Force tensor evaluation inside while-loop, scan and others. #13616

awav · 2017-10-10T22:21:20Z

Hello everyone,

I had big plans for tf.while_loop until I discovered that it is impossible to re-evaluate tensor inside it. Let's dive into the the issue and potential useful feature:

In [13]: import tensorflow as tf
    ...: import numpy as np
    ...:
    ...: def cond(i, _x, _sq):
    ...:       return tf.less(i, 10)
    ...:
    ...: def gen_body(v):
    ...:     def body(i, x, sq):
    ...:         x = tf.Print(x, [x, sq], "x and sq: ")
    ...:         v_assign = v.assign(x + 1)
    ...:         v_assign = tf.Print(v_assign, [v_assign], "v_assign: ")
    ...:         with tf.control_dependencies([v_assign]):
    ...:             sq_neg = tf.negative(sq)
    ...:         sq_neg = tf.Print(sq_neg, [i, sq_neg], message='i and sq_neg:')
    ...:         return tf.add(i, 1), sq_neg, sq
    ...:     return body
    ...:

In [14]: sess = tf.InteractiveSession()

In [15]: i = tf.Variable(0)
    ...: v = tf.Variable(2)
    ...: sq = tf.square(v)
    ...: l = tf.while_loop(cond, gen_body(v), (i, v, sq))
    ...: sess.run(tf.global_variables_initializer())
    ...: sess.run(l)
    ...:
2017-10-10 22:59:44.819271: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [2][4]
2017-10-10 22:59:44.819405: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [3]
2017-10-10 22:59:44.819466: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[0][-4]
2017-10-10 22:59:44.819553: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.819615: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.819680: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[1][-4]
2017-10-10 22:59:44.819827: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.819885: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.819932: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[2][-4]
2017-10-10 22:59:44.820034: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820094: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820111: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[3][-4]
2017-10-10 22:59:44.820162: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820250: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820265: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[4][-4]
2017-10-10 22:59:44.820315: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820379: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820408: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[5][-4]
2017-10-10 22:59:44.820428: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820438: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820446: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[6][-4]
2017-10-10 22:59:44.820464: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820490: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820500: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[7][-4]
2017-10-10 22:59:44.820519: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820532: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820542: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[8][-4]
2017-10-10 22:59:44.820559: I tensorflow/core/kernels/logging_ops.cc:79] x and sq: [-4][4]
2017-10-10 22:59:44.820580: I tensorflow/core/kernels/logging_ops.cc:79] v_assign: [-3]
2017-10-10 22:59:44.820593: I tensorflow/core/kernels/logging_ops.cc:79] i and sq_neg:[9][-4]
Out[15]: (10, -4, 4)

I created v variable and tensor sq which equals to v^2. In fact, we don't have control over them, they are our input as x and y and we know that y depends on x. I would like to assign new value inside TensorFlow loop to x (equavalent to v at example) and evaluate fresh y (sq inside example) at each iteration of the loop. Meanwhile we can do other evaluations inside while_loop, but most important is that I need to update x and get updated y. Currently, assigning operation doesn't propagate updates down to the dependant nodes and it shouldn't, but when someone calls tensor depending on value which were updated via assign inside while_loop, I suppose the tensor node must detect this change and evaluate new tensor value again.

Thanks!

The text was updated successfully, but these errors were encountered:

tatatodd · 2017-10-11T01:05:41Z

@skye Can you comment on this?

skye · 2017-10-11T17:55:16Z

Let me make sure I understand what you're requesting: you would like a tensor computed from variables to have a new updated value every time it's executed in a while loop.

This is a reasonable expectation. However, I don't think we'll change this behavior. Right now, only tensors defined inside the loop will be evaluated every loop iteration. All tensors defined outside a loop will be evaluated exactly once. So you would have to write something like:

import tensorflow as tf
import numpy as np

def cond(i, _x):
      return tf.less(i, 10)

def gen_body(v):
    def body(i, x):
        x = tf.Print(x, [x], "x: ")
        v_assign = v.assign(x + 1)
        v_assign = tf.Print(v_assign, [v_assign], "v_assign: ")
        with tf.control_dependencies([v_assign]):
          sq = tf.square(v)  # <---- put 'sq' definition inside loop
          sq = tf.Print(sq, [sq], "sq: ")
          sq_neg = tf.negative(sq)
        sq_neg = tf.Print(sq_neg, [i, sq_neg], message='i and sq_neg:')
        return tf.add(i, 1), sq_neg
    return body

sess = tf.InteractiveSession()

v = tf.Variable(2)
l = tf.while_loop(cond, gen_body(v), (1, v))
sess.run(tf.global_variables_initializer())
sess.run(l)

Note that I moved the sq = ... inside the loop (and also made it depend on v_assign, to make sure we pick up the new assignment).

Does this help?

skye · 2017-10-11T17:55:39Z

@ebrevdo @mrry FYI

ebrevdo · 2017-10-11T18:38:49Z

I believe if you want a variable to be evaluated at each value of the loop, you should access it using variable.ref

ebrevdo · 2017-10-11T18:39:50Z

or perhaps if you use a ResourceVariable (get_variable(..., use_resource=True)) it may do the right thing.

you may need to also set parallel_iterations=1, but probably not.

ebrevdo · 2017-10-11T18:51:24Z

This is an optimization that we turned on because you don't want a Variable sitting on a parameter server elsewhere in the datacenter to be accessed at each iteration of an RNN. you want to access it only once and cache it. in this case you have an assign inside the while_loop and converting a Variable to a Tensor doesn't look to see if an assign has been performed just prior to this, to know it must refresh the value. @mrry should we reset the caching identity tensor of a Variable after each assign?

ebrevdo · 2017-10-11T18:51:40Z

@skye also, wdyt?

skye · 2017-10-11T19:59:27Z

You're more familiar with these semantics than I am, but changing this might break (or at least regress) existing models, no?

ebrevdo · 2017-10-11T20:03:05Z

Not sure. Depends how many users perform a tf.assign/tf.assign_{add,sub} inside a while_loop (including applying an optimiser) and read the new value inside the while loop -- and what behavior they expect; and may be getting silent errors or models not converging because of this issue.

If this already just "works" with ResourceVariable, then that's a sign that the Variable semantics are just broken.

mrry · 2017-10-11T21:13:59Z

should we reset the caching identity tensor of a Variable after each assign?

I think this gets tricky because (from the perspective of a tf.Variable object) only some subset of the Variable.assign() (etc.) calls run in an any particular step, and so you can't use control dependencies to ensure that the identity is ordered after an assign. (Perhaps I am being too pessimistic and you have a more cunning solution in mind?)

If this already just "works" with ResourceVariable, then that's a sign that the Variable semantics are just broken.

+1. I'd strongly encourage people to switch to ResourceVariable in new code, and it's more compatible with other recent features like tf.data and some of the Estimator variants. I'm not sure what it would take to switch the default, however.

awav · 2017-10-11T21:14:29Z

Thank you for your responses. Here is some clarifications to the task:

There are a variable x and a tensor y - the result of computation upon x. y = tf.square(x). They created outside of the loop.
I pass implicitly to the while_loop both tensors - x and y
I assign new value to x inside loop (let's say x.assign(x + 1)) and expect y gets evaluated with new x value (x + 1) when it is used somewhere after assigning. So, the aim is to re-evaluate y at each iteration along with x assigning.

edited

@skye, so sq cannot be recreated inside while_loop, because we don't know what it does.

awav · 2017-10-11T21:38:48Z

I can be wrong, but in example below I never got different values for assign operation b and a variable outputs. I'm sure it can be a coincidence. Anyway, I got different values for c, and it is obvious because control_dependencies does not work in that case. I guess, that it would be useful to be able to construct an operation like read_value(), but for any tensor, so that whenever it is called it would re-evaluate that tensor.

In [91]: tf.reset_default_graph()
    ...: a = tf.get_variable('a', initializer=2, use_resource=True)
    ...: c = a + 1
    ...: b = a.assign(a * 10)
    ...: with tf.control_dependencies([b]):
    ...:     c = c # <<<<<<<<<<<<<<< 
    ...:              # here should be `c.eval_tensor()` or `c.read_tensor()`, 
    ...:              # which constructs an operation being equivalent
    ...:              # to read_value() for variables.
    ...:              # ***Edited***: tf.identity() doesn't work here!
    ...: sess = tf.InteractiveSession()
    ...: sess.run(tf.global_variables_initializer())
    ...: sess.run([a, b, c])
...
Out[91]: [20, 20, 3]   # <<< 3 
...
Out[92]: [20, 20, 21]   # <<< 21

awav · 2017-10-12T10:15:18Z

@skye, @mrry, @ebrevdo I'm sorry for bombarding you with multiple messages, here is a proof that assigning inside loop is not a problem:

import tensorflow as tf
import numpy as np

def cond(i, _x):
      return tf.less(i, 10000)

def gen_body(x):
    def body(i, xa_prev):
        xa = x.assign(xa_prev + 1)
        with tf.control_dependencies([xa]):
            with tf.control_dependencies([tf.assert_equal(x, xa)]):
                i = tf.add(i, 1)
        return i, xa
    return body

tf.reset_default_graph()
i = tf.get_variable('i', initializer=0, use_resource=True)
v = tf.get_variable('v', initializer=0, use_resource=True)
sess = tf.InteractiveSession()
loop = tf.while_loop(cond, gen_body(v), [i, v])
sess.run(tf.global_variables_initializer())
sess.run([loop])

Out[203]: [(10000, 10000)]

The issue is that there is no way to recompute tensors which depend on it and were constructed outside of the while_loop. :)

dchatterjee172 · 2017-10-12T10:56:34Z

I faced the same problem,

def body(rawstate,x,v,coef):
    rawstate.assign_add(tf.matmul(x[v],coef[v]))
    v.assign_add(1)
    return [rawstate,x,v,coef]

calrawstate=tf.while_loop(cond,body,[rawstate,x,v,coef])
And I faced this error,
AttributeError: 'Tensor' object has no attribute 'assign_add'
Here rawstate and v are two variables created using tf.get_variable()

Although this problem is trivial, and can be done without using assign or assign_add inside body
but sometimes it's needed in some case or in some custom model
suppose if we want to assign some specific value in rawstate[0] it's hard...

awav · 2017-10-12T11:13:26Z

@dchatterjee172, I think you have different issue. rawstate is a tensor, and by definition you can't assign value to it. If you expected to see same variable at each iteration, then it works a bit different. But it would be nice to pass variable to the loop as a reference.

@ebrevdo what is variable.ref?

ebrevdo · 2017-10-12T16:47:32Z

After playing with this for a bit, I reached two conclusions:

ResourceVariable has the same problem as Variable.
Variable.ref has been hidden from view; but you can access it using a hidden API (wherein your code will break in the future) via Variable._ref() or Variable._variable.

Here's code that does what I think you want it to do:

def cond(i, _x, _sq):
  return i < 5

def gen_body(v):
  def body(i, x, sq):
    x = tf.Print(x, [x, sq], "x and sq: ")
    with tf.control_dependencies([v.assign(x + 1)]):
      v_assign = v._ref()
    v_assign = tf.Print(v_assign, [v_assign], "v_assign: ")
    with tf.control_dependencies([v_assign]):
      sq_neg = tf.negative(sq)
      sq_neg = tf.Print(sq_neg, [i, sq_neg], message='i and sq_neg:')
      return tf.add(i, 1), sq_neg, sq
  return body

sess = tf.InteractiveSession()


i = tf.get_variable("i", initializer=0)
v = tf.get_variable("v", initializer=2)
sq = tf.square(v)
l = tf.while_loop(cond, gen_body(v), (i, v, sq))
sess.run(tf.global_variables_initializer())
sess.run((l, v))

@alextp any suggestions on how one would force accessing the updated value of a ResourceVariable within a calculation? This has implications not only for while loops but also other distributed computations where one wants to access a new value from another machine after waiting for a bit.

ebrevdo · 2017-10-12T16:47:56Z

@alextp perhaps a stateful read operator is in order?

alextp · 2017-10-12T17:02:00Z

It is possible to force a resource variable read to happen after an assignment, and indeed it will happen. @ebrevdo , why did resource variables not work? Do you have an example with them?

(I do not fully understand why we're using variables at all here instead of the regular while loop variable mechanism)

awav · 2017-10-12T19:34:08Z

@alextp, @ebrevdo Hello, your snippet doesn't do what I need, I simplified it a bit below. Let me explain what it does.
There is function f and variable x. At each iteration step I increment x and want to re-evaluate f and take negative of it. What's good: if I take assign tensorflow operation and pass it to the next iteration (even if I'm not using it), I can successfully increment external variable x. What's bad the function - f evaluated only once, but must be evaluated with freshly assigned x as many times as number of iterations. Check example below:

Here f(x) = x^2, and f_neg=-f(x):

def cond(i, _x_prev, _f_prev):
  return i < 3

def gen_body(x, f):
  def body(i, _x_prev, _f_prev):
    x_assign = x.assign(x + 1)
    # with tf.control_dependencies([x.assign(x + 1)]):
    #   x_assign = x._ref()
    with tf.control_dependencies([x_assign]):
      f_neg = tf.negative(f)
      i = tf.add(i, 1)
      i = tf.Print(i, [i], message='>>> Iteration ')
      i = tf.Print(i, [x], message='x = ')
      i = tf.Print(i, [x_assign], message='x_assign = ')
      i = tf.Print(i, [f], message='f = ')
      i = tf.Print(i, [f_neg], message='f_neg = ')
      return i, x_assign, f_neg
  return body

tf.reset_default_graph()
sess = tf.InteractiveSession()
i = tf.get_variable("i", initializer=0)
v = tf.get_variable("v", initializer=0)
func_v = tf.square(v)
l = tf.while_loop(cond, gen_body(v, func_v), (i, v, func_v))
sess.run(tf.global_variables_initializer())
sess.run((l, v))

Output. x incremented successfully from 0 to 3, but f remains non-updated! I need updated f.

... >>> Iteration [1]
... x = [1]
... x_assign = [1]
... f = [0]
... f_neg = [0]

... >>> Iteration [2]
... x = [2]
... x_assign = [2]
... f = [0]
... f_neg = [0]

... >>> Iteration [3]
... x = [3]
... x_assign = [3]
... f = [0]
... f_neg = [0]

In [259]: tf.__version__
Out[259]: '1.2.1'

I propose to add a method to the tensor structure which will be equivalent to read_value() for standard variable. It will form operation which re-evalutes the tensor when it is required. In example above that tensor is f.

awav · 2017-10-12T21:35:01Z

@alextp,

(I do not fully understand why we're using variables at all here instead of the regular while loop variable mechanism)

The task is update the variable and the dependant on it function which are defined outside of while_loop. In example above, the f is func_v tensor and x is v variable, both passed to gen_body.

alextp · 2017-10-13T13:15:05Z

Right. If you want the function to be computed inside the while loop you need to call it inside the while loop. Something like

def cond(i, _x_prev, _f_prev):
  return i < 3

def gen_body(x, f):
  def body(i, _x_prev, _f_prev):
    x_assign = x.assign(x + 1)
    # with tf.control_dependencies([x.assign(x + 1)]):
    #   x_assign = x._ref()
    with tf.control_dependencies([x_assign]):
      f_neg = tf.negative(f(x_assign))
      i = tf.add(i, 1)
      i = tf.Print(i, [i], message='>>> Iteration ')
      i = tf.Print(i, [x], message='x = ')
      i = tf.Print(i, [f(x_assign)], message='x_assign = ')
      i = tf.Print(i, [f(v)], message='f = ')
      i = tf.Print(i, [f_neg], message='f_neg = ')
      return i, x_assign, f_neg
  return body

tf.reset_default_graph()
sess = tf.InteractiveSession()
i = tf.get_variable("i", initializer=0)
v = tf.get_variable("v", initializer=0)
func_v = lambda v: tf.square(v)
l = tf.while_loop(cond, gen_body(v, func_v), (i, v, func_v(v)))
sess.run(tf.global_variables_initializer())
sess.run((l, v))

When you define func_v as a tensor outside the loop it has a fixed value which will stay fixed forever. If you want it to be recomputed you need to actually recompute it.

awav · 2017-10-13T14:46:14Z

@alextp, that's right. In fact, it was a feature request - to be able to recompute the tensor and introduce property which generates this operation. I'm sorry if it wasn't clear from previous messages That would be super useful in while_loop-like flow statements.

The solution which you proposed doesn't work for me, as I do not have control on input tensors x and f, and I do not know what's inside f.

How hard is it to implement re-computation operation for tensor?

alextp · 2017-10-13T14:52:57Z

With the current graph semantics it's essentially impossible.

…

On Fri, Oct 13, 2017 at 7:51 AM, Artem Artemev ***@***.***> wrote: @alextp <https://github.com/alextp>, that's right. In fact, it was a feature request - to be able to recompute the tensor and introduce property which generates this operation. I'm sorry if it wasn't clear from previous messages That would be super useful in while_loop-like flow statements. The solution which you proposed doesn't work for me, as I do have control on input tensors x and f, and I do not know what's inside f. How hard is it to implement re-computation operation for tensor? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#13616 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAATxcKI_k6pTejahsF7TvgFWFYvTgyCks5sr3j5gaJpZM4P0qPF> .

-- - Alex

awav · 2017-10-13T15:38:11Z

@alextp. oh, that's bad. Okay, If I manage to pass f(x) as python function, hence at each iteration I construct new tensor y = f(x), does it mean that the size of the graph will grow linearly with number of iterations? Can an unwise usage of while_loop lead to the graph "overflow", taking into account that graph size is limited?

alextp · 2017-10-13T15:45:27Z

No, the size of the graph is constant even if you call f(x) in each iteration, because tensorflow's while loop is symbolic and only creates the body graph once.

…

On Fri, Oct 13, 2017 at 8:43 AM, Artem Artemev ***@***.***> wrote: @alextp <https://github.com/alextp>. oh, that's bad. Okay, I managed to pass f(x) as python function and at each iteration I construct new tensor y = f(x), does it mean that the size of the graph will grow linearly with number of iterations? Can an unwise usage of while_loop lead to the graph "overflow", taking into account that graph size is limited? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#13616 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAATxXr6HKksKduSmuWFS3-emCehda_tks5sr4UrgaJpZM4P0qPF> .

-- - Alex

awav · 2017-10-13T20:32:24Z

@alextp, last question: If I create c++ operation which will accept x variable and f tensor, will I be able to manipulate an execution of f depending on changes in x. In other words, will I be able to execute f tensor each time when I change x. Thanks!

alextp · 2017-10-13T20:37:01Z

You need to explicitly include this C++ operation in the graph in all the places where you'd want it to run. There's no way to trigger it running only when x is modified.

…

On Fri, Oct 13, 2017 at 1:35 PM, Artem Artemev ***@***.***> wrote: @alextp <https://github.com/alextp>, last question: If I create c++ operation, which will accept x variable and f tensor, will I be able to manipulate execution of f depending on changes in x. In other words, will I be able to execute f tensor each time when I change x. Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#13616 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAATxUGGup22k2UCOAs2An65SDeEfn6fks5sr8mYgaJpZM4P0qPF> .

-- - Alex

awav · 2017-10-15T19:17:13Z

Thanks all for your responses!

alextp · 2018-11-05T17:47:20Z

Reevaluation does not work outside of a loop; a tensor is only evaluated a single time within a single call to session.run. Try to have more than one update per call to session.run and you'll see things breaking. (your example also relies on the lack of a memory model for legacy tf variables but this is not relevant)

…

On Sat, Nov 3, 2018 at 11:32 AM Jonas Eschle ***@***.***> wrote: A year later, did anything change on that? @alextp <https://github.com/alextp>, is it still impossible? I find that it somehow works and am confused about why it does/does not (I write here as the previous discussion is basically the same thing as I try to accomplish): Basically, the re-evaluation works *outside* of a loop, that's confusing for me Pseudo-code x = <Tensor depending on var> var_assign = tf.assign(var, value) with tf.control_dependencies([var_assign]) sess.run(x) # prints with the old value, of course but adding an identity before var_assign = tf.assign(var, value) with tf.control_dependencies([var_assign]) x_2 = tf.identity(x) sess.run(x_2) # prints with the new value! So in the second case, sess.run reevaluates the whole Tensor X. But this does not work in a while loop (of course, we cannot have a sess.run). So why does x_2 get's re-evaluated here but not in a while-loop? I did not expect this way to work — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#13616 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAATxfWbCkETnuov7i73C5uTctP_vaXXks5ureFJgaJpZM4P0qPF> .

-- - Alex

jonas-eschle · 2018-11-05T23:21:37Z

Ah, I see (so it's actually independent of a loop): one sess.run -> zero (if can be cached) or one evaluation of a tensor.

Thanks, @alextp!

tatatodd added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Oct 11, 2017

skye added stat:awaiting response Status - Awaiting response from author and removed stat:awaiting tensorflower Status - Awaiting response from tensorflower labels Oct 11, 2017

aselle removed the stat:awaiting response Status - Awaiting response from author label Oct 11, 2017

tatatodd added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Oct 12, 2017

awav closed this as completed Oct 12, 2017

awav reopened this Oct 12, 2017

awav closed this as completed Oct 15, 2017

gongzhitaao mentioned this issue Jan 27, 2018

Computing gradients of loop variables return None #16333

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Force tensor evaluation inside while-loop, scan and others. #13616

Force tensor evaluation inside while-loop, scan and others. #13616

awav commented Oct 10, 2017 •

edited

tatatodd commented Oct 11, 2017

skye commented Oct 11, 2017

skye commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

skye commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

mrry commented Oct 11, 2017

awav commented Oct 11, 2017 •

edited

awav commented Oct 11, 2017 •

edited

awav commented Oct 12, 2017 •

edited

dchatterjee172 commented Oct 12, 2017 •

edited

awav commented Oct 12, 2017 •

edited

ebrevdo commented Oct 12, 2017

ebrevdo commented Oct 12, 2017

alextp commented Oct 12, 2017

awav commented Oct 12, 2017 •

edited

awav commented Oct 12, 2017 •

edited

alextp commented Oct 13, 2017

awav commented Oct 13, 2017 •

edited

alextp commented Oct 13, 2017 via email

awav commented Oct 13, 2017 •

edited

alextp commented Oct 13, 2017 via email

awav commented Oct 13, 2017 •

edited

alextp commented Oct 13, 2017 via email

awav commented Oct 15, 2017

alextp commented Nov 5, 2018 via email

jonas-eschle commented Nov 5, 2018

Force tensor evaluation inside while-loop, scan and others. #13616

Force tensor evaluation inside while-loop, scan and others. #13616

Comments

awav commented Oct 10, 2017 • edited

tatatodd commented Oct 11, 2017

skye commented Oct 11, 2017

skye commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

skye commented Oct 11, 2017

ebrevdo commented Oct 11, 2017

mrry commented Oct 11, 2017

awav commented Oct 11, 2017 • edited

awav commented Oct 11, 2017 • edited

awav commented Oct 12, 2017 • edited

dchatterjee172 commented Oct 12, 2017 • edited

awav commented Oct 12, 2017 • edited

ebrevdo commented Oct 12, 2017

ebrevdo commented Oct 12, 2017

alextp commented Oct 12, 2017

awav commented Oct 12, 2017 • edited

awav commented Oct 12, 2017 • edited

alextp commented Oct 13, 2017

awav commented Oct 13, 2017 • edited

alextp commented Oct 13, 2017 via email

awav commented Oct 13, 2017 • edited

alextp commented Oct 13, 2017 via email

awav commented Oct 13, 2017 • edited

alextp commented Oct 13, 2017 via email

awav commented Oct 15, 2017

alextp commented Nov 5, 2018 via email

jonas-eschle commented Nov 5, 2018

awav commented Oct 10, 2017 •

edited

awav commented Oct 11, 2017 •

edited

awav commented Oct 11, 2017 •

edited

awav commented Oct 12, 2017 •

edited

dchatterjee172 commented Oct 12, 2017 •

edited

awav commented Oct 12, 2017 •

edited

awav commented Oct 12, 2017 •

edited

awav commented Oct 12, 2017 •

edited

awav commented Oct 13, 2017 •

edited

awav commented Oct 13, 2017 •

edited

awav commented Oct 13, 2017 •

edited