..., dimension)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operands | רצף של N XlaOp | מערכי N מסוג T עם מידות [L0, L1, ...]. דורש N> = 1. |
dimension | int64 | ערך במרווח [0, N) המכנה את המימד שיש לשרשר בין operands . |
למעט dimension
כל הממדים חייבים להיות זהים. הסיבה לכך היא ש- XLA אינו תומך במערכים "מרופטים". כמו כן, שים לב שלא ניתן לשרשר בין ערכי דירוג 0 (מכיוון שאי אפשר למנות את הממד שלאורכו מתרחש השרשור).
דוגמה חד מימדית:
Concat({ {2, 3}, {4, 5}, {6, 7} }, 0)
>>> {2, 3, 4, 5, 6, 7}
דוגמה דו ממדית:
let a = {
{1, 2},
{3, 4},
{5, 6},
};
let b = {
{7, 8},
};
Concat({a, b}, 0)
>>> {
{1, 2},
{3, 4},
{5, 6},
{7, 8},
}
תרשים:

מותנה
ראה גם XlaBuilder::Conditional
.
Conditional(pred, true_operand, true_computation, false_operand, false_computation)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
pred | XlaOp | סולם של סוג PRED |
true_operand | XlaOp | טיעון מסוג \(T_0\) |
true_computation | XlaComputation | חישוב Xla מסוג \(T_0 \to S\) |
false_operand | XlaOp | טיעון מסוג \(T_1\) |
false_computation | XlaComputation | חישוב Xla מסוג \(T_1 \to S\) |
מבצע true_computation
אם pred
true
, false_computation
אם pred
הוא false
ומחזיר את התוצאה.
true_computation
חייב לקבל ארגומנט יחיד מסוג \(T_0\) true_operand
באמצעות true_operand
שעליו להיות מאותו סוג. חישוב false_computation
חייב false_computation
ארגומנט יחיד מסוג \(T_1\) false_operand
באמצעות false_operand
שעליו להיות מאותו סוג. סוג הערך המוחזר של true_computation
ו- false_computation
חייב להיות זהה.
שים לב שרק אחד true_computation
false_computation
יבוצע בהתאם לערך pred
.
Conditional(branch_index, branch_computations, branch_operands)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
branch_index | XlaOp | מדרגות מסוג S32 |
branch_computations | רצף של N XlaComputation | Xla חישובים מסוג \( T_0 \to S , T_1 \to S , ..., T_{N-1} \to S \) |
branch_operands | רצף של N XlaOp | טיעונים מסוג \( T_0 , T_1 , ..., T_{N-1} \) |
מבצע branch_computations[branch_index]
ומחזיר את התוצאה. אם branch_index
הוא S32
שהוא <0 או> = N, אז branch_computations[N-1]
מבוצע כענף ברירת המחדל.
כל branch_computations[b]
חייבים לקחת ארגומנט יחיד מסוג T_b
branch_operands[b]
באמצעות branch_operands[b]
אשר חייבים להיות מאותו סוג. סוג הערך שהוחזר של כל branch_computations[b]
חייב להיות זהה.
שים לב שרק אחת מ- branch_computations
תבוצע בהתאם לערך של branch_index
.
המרה (פיתול)
ראה גם XlaBuilder::Conv
.
כ- ConvWithGeneralPadding, אך הריפוד מוגדר באופן קצר כ- SAME או VALID. ריפוד זהה מרפד את הקלט ( lhs
) lhs
כך lhs
תהיה אותה צורה כמו הקלט כאשר לא לוקחים בחשבון צעד. ריפוד תקף פירושו שאין ריפוד.
ConvWithGeneralPadding (קונבולציה)
ראה גם XlaBuilder::ConvWithGeneralPadding
.
מחשבת פיתול מהסוג המשמש ברשתות עצביות. כאן, ניתן לחשוב על קונבולוציה כחלון ממדי n הנע על פני שטח בסיס ממדי n ומבוצע חישוב לכל מיקום אפשרי של החלון.
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
lhs | XlaOp | דרג מערך תשומות n + 2 |
rhs | XlaOp | דרג מערך n + 2 של משקולות גרעין |
window_strides | ArraySlice<int64> | מערך נדיר של צעדי הליבה |
padding | ArraySlice< pair<int64, int64>> | מערך נדיר של ריפוד (נמוך, גבוה) |
lhs_dilation | ArraySlice<int64> | מערך גורמי התרחבות nd lhs |
rhs_dilation | ArraySlice<int64> | מערך גורמי התרחבות nd rhs |
feature_group_count | int64 | מספר קבוצות התכונות |
batch_group_count | int64 | מספר קבוצות האצווה |
תן n להיות מספר הממדים המרחבים. הטיעון lhs
הוא מערך n + 2 בדרגה המתאר את שטח הבסיס. זה נקרא קלט, למרות שכמובן שה- rhs הוא גם קלט. ברשת עצבית, אלה פעולות הקלט. הממדים n + 2 הם, בסדר זה:
-
batch
: כל קואורדינטות בממד זה מייצגות קלט עצמאי שעבורו מתבצע קונבולוציה. -
z/depth/features
: לכל מיקום (y, x) באזור הבסיס יש וקטור המשויך אליו, שנכנס לממד זה. -
spatial_dims
: מתאר אתn
הממדיםspatial_dims
המגדירים את שטח הבסיס שעליו החלון עובר.
הטיעון rhs
הוא מערך n + 2 של דרגה המתאר את המסנן / הליבה / החלון. המידות הן בסדר זה:
-
output-z
: ממדz
של הפלט. -
input-z
: גודל המימד הזה כפולfeature_group_count
צריך להיות שווה לגודל הממדz
ב- lhs. -
spatial_dims
: מתאר אתn
ממדים מרחביים המגדירים את חלון nd כי מהלך על פני שטח הבסיס.
הארגומנט של window_strides
מציין את window_strides
של החלון window_strides
בממדים המרחבים. לדוגמא, אם הצעד בממד המרחבי הראשון הוא 3, ניתן להציב את החלון רק בקואורדינטות בהן האינדקס המרחבי הראשון מתחלק ב -3.
טיעון padding
מציין את כמות ריפוד האפס שיש להחיל על שטח הבסיס. כמות הריפוד יכולה להיות שלילית - הערך המוחלט של ריפוד שלילי מציין את מספר האלמנטים שיש להסיר מהמימד שצוין לפני ביצוע הפיתול. padding[0]
מציין את הריפוד למימד y
padding[1]
מציין את הריפוד למימד x
. לכל זוג הריפוד הנמוך כאלמנט הראשון והריפוד הגבוה כאלמנט השני. הריפוד הנמוך מוחל לכיוון מדדים נמוכים יותר ואילו הריפוד הגבוה מוחל לכיוון מדדים גבוהים יותר. לדוגמא, אם padding[1]
הוא (2,3)
אז יהיה ריפוד על ידי 2 אפסים משמאל ועל ידי 3 אפסים מימין במימד המרחבי השני. שימוש בריפוד שווה ערך להכנסת אותם אפס ערכים לקלט ( lhs
) לפני ביצוע lhs
.
lhs_dilation
ו rhs_dilation
הטיעונים לציין את הגורם הרחב כדי להיות מיושם על LHS ובצד שמאל, בהתאמה, בכל ממד מרחבית. אם גורם ההתרחבות בממד מרחבי הוא d, אז חורים d-1 ממוקמים באופן מרומז בין כל אחד מהערכים בממד זה, מה שמגדיל את גודל המערך. החורים מלאים בערך no-op, שמשמעותו עבור פיתול היא אפסים.
התרחבות של ה- Rhs נקראת גם התפרצות חמורה. לפרטים נוספים, ראה tf.nn.atrous_conv2d
. התרחבות ה- lhs נקראת גם קונבולוציה מועברת. לפרטים נוספים, ראה tf.nn.conv2d_transpose
.
ניתן feature_group_count
בארגומנט feature_group_count
(ערך ברירת מחדל 1) feature_group_count
מקובצות. feature_group_count
צריכה להיות מחלק הן את קלט והן את ממד תכונות הפלט. אם feature_group_count
גדולה מ- 1, המשמעות היא rhs
ממד תכונות הקלט והפלט ומימד תכונות הפלט של ה- rhs
מפוצלים באופן שווה לקבוצות רבות של feature_group_count
, שכל קבוצה מורכבת feature_group_count
תכונות רצוף. ממד תכונות הקלט של ה- rhs
צריך להיות שווה lhs
המאפיינים של קלט ה- lhs
מחולק על ידי feature_group_count
(כך שכבר יש לו גודל של קבוצת תכונות קלט). הקבוצות i-th משמשות יחד כדי לחשב את feature_group_count
נפרדות רבות. התוצאות של התפתלות אלה משורשרות יחד בממד תכונות הפלט.
לפיתול feature_group_count
ארגומנט feature_group_count
יוגדר לממד תכונת הקלט [filter_height, filter_width, in_channels, channel_multiplier]
מחדש מ- [filter_height, filter_width, in_channels, channel_multiplier]
ל- [filter_height, filter_width, 1, in_channels * channel_multiplier]
. לפרטים נוספים, ראה tf.nn.depthwise_conv2d
.
ניתן להשתמש batch_group_count
(ערך ברירת מחדל 1) למסננים מקובצים במהלך ההיסטוריה batch_group_count
. batch_group_count
צריך להיות מחלק את גודל ממד האצווה lhs
(קלט). אם batch_group_count
גדול מ- 1, המשמעות היא שממד האצווה של הפלט צריך להיות של input batch / batch_group_count
. על batch_group_count
להיות מחלק את גודל תכונת הפלט.
לצורת הפלט יש מימדים אלה, בסדר זה:
-
batch
: גודל הממד הזה כפולbatch_group_count
צריך להיות שווה לגודל מימדbatch
ב- lhs. -
z
: גודל זההrhs
output-z
על הגרעין (rhs
). -
spatial_dims
: ערך אחד לכל מיקום חוקי של חלון ההתפתחות.
המיקומים התקפים של חלון ההתפחחות נקבעים על פי הצעדים וגודל שטח הבסיס לאחר הריפוד.
כדי לתאר מה עושה פיתול, שקול פיתול דו-ממדי ובחר כמה קבוצות batch
קבועה, z
, y
, x
בפלט. ואז (y,x)
הוא מיקום של פינה של החלון בתוך שטח הבסיס (למשל בפינה השמאלית העליונה, תלוי איך אתה מפרש את המידות המרחביות). כעת יש לנו חלון דו-ממדי, שנלקח מאזור הבסיס, בו כל נקודה דו-ממדית משויכת לווקטור 1d, כך שנקבל תיבה תלת-ממדית. מגרעין הפיתול, מכיוון שקיבענו את קואורדינטות הפלט z
, יש לנו גם תיבה תלת-ממדית. שתי התיבות בעלות אותם מידות, כך שנוכל לקחת את סכום המוצרים החכמים אלמנט בין שתי התיבות (בדומה למוצר נקודה). זהו ערך הפלט.
שים לב שאם output-z
הוא למשל 5, אז כל מיקום של החלון מייצר 5 ערכים בפלט לממד z
של הפלט. ערכים אלה נבדלים באיזה חלק מגרעין הפיתול משתמשים - יש תיבת תלת-ממד נפרדת של ערכים המשמשת לכל קואורדינטות output-z
. אז אתה יכול לחשוב על זה כ -5 התפתלות נפרדת עם פילטר שונה לכל אחד מהם.
הנה קוד פסאודו לפיתול דו-ממדי עם ריפוד וצעדים:
for (b, oz, oy, ox) { // output coordinates
value = 0;
for (iz, ky, kx) { // kernel coordinates and input z
iy = oy*stride_y + ky - pad_low_y;
ix = ox*stride_x + kx - pad_low_x;
if ((iy, ix) inside the base area considered without padding) {
value += input(b, iz, iy, ix) * kernel(oz, iz, ky, kx);
}
}
output(b, oz, oy, ox) = value;
}
ConvertElementType
ראה גם XlaBuilder::ConvertElementType
.
בדומה ל- static_cast
אלמנט ב- C ++, מבצע פעולת המרה static_cast
מצורת נתונים לצורת יעד. המימדים חייבים להתאים, והמרה היא אלמנטית; למשל s32
אלמנטים הופכים f32
אלמנטים באמצעות s32
-to- f32
שגרת ההמרה.
ConvertElementType(operand, new_element_type)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך מסוג T עם דימומים D |
new_element_type | PrimitiveType | הקלד U |
מידות האופראנד וצורת היעד חייבות להתאים. אסור שסוגי אלמנט המקור והיעד יהיו כפולות.
המרה כגון T=s32
ל- U=f32
תבצע שגרת המרה של U=f32
-לצוף U=f32
כגון עגול-עד-קרוב ביותר.
let a: s32[3] = {0, 1, 2};
let b: f32[3] = convert(a, f32);
then b == f32[3]{0.0, 1.0, 2.0}
CrossReplicaSum
מבצע את AllReduce
באמצעות חישוב סיכום.
CustomCall
ראה גם XlaBuilder::CustomCall
.
התקשר לפונקציה המסופקת על ידי המשתמש בתוך חישוב.
CustomCall(target_name, args..., shape)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
target_name | string | שם הפונקציה. תונפק הוראת שיחה המכוונת לשם סמל זה. |
args | רצף של N XlaOp s | טיעוני N מסוג שרירותי, שיעברו לפונקציה. |
shape | Shape | צורת הפלט של הפונקציה |
חתימת הפונקציה זהה, ללא קשר לאריבות או סוג הטענות:
extern "C" void target_name(void* out, void** in);
לדוגמה, אם משתמשים ב- CustomCall באופן הבא:
let x = f32[2] {1,2};
let y = f32[2x3] { {10, 20, 30}, {40, 50, 60} };
CustomCall("myfunc", {x, y}, f32[3x3])
הנה דוגמה ליישום של myfunc
:
extern "C" void myfunc(void* out, void** in) {
float (&x)[2] = *static_cast<float(*)[2]>(in[0]);
float (&y)[2][3] = *static_cast<float(*)[2][3]>(in[1]);
EXPECT_EQ(1, x[0]);
EXPECT_EQ(2, x[1]);
EXPECT_EQ(10, y[0][0]);
EXPECT_EQ(20, y[0][1]);
EXPECT_EQ(30, y[0][2]);
EXPECT_EQ(40, y[1][0]);
EXPECT_EQ(50, y[1][1]);
EXPECT_EQ(60, y[1][2]);
float (&z)[3][3] = *static_cast<float(*)[3][3]>(out);
z[0][0] = x[1] + y[1][0];
// ...
}
לפונקציה המסופקת על ידי המשתמש לא יכולות להיות תופעות לוואי והביצוע שלה חייב להיות חסר יכולת.
נְקוּדָה
ראה גם XlaBuilder::Dot
.
Dot(lhs, rhs)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
lhs | XlaOp | מערך מסוג T |
rhs | XlaOp | מערך מסוג T |
הסמנטיקה המדויקת של פעולה זו תלויה בשורות האופרנדים:
קֶלֶט | תְפוּקָה | סֵמַנטִיקָה |
---|---|---|
וקטור [n] וקטור dot [n] | סקלר | מוצר נקודה וקטורית |
מטריצה [mxk] וקטור dot [k] | וקטור [m] | כפל מטריצה-וקטורי |
מטריצה [mxk] מטריצת dot [kxn] | מטריצה [mxn] | כפל מטריצה-מטריצה |
הפעולה מבצעת סכום של מוצרים על פני הממד השני של lhs
(או הראשון אם יש לו דרגה 1) ואת הממד הראשון של rhs
. אלה הממדים ה"מכווצים ". המידות lhs
של lhs
ו- rhs
חייבות להיות באותו גודל. בפועל, ניתן להשתמש בו לביצוע מוצרי נקודה בין וקטורים, כפל וקטור / מטריצה או כפל מטריצה / מטריצה.
DotGeneral
ראה גם XlaBuilder::DotGeneral
.
DotGeneral(lhs, rhs, dimension_numbers)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
lhs | XlaOp | מערך מסוג T |
rhs | XlaOp | מערך מסוג T |
dimension_numbers | DotDimensionNumbers | מספרי מיצוי חוזה ואצווה |
כנקודה, אך מאפשר לציין מספרי מימד חוזה ואצווה הן עבור 'lhs' והן עבור 'rhs'.
שדות DotDimensionNumbers | סוּג | סֵמַנטִיקָה |
---|---|---|
'lhs_contraction_dimensions' | int64 חוזר | מספרי המימדים החוזים של 'lhs' |
'ממדי rhs_contracting_dimensions' | int64 חוזר | מספרי ממדים מכווצים של 'rhs' |
'lhs_batch_dimensions' | int64 חוזר | מספרי מידות אצווה של 'lhs' |
'rhs_batch_dimensions' | int64 חוזר | מספרי מידות אצווה של 'rhs' |
DotGeneral מבצעת את סכום המוצרים על פני ממדים מתכווצים שצוינו ב'ממדים_מספרים '.
מספרי מימד חוזה משויכים מה- 'lhs' ו- 'rhs' אינם צריכים להיות זהים אלא חייבים להיות בעלי אותם מידות מימד.
דוגמה עם מספרי מימד חוזה:
lhs = { {1.0, 2.0, 3.0},
{4.0, 5.0, 6.0} }
rhs = { {1.0, 1.0, 1.0},
{2.0, 2.0, 2.0} }
DotDimensionNumbers dnums;
dnums.add_lhs_contracting_dimensions(1);
dnums.add_rhs_contracting_dimensions(1);
DotGeneral(lhs, rhs, dnums) -> { {6.0, 12.0},
{15.0, 30.0} }
מספרים של מאפייני אצווה משויכים מה- 'lhs' ו- 'rhs' חייבים להיות בעלי אותם מידות מימד.
דוגמה עם מספרי מימד אצווה (גודל אצווה 2, מטריצות 2x2):
lhs = { { {1.0, 2.0},
{3.0, 4.0} },
{ {5.0, 6.0},
{7.0, 8.0} } }
rhs = { { {1.0, 0.0},
{0.0, 1.0} },
{ {1.0, 0.0},
{0.0, 1.0} } }
DotDimensionNumbers dnums;
dnums.add_lhs_contracting_dimensions(2);
dnums.add_rhs_contracting_dimensions(1);
dnums.add_lhs_batch_dimensions(0);
dnums.add_rhs_batch_dimensions(0);
DotGeneral(lhs, rhs, dnums) -> { { {1.0, 2.0},
{3.0, 4.0} },
{ {5.0, 6.0},
{7.0, 8.0} } }
קֶלֶט | תְפוּקָה | סֵמַנטִיקָה |
---|---|---|
[b0, m, k] dot [b0, k, n] | [b0, m, n] | מטמול אצווה |
[b0, b1, m, k] dot [b0, b1, k, n] | [b0, b1, m, n] | מטמול אצווה |
מכאן נובע שמספר המימדים המתקבל מתחיל בממד האצווה, ואז המימד 'lhs' שאינו חוזה / שאינו אצווה, ולבסוף הממד 'rhs' שאינו חוזה / שאינו אצווה.
DynamicSlice
ראה גם XlaBuilder::DynamicSlice
.
DynamicSlice מחלץ מערך משנה ממערך הקלט ב start_indices
דינמי. גודל הפרוסה בכל מימד מועבר ב- size_indices
, המציינים את נקודת הסיום של מרווחי פרוסות בלעדיים בכל מימד: [התחלה, התחלה + גודל]. הצורה של start_indices
חייבת להיות דרגה == 1, וגודל הממד שווה לדרגת operand
.
DynamicSlice(operand, start_indices, size_indices)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך ממדי N מסוג T |
start_indices | רצף של N XlaOp | רשימה של מספרים שלמים N סקלריים המכילים את מדדי ההתחלה של הנתח לכל מימד. הערך חייב להיות גדול מאפס או שווה לו. |
size_indices | ArraySlice<int64> | רשימה של מספרים N שלמים המכילים את גודל הפרוסה לכל מימד. על כל ערך להיות גדול מאפס, וגודל ההתחלה + חייב להיות קטן או שווה לגודל המימד כדי להימנע מגודל מימד מודולו עטוף. |
מדדי הפרוסות היעילים מחושבים על ידי יישום הטרנספורמציה הבאה לכל אינדקס i
ב- [1, N)
לפני ביצוע הפרוסה:
start_indices[i] = clamp(start_indices[i], 0, operand.dimension_size[i] - size_indices[i])
זה מבטיח שהפרוסה שחולצה תמיד תהיה בגבול ביחס למערך האופרנד. אם הנתח נמצא בגבול לפני החלת הטרנספורמציה, לטרנספורמציה אין כל השפעה.
דוגמה חד מימדית:
let a = {0.0, 1.0, 2.0, 3.0, 4.0}
let s = {2}
DynamicSlice(a, s, {2}) produces:
{2.0, 3.0}
דוגמה דו ממדית:
let b =
{ {0.0, 1.0, 2.0},
{3.0, 4.0, 5.0},
{6.0, 7.0, 8.0},
{9.0, 10.0, 11.0} }
let s = {2, 1}
DynamicSlice(b, s, {2, 2}) produces:
{ { 7.0, 8.0},
{10.0, 11.0} }
DynamicUpdateSlice
ראה גם XlaBuilder::DynamicUpdateSlice
.
DynamicUpdateSlice מייצר תוצאה המהווה את הערך של מערך הקלט operand
, עם פרוס update
מוחלף על start_indices
. צורת update
קובעת את צורת מערך המשנה של התוצאה המתעדכן. הצורה של start_indices
חייבת להיות דרגה == 1, וגודל הממד שווה לדרגת operand
.
DynamicUpdateSlice(operand, update, start_indices)
ויכוחים | סוּג | סֵמַנטִיקָה |
---|---|---|
operand | XlaOp | מערך ממדי N מסוג T |
update | XlaOp | מערך ממדי N מסוג T המכיל את עדכון הפרוסות. Each dimension of update shape must be strictly greater than zero, and start + update must be less than or equal to the operand size for each dimension to avoid generating out-of-bounds update indices. |
start_indices | sequence of N XlaOp | List of N scalar integers containing the starting indices of the slice for each dimension. Value must be greater than or equal to zero. |
The effective slice indices are computed by applying the following transformation for each index i
in [1, N)
before performing the slice:
start_indices[i] = clamp(start_indices[i], 0, operand.dimension_size[i] - update.dimension_size[i])
This ensures that the updated slice is always in-bounds with respect to the operand array. If the slice is in-bounds before the transformation is applied, the transformation has no effect.
1-dimensional example:
let a = {0.0, 1.0, 2.0, 3.0, 4.0}
let u = {5.0, 6.0}
let s = {2}
DynamicUpdateSlice(a, u, s) produces:
{0.0, 1.0, 5.0, 6.0, 4.0}
2-dimensional example:
let b =
{ {0.0, 1.0, 2.0},
{3.0, 4.0, 5.0},
{6.0, 7.0, 8.0},
{9.0, 10.0, 11.0} }
let u =
{ {12.0, 13.0},
{14.0, 15.0},
{16.0, 17.0} }
let s = {1, 1}
DynamicUpdateSlice(b, u, s) produces:
{ {0.0, 1.0, 2.0},
{3.0, 12.0, 13.0},
{6.0, 14.0, 15.0},
{9.0, 16.0, 17.0} }
Element-wise binary arithmetic operations
See also XlaBuilder::Add
.
A set of element-wise binary arithmetic operations is supported.
Op(lhs, rhs)
Where Op
is one of Add
(addition), Sub
(subtraction), Mul
(multiplication), Div
(division), Rem
(remainder), Max
(maximum), Min
(minimum), LogicalAnd
(logical AND), or LogicalOr
(logical OR).
Arguments | Type | Semantics |
---|---|---|
lhs | XlaOp | left-hand-side operand: array of type T |
rhs | XlaOp | right-hand-side operand: array of type T |
The arguments' shapes have to be either similar or compatible. See the broadcasting documentation about what it means for shapes to be compatible. The result of an operation has a shape which is the result of broadcasting the two input arrays. In this variant, operations between arrays of different ranks are not supported, unless one of the operands is a scalar.
When Op
is Rem
, the sign of the result is taken from the dividend, and the absolute value of the result is always less than the divisor's absolute value.
Integer division overflow (signed/unsigned division/remainder by zero or signed division/remainder of INT_SMIN
with -1
) produces an implementation defined value.
An alternative variant with different-rank broadcasting support exists for these operations:
Op(lhs, rhs, broadcast_dimensions)
Where Op
is the same as above. This variant of the operation should be used for arithmetic operations between arrays of different ranks (such as adding a matrix to a vector).
The additional broadcast_dimensions
operand is a slice of integers used to expand the rank of the lower-rank operand up to the rank of the higher-rank operand. broadcast_dimensions
maps the dimensions of the lower-rank shape to the dimensions of the higher-rank shape. The unmapped dimensions of the expanded shape are filled with dimensions of size one. Degenerate-dimension broadcasting then broadcasts the shapes along these degenerate dimensions to equalize the shapes of both operands. The semantics are described in detail on the broadcasting page .
Element-wise comparison operations
See also XlaBuilder::Eq
.
A set of standard element-wise binary comparison operations is supported. Note that standard IEEE 754 floating-point comparison semantics apply when comparing floating-point types.
Op(lhs, rhs)
Where Op
is one of Eq
(equal-to), Ne
(not equal-to), Ge
(greater-or-equal-than), Gt
(greater-than), Le
(less-or-equal-than), Lt
(less-than). Another set of operators, EqTotalOrder, NeTotalOrder, GeTotalOrder, GtTotalOrder, LeTotalOrder, and LtTotalOrder, provide the same functionalities, except that they additionally support a total order over the floating point numbers, by enforcing -NaN < -Inf < -Finite < -0 < +0 < +Finite < +Inf < +NaN.
Arguments | Type | Semantics |
---|---|---|
lhs | XlaOp | left-hand-side operand: array of type T |
rhs | XlaOp | right-hand-side operand: array of type T |
The arguments' shapes have to be either similar or compatible. See the broadcasting documentation about what it means for shapes to be compatible. The result of an operation has a shape which is the result of broadcasting the two input arrays with the element type PRED
. In this variant, operations between arrays of different ranks are not supported, unless one of the operands is a scalar.
An alternative variant with different-rank broadcasting support exists for these operations:
Op(lhs, rhs, broadcast_dimensions)
Where Op
is the same as above. This variant of the operation should be used for comparison operations between arrays of different ranks (such as adding a matrix to a vector).
The additional broadcast_dimensions
operand is a slice of integers specifying the dimensions to use for broadcasting the operands. The semantics are described in detail on the broadcasting page .
Element-wise unary functions
XlaBuilder supports these element-wise unary functions:
Abs(operand)
Element-wise abs x -> |x|
.
Ceil(operand)
Element-wise ceil x -> ⌈x⌉
.
Cos(operand)
Element-wise cosine x -> cos(x)
.
Exp(operand)
Element-wise natural exponential x -> e^x
.
Floor(operand)
Element-wise floor x -> ⌊x⌋
.
Imag(operand)
Element-wise imaginary part of a complex (or real) shape. x -> imag(x)
. If the operand is a floating point type, returns 0.
IsFinite(operand)
Tests whether each element of operand
is finite, ie, is not positive or negative infinity, and is not NaN
. Returns an array of PRED
values with the same shape as the input, where each element is true
if and only if the corresponding input element is finite.
Log(operand)
Element-wise natural logarithm x -> ln(x)
.
LogicalNot(operand)
Element-wise logical not x -> !(x)
.
Logistic(operand)
Element-wise logistic function computation x -> logistic(x)
.
PopulationCount(operand)
Computes the number of bits set in each element of operand
.
Neg(operand)
Element-wise negation x -> -x
.
Real(operand)
Element-wise real part of a complex (or real) shape. x -> real(x)
. If the operand is a floating point type, returns the same value.
Rsqrt(operand)
Element-wise reciprocal of square root operation x -> 1.0 / sqrt(x)
.
Sign(operand)
Element-wise sign operation x -> sgn(x)
where
using the comparison operator of the element type of operand
.
Sqrt(operand)
Element-wise square root operation x -> sqrt(x)
.
Cbrt(operand)
Element-wise cubic root operation x -> cbrt(x)
.
Tanh(operand)
Element-wise hyperbolic tangent x -> tanh(x)
.
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | The operand to the function |
The function is applied to each element in the operand
array, resulting in an array with the same shape. It is allowed for operand
to be a scalar (rank 0).
Fft
The XLA FFT operation implements the forward and inverse Fourier Transforms for real and complex inputs/outputs. Multidimensional FFTs on up to 3 axes are supported, except on TPU, where only a single axis is supported (please file a GitHub issue if you require higher order).
See also XlaBuilder::Fft
.
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | The array we are Fourier transforming. |
fft_type | FftType | See the table below. |
fft_length | ArraySlice<int64> | The time-domain lengths of the axes being transformed. This is needed in particular for IRFFT to right-size the innermost axis, since RFFT(fft_length=[16]) has the same output shape as RFFT(fft_length=[17]) . |
FftType | Semantics |
---|---|
FFT | Forward complex-to-complex FFT. Shape is unchanged. |
IFFT | Inverse complex-to-complex FFT. Shape is unchanged. |
RFFT | Forward real-to-complex FFT. Shape of the innermost axis is reduced to fft_length[-1] // 2 + 1 if fft_length[-1] is a non-zero value, omitting the reversed conjugate part of the transformed signal beyond the Nyquist frequency. |
IRFFT | Inverse real-to-complex FFT (ie takes complex, returns real). Shape of the innermost axis is expanded to fft_length[-1] if fft_length[-1] is a non-zero value, inferring the part of the transformed signal beyond the Nyquist frequency from the reverse conjugate of the 1 to fft_length[-1] // 2 + 1 entries. |
Multidimensional FFT
When more than 1 fft_length
is provided, this is equivalent to applying a cascade of FFT operations to each of the innermost axes. Note that for the real->complex and complex->real cases, the innermost axis transform is (effectively) performed first (RFFT; last for IRFFT), which is why the innermost axis is the one which changes size. Other axis transforms will then be complex->complex.
Implementation details
CPU FFT is backed by Eigen's TensorFFT. GPU FFT uses cuFFT.
Gather
The XLA gather operation stitches together several slices (each slice at a potentially different runtime offset) of an input array.
General Semantics
See also XlaBuilder::Gather
. For a more intuitive description, see the "Informal Description" section below.
gather(operand, start_indices, offset_dims, collapsed_slice_dims, slice_sizes, start_index_map)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | The array we're gathering from. |
start_indices | XlaOp | Array containing the starting indices of the slices we gather. |
index_vector_dim | int64 | The dimension in start_indices that "contains" the starting indices. See below for a detailed description. |
offset_dims | ArraySlice<int64> | The set of dimensions in the output shape that offset into an array sliced from operand. |
slice_sizes | ArraySlice<int64> | slice_sizes[i] is the bounds for the slice on dimension i . |
collapsed_slice_dims | ArraySlice<int64> | The set of dimensions in each slice that are collapsed away. These dimensions must have size 1. |
start_index_map | ArraySlice<int64> | A map that describes how to map indices in start_indices to legal indices into operand. |
indices_are_sorted | bool | Whether the indices are guaranteed to be sorted by the caller. |
unique_indices | bool | Whether the indices are guaranteed to be unique by the caller. |
For convenience, we label dimensions in the output array not in offset_dims
as batch_dims
.
The output is an array of rank batch_dims.size
+ offset_dims.size
.
The operand.rank
must equal the sum of offset_dims.size
and collapsed_slice_dims
. Also, slice_sizes.size
has to be equal to operand.rank
.
If index_vector_dim
is equal to start_indices.rank
we implicitly consider start_indices
to have a trailing 1
dimension (ie if start_indices
was of shape [6,7]
and index_vector_dim
is 2
then we implicitly consider the shape of start_indices
to be [6,7,1]
).
The bounds for the output array along dimension i
is computed as follows:
If
i
is present inbatch_dims
(ie is equal tobatch_dims[k]
for somek
) then we pick the corresponding dimension bounds out ofstart_indices.shape
, skippingindex_vector_dim
(ie pickstart_indices.shape.dims
[k
] ifk
<index_vector_dim
andstart_indices.shape.dims
[k
+1
] otherwise).If
i
is present inoffset_dims
(ie equal tooffset_dims
[k
] for somek
) then we pick the corresponding bound out ofslice_sizes
after accounting forcollapsed_slice_dims
(ie we pickadjusted_slice_sizes
[k
] whereadjusted_slice_sizes
isslice_sizes
with the bounds at indicescollapsed_slice_dims
removed).
Formally, the operand index In
corresponding to a given output index Out
is calculated as follows:
Let
G
= {Out
[k
] fork
inbatch_dims
}. UseG
to slice out a vectorS
such thatS
[i
] =start_indices
[Combine(G
,i
)] where Combine(A, b) inserts b at positionindex_vector_dim
into A. Note that this is well defined even ifG
is empty -- ifG
is empty thenS
=start_indices
.Create a starting index,
S
in
, intooperand
usingS
by scatteringS
usingstart_index_map
. More precisely:S
in
[start_index_map
[k
]] =S
[k
] ifk
<start_index_map.size
.S
in
[_
] =0
otherwise.
Create an index
O
in
intooperand
by scattering the indices at the offset dimensions inOut
according to thecollapsed_slice_dims
set. More precisely:O
in
[remapped_offset_dims
(k
)] =Out
[offset_dims
[k
]] ifk
<offset_dims.size
(remapped_offset_dims
is defined below).O
in
[_
] =0
otherwise.
In
isO
in
+S
in
where + is element-wise addition.
remapped_offset_dims
is a monotonic function with domain [ 0
, offset.size
) and range [ 0
, operand.rank
) \ collapsed_slice_dims
. So if, eg, offset.size
is 4
, operand.rank
is 6
and collapsed_slice_dims
is { 0
, 2
} then remapped_offset_dims
is { 0
→ 1
, 1
→ 3
, 2
→ 4
, 3
→ 5
}.
If indices_are_sorted
is set to true then XLA can assume that start_indices
are sorted (in ascending start_index_map
order) by the user. If they are not then the semantics is implementation defined.
If unique_indices
is set to true then XLA can assume that all element scattered to are unique. So XLA could use non-atomic operations. If unique_indices
is set to true and the indices being scattered to are not unique then the semantics is implementation defined.
Informal Description and Examples
Informally, every index Out
in the output array corresponds to an element E
in the operand array, computed as follows:
We use the batch dimensions in
Out
to look up a starting index fromstart_indices
.We use
start_index_map
to map the starting index (whose size may be less than operand.rank) to a "full" starting index into theoperand
.We dynamic-slice out a slice with size
slice_sizes
using the full starting index.We reshape the slice by collapsing the
collapsed_slice_dims
dimensions. Since all collapsed slice dimensions must have a bound of 1, this reshape is always legal.We use the offset dimensions in
Out
to index into this slice to get the input element,E
, corresponding to output indexOut
.
index_vector_dim
is set to start_indices.rank
- 1
in all of the examples that follow. More interesting values for index_vector_dim
do not change the operation fundamentally, but make the visual representation more cumbersome.
To get an intuition on how all of the above fits together, let's look at an example that gathers 5 slices of shape [8,6]
from a [16,11]
array. The position of a slice into the [16,11]
array can be represented as an index vector of shape S64[2]
, so the set of 5 positions can be represented as a S64[5,2]
array.
The behavior of the gather operation can then be depicted as an index transformation that takes [ G
, O
0
, O
1
], an index in the output shape, and maps it to an element in the input array in the following way:
We first select an ( X
, Y
) vector from the gather indices array using G
. The element in the output array at index [ G
, O
0
, O
1
] is then the element in the input array at index [ X
+ O
0
, Y
+ O
1
].
slice_sizes
is [8,6]
, which decides the range of O 0
and O 1
, and this in turn decides the bounds of the slice.
This gather operation acts as a batch dynamic slice with G
as the batch dimension.
The gather indices may be multidimensional. For instance, a more general version of the example above using a "gather indices" array of shape [4,5,2]
would translate indices like this:
Again, this acts as a batch dynamic slice G
0
and G
1
as the batch dimensions. The slice size is still [8,6]
.
The gather operation in XLA generalizes the informal semantics outlined above in the following ways:
We can configure which dimensions in the output shape are the offset dimensions (dimensions containing
O
0
,O
1
in the last example). The output batch dimensions (dimensions containingG
0
,G
1
in the last example) are defined to be the output dimensions that are not offset dimensions.The number of output offset dimensions explicitly present in the output shape may be smaller than the input rank. These "missing" dimensions, which are listed explicitly as
collapsed_slice_dims
, must have a slice size of1
. Since they have a slice size of1
the only valid index for them is0
and eliding them does not introduce ambiguity.The slice extracted from the "Gather Indices" array ((
X
,Y
) in the last example) may have fewer elements than the input array rank, and an explicit mapping dictates how the index should be expanded to have the same rank as the input.
As a final example, we use (2) and (3) to implement tf.gather_nd
:
G
0
and G
1
are used to slice out a starting index from the gather indices array as usual, except the starting index has only one element, X
. Similarly, there is only one output offset index with the value O
0
. However, before being used as indices into the input array, these are expanded in accordance to "Gather Index Mapping" ( start_index_map
in the formal description) and "Offset Mapping" ( remapped_offset_dims
in the formal description) into [ X
, 0
] and [ 0
, O
0
] respectively, adding up to [ X
, O
0
]. In other words, the output index [ G
0
, G
1
, O
0
] maps to the input index [ GatherIndices
[ G
0
, G
1
, 0
], X
] which gives us the semantics for tf.gather_nd
.
slice_sizes
for this case is [1,11]
. Intuitively this means that every index X
in the gather indices array picks an entire row and the result is the concatenation of all these rows.
GetDimensionSize
See also XlaBuilder::GetDimensionSize
.
Returns the size of the given dimension of the operand. The operand must be array shaped.
GetDimensionSize(operand, dimension)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | n dimensional input array |
dimension | int64 | A value in the interval [0, n) that specifies the dimension |
SetDimensionSize
See also XlaBuilder::SetDimensionSize
.
Sets the dynamic size of XlaOp's given dimension. The operand must be array shaped.
SetDimensionSize(operand, size, dimension)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | n dimensional input array. |
size | XlaOp | int32 representing the runtime dynamic size. |
dimension | int64 | A value in the interval [0, n) that specifies the dimension. |
Pass through the operand as result, with dynamic dimension tracked by the compiler.
Padded values will be ignored by downstream reduction ops.
let v: f32[10] = f32[10]{1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
let five: s32 = 5;
let six: s32 = 6;
// Setting dynamic dimension size doesn't change the upper bound of the static
// shape.
let padded_v_five: f32[10] = set_dimension_size(v, five, /*dimension=*/0);
let padded_v_six: f32[10] = set_dimension_size(v, six, /*dimension=*/0);
// sum == 1 + 2 + 3 + 4 + 5
let sum:f32[] = reduce_sum(padded_v_five);
// product == 1 * 2 * 3 * 4 * 5
let product:f32[] = reduce_product(padded_v_five);
// Changing padding size will yield different result.
// sum == 1 + 2 + 3 + 4 + 5 + 6
let sum':f32[] = reduce_sum(padded_v_six);
GetTupleElement
See also XlaBuilder::GetTupleElement
.
Indexes into a tuple with a compile-time-constant value.
The value must be a compile-time-constant so that shape inference can determine the type of the resulting value.
This is analogous to std::get<int N>(t)
in C++. Conceptually:
let v: f32[10] = f32[10]{0, 1, 2, 3, 4, 5, 6, 7, 8, 9};
let s: s32 = 5;
let t: (f32[10], s32) = tuple(v, s);
let element_1: s32 = gettupleelement(t, 1); // Inferred shape matches s32.
See also tf.tuple
.
Infeed
See also XlaBuilder::Infeed
.
Infeed(shape)
Argument | Type | Semantics |
---|---|---|
shape | Shape | Shape of the data read from the Infeed interface. The layout field of the shape must be set to match the layout of the data sent to the device; otherwise its behavior is undefined. |
Reads a single data item from the implicit Infeed streaming interface of the device, interpreting the data as the given shape and its layout, and returns a XlaOp
of the data. Multiple Infeed operations are allowed in a computation, but there must be a total order among the Infeed operations. For example, two Infeeds in the code below have a total order since there is a dependency between the while loops.
result1 = while (condition, init = init_value) {
Infeed(shape)
}
result2 = while (condition, init = result1) {
Infeed(shape)
}
Nested tuple shapes are not supported. For an empty tuple shape, the Infeed operation is effectively a no-op and proceeds without reading any data from the Infeed of the device.
Iota
Iota()
Builds a constant literal on device rather than a potentially large host transfer. Creates a rank 1 array of values starting at zero and incrementing by one. For floating-point types, the produced array is equivalent to ConvertElementType(Iota(...))
where the Iota
is of integral type and the conversion is to the floating-point type.
Arguments | Type | Semantics |
---|---|---|
type | PrimitiveType | type U |
size | int64 | The number of elements in the array. |
iota_dimension | int64 | The dimension to increment along. |
Map
See also XlaBuilder::Map
.
Map(operands..., computation)
Arguments | Type | Semantics |
---|---|---|
operands | sequence of N XlaOp s | N arrays of types T 0..T {N-1} |
computation | XlaComputation | computation of type T_0, T_1, ..., T_{N + M -1} -> S with N parameters of type T and M of arbitrary type |
dimensions | int64 array | array of map dimensions |
Applies a scalar function over the given operands
arrays, producing an array of the same dimensions where each element is the result of the mapped function applied to the corresponding elements in the input arrays.
The mapped function is an arbitrary computation with the restriction that it has N inputs of scalar type T
and a single output with type S
. The output has the same dimensions as the operands except that the element type T is replaced with S.
For example: Map(op1, op2, op3, computation, par1)
maps elem_out <- computation(elem1, elem2, elem3, par1)
at each (multi-dimensional) index in the input arrays to produce the output array.
Pad
See also XlaBuilder::Pad
.
Pad(operand, padding_value, padding_config)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | array of type T |
padding_value | XlaOp | scalar of type T to fill in the added padding |
padding_config | PaddingConfig | padding amount on both edges (low, high) and between the elements of each dimension |
Expands the given operand
array by padding around the array as well as between the elements of the array with the given padding_value
. padding_config
specifies the amount of edge padding and the interior padding for each dimension.
PaddingConfig
is a repeated field of PaddingConfigDimension
, which contains three fields for each dimension: edge_padding_low
, edge_padding_high
, and interior_padding
.
edge_padding_low
and edge_padding_high
specify the amount of padding added at the low-end (next to index 0) and the high-end (next to the highest index) of each dimension respectively. The amount of edge padding can be negative -- the absolute value of negative padding indicates the number of elements to remove from the specified dimension.
interior_padding
specifies the amount of padding added between any two elements in each dimension; it may not be negative. Interior padding occurs logically before edge padding, so in the case of negative edge padding, elements are removed from the interior-padded operand.
This operation is a no-op if the edge padding pairs are all (0, 0) and the interior padding values are all 0. The figure below shows examples of different edge_padding
and interior_padding
values for a two-dimensional array.

Recv
See also XlaBuilder::Recv
.
Recv(shape, channel_handle)
Arguments | Type | Semantics |
---|---|---|
shape | Shape | shape of the data to receive |
channel_handle | ChannelHandle | unique identifier for each send/recv pair |
Receives data of the given shape from a Send
instruction in another computation that shares the same channel handle. Returns a XlaOp for the received data.
The client API of Recv
operation represents synchronous communication. However, the instruction is internally decomposed into 2 HLO instructions ( Recv
and RecvDone
) to enable asynchronous data transfers. See also HloInstruction::CreateRecv
and HloInstruction::CreateRecvDone
.
Recv(const Shape& shape, int64 channel_id)
Allocates resources required to receive data from a Send
instruction with the same channel_id. Returns a context for the allocated resources, which is used by a following RecvDone
instruction to wait for the completion of the data transfer. The context is a tuple of {receive buffer (shape), request identifier (U32)} and it can only be used by a RecvDone
instruction.
RecvDone(HloInstruction context)
Given a context created by a Recv
instruction, waits for the data transfer to complete and returns the received data.
Reduce
See also XlaBuilder::Reduce
.
Applies a reduction function to one or more arrays in parallel.
Reduce(operands..., init_values..., computation, dimensions)
Arguments | Type | Semantics |
---|---|---|
operands | Sequence of N XlaOp | N arrays of types T_0, ..., T_{N-1} . |
init_values | Sequence of N XlaOp | N scalars of types T_0, ..., T_{N-1} . |
computation | XlaComputation | computation of type T_0, ..., T_{N-1}, T_0, ..., T_{N-1} -> Collate(T_0, ..., T_{N-1}) . |
dimensions | int64 array | unordered array of dimensions to reduce. |
Where:
- N is required to be greater or equal to 1.
- All input arrays must have the same dimensions.
- If
N = 1
,Collate(T)
isT
. - If
N > 1
,Collate(T_0, ..., T_{N-1})
is a tuple ofN
elements of typeT
.
The output of the op is Collate(Q_0, ..., Q_N)
where Q_i
is an array of type T_i
, the dimensions of which are described below.
This operation reduces one or more dimensions of each input array into scalars. The rank of each returned array is rank(operand) - len(dimensions)
. The initial value used for every reduction is init_value
, and it may be inserted anywhere during computation by the back-end. In most cases, init_value
is an identity of the reduction function (for example, 0
for addition). The applied computation
is always passed the init_value
on the left-hand side.
The evaluation order of the reduction function is arbitrary and may be non-deterministic. Therefore, the reduction function should not be overly sensitive to reassociation.
Some reduction functions like addition are not strictly associative for floats. However, if the range of the data is limited, floating-point addition is close enough to being associative for most practical uses. It is possible to conceive of some completely non-associative reductions, however, and these will produce incorrect or unpredictable results in XLA.
As an example, when reducing across one dimension in a single 1D array with values [10, 11, 12, 13]
, with reduction function f
(this is computation
) then that could be computed as
f(10, f(11, f(12, f(init_value, 13)))
but there are also many other possibilities, eg
f(init_value, f(f(10, f(init_value, 11)), f(f(init_value, 12), f(init_value, 13))))
The following is a rough pseudo-code example of how reduction could be implemented, using summation as the reduction computation with an initial value of 0.
result_shape <- remove all dims in dimensions from operand_shape
# Iterate over all elements in result_shape. The number of r's here is equal
# to the rank of the result
for r0 in range(result_shape[0]), r1 in range(result_shape[1]), ...:
# Initialize this result element
result[r0, r1...] <- 0
# Iterate over all the reduction dimensions
for d0 in range(dimensions[0]), d1 in range(dimensions[1]), ...:
# Increment the result element with the value of the operand's element.
# The index of the operand's element is constructed from all ri's and di's
# in the right order (by construction ri's and di's together index over the
# whole operand shape).
result[r0, r1...] += operand[ri... di]
Here's an example of reducing a 2D array (matrix). The shape has rank 2, dimension 0 of size 2 and dimension 1 of size 3:

Results of reducing dimensions 0 or 1 with an "add" function:

Note that both reduction results are 1D arrays. The diagram shows one as column and another as row just for visual convenience.
For a more complex example, here is a 3D array. Its rank is 3, dimension 0 of size 4, dimension 1 of size 2 and dimension 2 of size 3. For simplicity, the values 1 to 6 are replicated across dimension 0.

Similarly to the 2D example, we can reduce just one dimension. If we reduce dimension 0, for example, we get a rank-2 array where all values across dimension 0 were folded into a scalar:
| 4 8 12 |
| 16 20 24 |
If we reduce dimension 2, we also get a rank-2 array where all values across dimension 2 were folded into a scalar:
| 6 15 |
| 6 15 |
| 6 15 |
| 6 15 |
Note that the relative order between the remaining dimensions in the input is preserved in the output, but some dimensions may get assigned new numbers (since the rank changes).
We can also reduce multiple dimensions. Add-reducing dimensions 0 and 1 produces the 1D array [20, 28, 36]
.
Reducing the 3D array over all its dimensions produces the scalar 84
.
Variadic Reduce
When N > 1
, reduce function application is slightly more complex, as it is applied simultaneously to all inputs. The operands are supplied to the computation in the following order:
- Running reduced value for the first operand
- ...
- Running reduced value for the N'th operand
- Input value for the first operand
- ...
- Input value for the N'th operand
For example, consider the following reduction function, which can be used to compute the max and the argmax of a 1-D array in parallel:
f: (Float, Int, Float, Int) -> Float, Int
f(max, argmax, value, index):
if value >= max:
return (value, index)
else:
return (max, argmax)
For 1-D Input arrays V = Float[N], K = Int[N]
, and init values I_V = Float, I_K = Int
, the result f_(N-1)
of reducing across the only input dimension is equivalent to the following recursive application:
f_0 = f(I_V, I_K, V_0, K_0)
f_1 = f(f_0.first, f_0.second, V_1, K_1)
...
f_(N-1) = f(f_(N-2).first, f_(N-2).second, V_(N-1), K_(N-1))
Applying this reduction to an array of values, and an array of sequential indices (ie iota), will co-iterate over the arrays, and return a tuple containing the maximal value and the matching index.
ReducePrecision
See also XlaBuilder::ReducePrecision
.
Models the effect of converting floating-point values to a lower-precision format (such as IEEE-FP16) and back to the original format. The number of exponent and mantissa bits in the lower-precision format can be specified arbitrarily, although all bit sizes may not be supported on all hardware implementations.
ReducePrecision(operand, mantissa_bits, exponent_bits)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | array of floating-point type T . |
exponent_bits | int32 | number of exponent bits in lower-precision format |
mantissa_bits | int32 | number of mantissa bits in lower-precision format |
The result is an array of type T
. The input values are rounded to the nearest value representable with the given number of mantissa bits (using "ties to even" semantics), and any values that exceed the range specified by the number of exponent bits are clamped to positive or negative infinity. NaN
values are retained, although they may be converted to canonical NaN
values.
The lower-precision format must have at least one exponent bit (in order to distinguish a zero value from an infinity, since both have a zero mantissa), and must have a non-negative number of mantissa bits. The number of exponent or mantissa bits may exceed the corresponding value for type T
; the corresponding portion of the conversion is then simply a no-op.
ReduceWindow
See also XlaBuilder::ReduceWindow
.
Applies a reduction function to all elements in each window of the input multi-dimensional array, producing an output multi-dimensional array with the same number of elements as the number of valid positions of the window. A pooling layer can be expressed as a ReduceWindow
. Similar to Reduce
, the applied computation
is always passed the init_value
on the left-hand side.
ReduceWindow(operand, init_value, computation, window_dimensions, window_strides, padding)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | N dimensional array containing elements of type T. This is the base area on which the window is placed. |
init_value | XlaOp | Starting value for the reduction. See Reduce for details. |
computation | XlaComputation | Reduction function of type T, T -> T , to apply to all elements in each window |
window_dimensions | ArraySlice<int64> | array of integers for window dimension values |
window_strides | ArraySlice<int64> | array of integers for window stride values |
base_dilations | ArraySlice<int64> | array of integers for base dilation values |
window_dilations | ArraySlice<int64> | array of integers for window dilation values |
padding | Padding | padding type for window (Padding::kSame, which pads so as to have the same output shape as input if the stride is 1, or Padding::kValid, which uses no padding and "stops" the window once it no longer fits) |
Below code and figure shows an example of using ReduceWindow
. Input is a matrix of size [4x6] and both window_dimensions and window_stride_dimensions are [2x3].
// Create a computation for the reduction (maximum).
XlaComputation max;
{
XlaBuilder builder(client_, "max");
auto y = builder.Parameter(0, ShapeUtil::MakeShape(F32, {}), "y");
auto x = builder.Parameter(1, ShapeUtil::MakeShape(F32, {}), "x");
builder.Max(y, x);
max = builder.Build().ConsumeValueOrDie();
}
// Create a ReduceWindow computation with the max reduction computation.
XlaBuilder builder(client_, "reduce_window_2x3");
auto shape = ShapeUtil::MakeShape(F32, {4, 6});
auto input = builder.Parameter(0, shape, "input");
builder.ReduceWindow(
input,
/*init_val=*/builder.ConstantLiteral(LiteralUtil::MinValue(F32)),
*max,
/*window_dimensions=*/{2, 3},
/*window_stride_dimensions=*/{2, 3},
Padding::kValid);

Stride of 1 in a dimension specifies that the position of a window in the dimension is 1 element away from its adjacent window. In order to specify that no windows overlap with each other, window_stride_dimensions should be equal to window_dimensions. The figure below illustrates the use of two different stride values. Padding is applied to each dimension of the input and the calculations are the same as though the input came in with the dimensions it has after padding.

For a non-trivial padding example, consider computing reduce-window minimum (initial value is MAX_FLOAT
) with dimension 3
and stride 2
over the input array [10000, 1000, 100, 10, 1]
. Padding kValid
computes minimums over two valid windows: [10000, 1000, 100]
and [100, 10, 1]
, resulting in the output [100, 1]
. Padding kSame
first pads the array so that the shape after the reduce-window would be the same as input for stride one by adding initial elements on both sides, getting [MAX_VALUE, 10000, 1000, 100, 10, 1, MAX_VALUE]
. Running reduce-window over the padded array operates on three windows [MAX_VALUE, 10000, 1000]
, [1000, 100, 10]
, [10, 1, MAX_VALUE]
, and yields [1000, 10, 1]
.
The evaluation order of the reduction function is arbitrary and may be non-deterministic. Therefore, the reduction function should not be overly sensitive to reassociation. See the discussion about associativity in the context of Reduce
for more details.
ReplicaId
See also XlaBuilder::ReplicaId
.
Returns the unique ID (U32 scalar) of the replica.
ReplicaId()
The unique ID of each replica is an unsigned integer in the interval [0, N)
, where N
is the number of replicas. Since all the replicas are running the same program, a ReplicaId()
call in the program will return a different value on each replica.
Reshape
See also XlaBuilder::Reshape
and the Collapse
operation.
Reshapes the dimensions of an array into a new configuration.
Reshape(operand, new_sizes)
Reshape(operand, dimensions, new_sizes)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | array of type T |
dimensions | int64 vector | order in which dimensions are collapsed |
new_sizes | int64 vector | vector of sizes of new dimensions |
Conceptually, reshape first flattens an array into a one-dimensional vector of data values, and then refines this vector into a new shape. The input arguments are an arbitrary array of type T, a compile-time-constant vector of dimension indices, and a compile-time-constant vector of dimension sizes for the result. The values in the dimension
vector, if given, must be a permutation of all of T's dimensions; the default if not given is {0, ..., rank - 1}
. The order of the dimensions in dimensions
is from slowest-varying dimension (most major) to fastest-varying dimension (most minor) in the loop nest which collapses the input array into a single dimension. The new_sizes
vector determines the size of the output array. The value at index 0 in new_sizes
is the size of dimension 0, the value at index 1 is the size of dimension 1, and so on. The product of the new_size
dimensions must equal the product of the operand's dimension sizes. When refining the collapsed array into the multidimensional array defined by new_sizes
, the dimensions in new_sizes
are ordered from slowest varying (most major) and to fastest varying (most minor).
For example, let v be an array of 24 elements:
let v = f32[4x2x3] { { {10, 11, 12}, {15, 16, 17} },
{ {20, 21, 22}, {25, 26, 27} },
{ {30, 31, 32}, {35, 36, 37} },
{ {40, 41, 42}, {45, 46, 47} } };
In-order collapse:
let v012_24 = Reshape(v, {0,1,2}, {24});
then v012_24 == f32[24] {10, 11, 12, 15, 16, 17, 20, 21, 22, 25, 26, 27,
30, 31, 32, 35, 36, 37, 40, 41, 42, 45, 46, 47};
let v012_83 = Reshape(v, {0,1,2}, {8,3});
then v012_83 == f32[8x3] { {10, 11, 12}, {15, 16, 17},
{20, 21, 22}, {25, 26, 27},
{30, 31, 32}, {35, 36, 37},
{40, 41, 42}, {45, 46, 47} };
Out-of-order collapse:
let v021_24 = Reshape(v, {1,2,0}, {24});
then v012_24 == f32[24] {10, 20, 30, 40, 11, 21, 31, 41, 12, 22, 32, 42,
15, 25, 35, 45, 16, 26, 36, 46, 17, 27, 37, 47};
let v021_83 = Reshape(v, {1,2,0}, {8,3});
then v021_83 == f32[8x3] { {10, 20, 30}, {40, 11, 21},
{31, 41, 12}, {22, 32, 42},
{15, 25, 35}, {45, 16, 26},
{36, 46, 17}, {27, 37, 47} };
let v021_262 = Reshape(v, {1,2,0}, {2,6,2});
then v021_262 == f32[2x6x2] { { {10, 20}, {30, 40},
{11, 21}, {31, 41},
{12, 22}, {32, 42} },
{ {15, 25}, {35, 45},
{16, 26}, {36, 46},
{17, 27}, {37, 47} } };
As a special case, reshape can transform a single-element array to a scalar and vice versa. For example,
Reshape(f32[1x1] { {5} }, {0,1}, {}) == 5;
Reshape(5, {}, {1,1}) == f32[1x1] { {5} };
Rev (reverse)
See also XlaBuilder::Rev
.
Rev(operand, dimensions)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | array of type T |
dimensions | ArraySlice<int64> | dimensions to reverse |
Reverses the order of elements in the operand
array along the specified dimensions
, generating an output array of the same shape. Each element of the operand array at a multidimensional index is stored into the output array at a transformed index. The multidimensional index is transformed by reversing the index in each dimension to be reversed (ie, if a dimension of size N is one of the reversing dimensions, its index i is transformed into N - 1 - i).
One use for the Rev
operation is to reverse the convolution weight array along the two window dimensions during the gradient computation in neural networks.
RngNormal
See also XlaBuilder::RngNormal
.
Constructs an output of a given shape with random numbers generated following the
normal distribution. The parameters
and
, and output shape have to have a floating point elemental type. The parameters furthermore have to be scalar valued.
RngNormal(mu, sigma, shape)
Arguments | Type | Semantics |
---|---|---|
mu | XlaOp | Scalar of type T specifying mean of generated numbers |
sigma | XlaOp | Scalar of type T specifying standard deviation of generated numbers |
shape | Shape | Output shape of type T |
RngUniform
See also XlaBuilder::RngUniform
.
Constructs an output of a given shape with random numbers generated following the uniform distribution over the interval
. The parameters and output element type have to be a boolean type, an integral type or a floating point types, and the types have to be consistent. The CPU and GPU backends currently only support F64, F32, F16, BF16, S64, U64, S32 and U32. Furthermore, the parameters need to be scalar valued. If
the result is implementation-defined.
RngUniform(a, b, shape)
Arguments | Type | Semantics |
---|---|---|
a | XlaOp | Scalar of type T specifying lower limit of interval |
b | XlaOp | Scalar of type T specifying upper limit of interval |
shape | Shape | Output shape of type T |
RngBitGenerator
Generates an output with a given shape filled with uniform random bits using the specified algorithm (or backend default) and returns an updated state (with the same shape as initial state) and the generated random data.
Initial state is the initial state of the current random number generation. It and the required shape and valid values are dependent on the algorithm used.
The output is guaranteed to be a deterministic function of the initial state but it is not guaranteed to be deterministic between backends and different compiler versions.
RngBitGenerator(algorithm, key, shape)
Arguments | Type | Semantics |
---|---|---|
algorithm | RandomAlgorithm | PRNG algorithm to be used. |
initial_state | XlaOp | Initial state for the PRNG algorithm. |
shape | Shape | Output shape for generated data. |
Available values for algorithm
:
rng_default
: Backend specific algorithm with backend specific shape requirements.rng_three_fry
: ThreeFry counter-based PRNG algorithm. Theinitial_state
shape isu64[2]
with arbitrary values. Salmon et al. SC 2011. Parallel random numbers: as easy as 1, 2, 3.rng_philox
: Philox algorithm to generate random numbers in parallel. Theinitial_state
shape isu64[3]
with arbitrary values. Salmon et al. SC 2011. Parallel random numbers: as easy as 1, 2, 3.
Scatter
The XLA scatter operation generates a result which is the value of the input array operand
, with several slices (at indices specified by scatter_indices
) updated with the values in updates
using update_computation
.
See also XlaBuilder::Scatter
.
scatter(operand, scatter_indices, updates, update_computation, index_vector_dim, update_window_dims, inserted_window_dims, scatter_dims_to_operand_dims)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | Array to be scattered into. |
scatter_indices | XlaOp | Array containing the starting indices of the slices that must be scattered to. |
updates | XlaOp | Array containing the values that must be used for scattering. |
update_computation | XlaComputation | Computation to be used for combining the existing values in the input array and the updates during scatter. This computation should be of type (T, T) -> T . |
index_vector_dim | int64 | The dimension in scatter_indices that contains the starting indices. |
update_window_dims | ArraySlice<int64> | The set of dimensions in updates shape that are window dimensions . |
inserted_window_dims | ArraySlice<int64> | The set of window dimensions that must be inserted into updates shape. |
scatter_dims_to_operand_dims | ArraySlice<int64> | A dimensions map from the scatter indices to the operand index space. This array is interpreted as mapping i to scatter_dims_to_operand_dims[i] . It has to be one-to-one and total. |
indices_are_sorted | bool | Whether the indices are guaranteed to be sorted by the caller. |
If index_vector_dim
is equal to scatter_indices.rank
we implicitly consider scatter_indices
to have a trailing 1
dimension.
We define update_scatter_dims
of type ArraySlice<int64>
as the set of dimensions in updates
shape that are not in update_window_dims
, in ascending order.
The arguments of scatter should follow these constraints:
updates
array must be of rankupdate_window_dims.size + scatter_indices.rank - 1
.Bounds of dimension
i
inupdates
must conform to the following:- If
i
is present inupdate_window_dims
(ie equal toupdate_window_dims
[k
] for somek
), then the bound of dimensioni
inupdates
must not exceed the corresponding bound ofoperand
after accounting for theinserted_window_dims
(ieadjusted_window_bounds
[k
], whereadjusted_window_bounds
contains the bounds ofoperand
with the bounds at indicesinserted_window_dims
removed). - If
i
is present inupdate_scatter_dims
(ie equal toupdate_scatter_dims
[k
] for somek
), then the bound of dimensioni
inupdates
must be equal to the corresponding bound ofscatter_indices
, skippingindex_vector_dim
(iescatter_indices.shape.dims
[k
], ifk
<index_vector_dim
andscatter_indices.shape.dims
[k+1
] otherwise).
- If
update_window_dims
must be in ascending order, not have any repeating dimension numbers, and be in the range[0, updates.rank)
.inserted_window_dims
must be in ascending order, not have any repeating dimension numbers, and be in the range[0, operand.rank)
.operand.rank
must equal the sum ofupdate_window_dims.size
andinserted_window_dims.size
.scatter_dims_to_operand_dims.size
must be equal toscatter_indices
[index_vector_dim
], and its values must be in the range[0, operand.rank)
.
For a given index U
in the updates
array, the corresponding index I
in the operand
array into which this update has to be applied is computed as follows:
- Let
G
= {U
[k
] fork
inupdate_scatter_dims
}. UseG
to look up an index vectorS
in thescatter_indices
array such thatS
[i
] =scatter_indices
[Combine(G
,i
)] where Combine(A, b) inserts b at positionsindex_vector_dim
into A. - Create an index
S
in
intooperand
usingS
by scatteringS
using thescatter_dims_to_operand_dims
map. More formally:-
S
in
[scatter_dims_to_operand_dims
[k
]] =S
[k
] ifk
<scatter_dims_to_operand_dims.size
. -
S
in
[_
] =0
otherwise.
-
- Create an index
W
in
intooperand
by scattering the indices atupdate_window_dims
inU
according toinserted_window_dims
. More formally:-
W
in
[window_dims_to_operand_dims
(k
)] =U
[k
] ifk
is inupdate_window_dims
, wherewindow_dims_to_operand_dims
is the monotonic function with domain [0
,update_window_dims.size
) and range [0
,operand.rank
) \inserted_window_dims
. (For example, ifupdate_window_dims.size
is4
,operand.rank
is6
, andinserted_window_dims
is {0
,2
} thenwindow_dims_to_operand_dims
is {0
→1
,1
→3
,2
→4
,3
→5
}). -
W
in
[_
] =0
otherwise.
-
-
I
isW
in
+S
in
where + is element-wise addition.
In summary, the scatter operation can be defined as follows.
- Initialize
output
withoperand
, ie for all indicesO
in theoperand
array:
output
[O
] =operand
[O
] - For every index
U
in theupdates
array and the corresponding indexO
in theoperand
array, ifO
is a valid index foroutput
:
output
[O
] =update_computation
(output
[O
],updates
[U
])
The order in which updates are applied is non-deterministic. So, when multiple indices in updates
refer to the same index in operand
, the corresponding value in output
will be non-deterministic.
Note that the first parameter that is passed into the update_computation
will always be the current value from the output
array and the second parameter will always be the value from the updates
array. This is important specifically for cases when the update_computation
is not commutative .
If indices_are_sorted
is set to true then XLA can assume that start_indices
are sorted (in ascending start_index_map
order) by the user. If they are not then the semantics is implementation defined.
Informally, the scatter op can be viewed as an inverse of the gather op, ie the scatter op updates the elements in the input that are extracted by the corresponding gather op.
For a detailed informal description and examples, refer to the "Informal Description" section under Gather
.
Select
See also XlaBuilder::Select
.
Constructs an output array from elements of two input arrays, based on the values of a predicate array.
Select(pred, on_true, on_false)
Arguments | Type | Semantics |
---|---|---|
pred | XlaOp | array of type PRED |
on_true | XlaOp | array of type T |
on_false | XlaOp | array of type T |
The arrays on_true
and on_false
must have the same shape. This is also the shape of the output array. The array pred
must have the same dimensionality as on_true
and on_false
, with the PRED
element type.
For each element P
of pred
, the corresponding element of the output array is taken from on_true
if the value of P
is true
, and from on_false
if the value of P
is false
. As a restricted form of broadcasting , pred
can be a scalar of type PRED
. In this case, the output array is taken wholly from on_true
if pred
is true
, and from on_false
if pred
is false
.
Example with non-scalar pred
:
let pred: PRED[4] = {true, false, false, true};
let v1: s32[4] = {1, 2, 3, 4};
let v2: s32[4] = {100, 200, 300, 400};
==>
Select(pred, v1, v2) = s32[4]{1, 200, 300, 4};
Example with scalar pred
:
let pred: PRED = true;
let v1: s32[4] = {1, 2, 3, 4};
let v2: s32[4] = {100, 200, 300, 400};
==>
Select(pred, v1, v2) = s32[4]{1, 2, 3, 4};
Selections between tuples are supported. Tuples are considered to be scalar types for this purpose. If on_true
and on_false
are tuples (which must have the same shape!) then pred
has to be a scalar of type PRED
.
SelectAndScatter
See also XlaBuilder::SelectAndScatter
.
This operation can be considered as a composite operation that first computes ReduceWindow
on the operand
array to select an element from each window, and then scatters the source
array to the indices of the selected elements to construct an output array with the same shape as the operand array. The binary select
function is used to select an element from each window by applying it across each window, and it is called with the property that the first parameter's index vector is lexicographically less than the second parameter's index vector. The select
function returns true
if the first parameter is selected and returns false
if the second parameter is selected, and the function must hold transitivity (ie, if select(a, b)
and select(b, c)
are true
, then select(a, c)
is also true
) so that the selected element does not depend on the order of the elements traversed for a given window.
The function scatter
is applied at each selected index in the output array. It takes two scalar parameters:
- Current value at the selected index in the output array
- The scatter value from
source
that applies to the selected index
It combines the two parameters and returns a scalar value that's used to update the value at the selected index in the output array. Initially, all indices of the output array are set to init_value
.
The output array has the same shape as the operand
array and the source
array must have the same shape as the result of applying a ReduceWindow
operation on the operand
array. SelectAndScatter
can be used to backpropagate the gradient values for a pooling layer in a neural network.
SelectAndScatter(operand, select, window_dimensions, window_strides, padding, source, init_value, scatter)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | array of type T over which the windows slide |
select | XlaComputation | binary computation of type T, T -> PRED , to apply to all elements in each window; returns true if the first parameter is selected and returns false if the second parameter is selected |
window_dimensions | ArraySlice<int64> | array of integers for window dimension values |
window_strides | ArraySlice<int64> | array of integers for window stride values |
padding | Padding | padding type for window (Padding::kSame or Padding::kValid) |
source | XlaOp | array of type T with the values to scatter |
init_value | XlaOp | scalar value of type T for the initial value of the output array |
scatter | XlaComputation | binary computation of type T, T -> T , to apply each scatter source element with its destination element |
The figure below shows examples of using SelectAndScatter
, with the select
function computing the maximal value among its parameters. Note that when the windows overlap, as in the figure (2) below, an index of the operand
array may be selected multiple times by different windows. In the figure, the element of value 9 is selected by both of the top windows (blue and red) and the binary addition scatter
function produces the output element of value 8 (2 + 6).

The evaluation order of the scatter
function is arbitrary and may be non-deterministic. Therefore, the scatter
function should not be overly sensitive to reassociation. See the discussion about associativity in the context of Reduce
for more details.
Send
See also XlaBuilder::Send
.
Send(operand, channel_handle)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | data to send (array of type T) |
channel_handle | ChannelHandle | unique identifier for each send/recv pair |
Sends the given operand data to a Recv
instruction in another computation that shares the same channel handle. Does not return any data.
Similar to the Recv
operation, the client API of Send
operation represents synchronous communication, and is internally decomposed into 2 HLO instructions ( Send
and SendDone
) to enable asynchronous data transfers. See also HloInstruction::CreateSend
and HloInstruction::CreateSendDone
.
Send(HloInstruction operand, int64 channel_id)
Initiates an asynchronous transfer of the operand to the resources allocated by the Recv
instruction with the same channel id. Returns a context, which is used by a following SendDone
instruction to wait for the completion of the data transfer. The context is a tuple of {operand (shape), request identifier (U32)} and it can only be used by a SendDone
instruction.
SendDone(HloInstruction context)
Given a context created by a Send
instruction, waits for the data transfer to complete. The instruction does not return any data.
Scheduling of channel instructions
The execution order of the 4 instructions for each channel ( Recv
, RecvDone
, Send
, SendDone
) is as below.

-
Recv
happens beforeSend
-
Send
happens beforeRecvDone
-
Recv
happens beforeRecvDone
-
Send
happens beforeSendDone
When the backend compilers generate a linear schedule for each computation that communicates via channel instructions, there must not be cycles across the computations. For example, below schedules lead to deadlocks.

Slice
See also XlaBuilder::Slice
.
Slicing extracts a sub-array from the input array. The sub-array is of the same rank as the input and contains the values inside a bounding box within the input array where the dimensions and indices of the bounding box are given as arguments to the slice operation.
Slice(operand, start_indices, limit_indices)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | N dimensional array of type T |
start_indices | ArraySlice<int64> | List of N integers containing the starting indices of the slice for each dimension. Values must be greater than or equal to zero. |
limit_indices | ArraySlice<int64> | List of N integers containing the ending indices (exclusive) for the slice for each dimension. Each value must be greater than or equal to the respective start_indices value for the dimension and less than or equal to the size of the dimension. |
strides | ArraySlice<int64> | List of N integers that decides the input stride of the slice. The slice picks every strides[d] element in dimension d . |
1-dimensional example:
let a = {0.0, 1.0, 2.0, 3.0, 4.0}
Slice(a, {2}, {4}) produces:
{2.0, 3.0}
2-dimensional example:
let b =
{ {0.0, 1.0, 2.0},
{3.0, 4.0, 5.0},
{6.0, 7.0, 8.0},
{9.0, 10.0, 11.0} }
Slice(b, {2, 1}, {4, 3}) produces:
{ { 7.0, 8.0},
{10.0, 11.0} }
Sort
See also XlaBuilder::Sort
.
Sort(operands, comparator, dimension, is_stable)
Arguments | Type | Semantics |
---|---|---|
operands | ArraySlice<XlaOp> | The operands to sort. |
comparator | XlaComputation | The comparator computation to use. |
dimension | int64 | The dimension along which to sort. |
is_stable | bool | Whether stable sorting should be used. |
If only one operand is provided:
If the operand is a rank-1 tensor (an array), the result is a sorted array. If you want to sort the array into ascending order, the comparator should perform a less-than comparison. Formally, after the array is sorted, it holds for all index positions
i, j
withi < j
that eithercomparator(value[i], value[j]) = comparator(value[j], value[i]) = false
orcomparator(value[i], value[j]) = true
.If the operand has higher rank, the operand is sorted along the provided dimension. For example, for a rank-2 tensor (a matrix), a dimension value of
0
will independently sort every column, and a dimension value of1
will independently sort each row. If no dimension number is provided, then the last dimension is chosen by default. For the dimension which is sorted, the same sorting order applies as in the rank-1 case.
If n > 1
operands are provided:
All
n
operands must be tensors with the same dimensions. The element types of the tensors may be different.All operands are sorted together, not individually. Conceptually the operands are treated as a tuple. When checking whether the elements of each operand at index positions
i
andj
need to be swapped, the comparator is called with2 * n
scalar parameters, where parameter2 * k
corresponds to the value at positioni
from thek-th
operand, and parameter2 * k + 1
corresponds to the value at positionj
from thek-th
operand. Usually, the comparator would thus compare parameters2 * k
and2 * k + 1
with each other and possibly use other parameter pairs as tie breakers.The result is a tuple that consists of the operands in sorted order (along the provided dimension, as above). The
i-th
operand of the tuple corresponds to thei-th
operand of Sort.
For example, if there are three operands operand0 = [3, 1]
, operand1 = [42, 50]
, operand2 = [-3.0, 1.1]
, and the comparator compares only the values of operand0
with less-than, then the output of the sort is the tuple ([1, 3], [50, 42], [1.1, -3.0])
.
If is_stable
is set to true, the sort is guaranteed to be stable, that is, if there are elements which are considered to be equal by the comparator, the relative order of the equal values is preserved. By default, is_stable
is set to false.
Transpose
See also the tf.reshape
operation.
Transpose(operand)
Arguments | Type | Semantics |
---|---|---|
operand | XlaOp | The operand to transpose. |
permutation | ArraySlice<int64> | How to permute the dimensions. |
Permutes the operand dimensions with the given permutation, so ∀ i . 0 ≤ i < rank ⇒ input_dimensions[permutation[i]] = output_dimensions[i]
.
This is the same as Reshape(operand, permutation, Permute(permutation, operand.shape.dimensions)).
TriangularSolve
See also XlaBuilder::TriangularSolve
.
Solves systems of linear equations with lower or upper triangular coefficient matrices by forward- or back-substitution. Broadcasting along leading dimensions, this routine solves one of the matrix systems op(a) * x = b
, or x * op(a) = b
, for the variable x
, given a
and b
, where op(a)
is either op(a) = a
, or op(a) = Transpose(a)
, or op(a) = Conj(Transpose(a))
.
TriangularSolve(a, b, left_side, lower, unit_diagonal, transpose_a)
Arguments | Type | Semantics |
---|---|---|
a | XlaOp | a rank > 2 array of a complex or floating-point type with shape [..., M, M] . |
b | XlaOp | a rank > 2 array of the same type with shape [..., M, K] if left_side is true, [..., K, M] otherwise. |
left_side | bool | indicates whether to solve a system of the form op(a) * x = b ( true ) or x * op(a) = b ( false ). |
lower | bool | whether to use the upper or lower triangle of a . |
unit_diagonal | bool | if true , the diagonal elements of a are assumed to be 1 and not accessed. |
transpose_a | Transpose | whether to use a as is, transpose it or take its conjugate transpose. |
Input data is read only from the lower/upper triangle of a
, depending on the value of lower
. Values from the other triangle are ignored. Output data is returned in the same triangle; the values in the other triangle are implementation-defined and may be anything.
If the rank of a
and b
are greater than 2, they are treated as batches of matrices, where all except the minor 2 dimensions are batch dimensions. a
and b
must have equal batch dimensions.
Tuple
See also XlaBuilder::Tuple
.
A tuple containing a variable number of data handles, each of which has its own shape.
This is analogous to std::tuple
in C++. Conceptually:
let v: f32[10] = f32[10]{0, 1, 2, 3, 4, 5, 6, 7, 8, 9};
let s: s32 = 5;
let t: (f32[10], s32) = tuple(v, s);
Tuples can be deconstructed (accessed) via the GetTupleElement
operation.
While
See also XlaBuilder::While
.
While(condition, body, init)
Arguments | Type | Semantics |
---|---|---|
condition | XlaComputation | XlaComputation of type T -> PRED which defines the termination condition of the loop. |
body | XlaComputation | XlaComputation of type T -> T which defines the body of the loop. |
init | T | Initial value for the parameter of condition and body . |
Sequentially executes the body
until the condition
fails. This is similar to a typical while loop in many other languages except for the differences and restrictions listed below.
- A
While
node returns a value of typeT
, which is the result from the last execution of thebody
. - The shape of the type
T
is statically determined and must be the same across all iterations.
The T parameters of the computations are initialized with the init
value in the first iteration and are automatically updated to the new result from body
in each subsequent iteration.
One main use case of the While
node is to implement the repeated execution of training in neural networks. Simplified pseudocode is shown below with a graph that represents the computation. The code can be found in while_test.cc
. The type T
in this example is a Tuple
consisting of an int32
for the iteration count and a vector[10]
for the accumulator. For 1000 iterations, the loop keeps adding a constant vector to the accumulator.
// Pseudocode for the computation.
init = {0, zero_vector[10]} // Tuple of int32 and float[10].
result = init;
while (result(0) < 1000) {
iteration = result(0) + 1;
new_vector = result(1) + constant_vector[10];
result = {iteration, new_vector};
}
