@srkreddy1238 @yzhliu Thanks comments!
If all of you agree, I will make TFLite frontend support from NCHW to NHWC.
@yzhliu Yes. quantization part support is not been upstreamed yet. It has many
changes. I plan to upstream it in dev 0.6. My original plan is to support
TFLite NHWC the reason is w
Dear podling,
This email was sent by an automated system on behalf of the Apache
Incubator PMC. It is an initial reminder to give you plenty of time to
prepare your quarterly board report.
The board meeting is scheduled for Wed, 15 May 2019, 10:30 am PDT.
The report for your podling will form a p
I prefer not to do any layout conversion in frontend.
Just to clarify, tflite quantization has not been upstreamed, right?
@FrozenGene
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/dmlc/tvm/issues/2519
@masahi In your exmaple of two strided_slice, given that we do not yet
cooperatively fuse the two parallel strided_slice together atm, the second
version is mostly as efficient as the first one.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or
I added a section on PassUpDomain's limitations, and made a pull request
#[3040](https://github.com/dmlc/tvm/pull/3040).
@YuanLin, I didn't end up including the content on threads. I think it would be
helpful to add more explanation of basic thread concepts, to help the reader
understand why
@tqchen I'm looking into the following approach:
* Initially, the tuple and its fields are marked as Injective, so that they can
be fused as much as possible.
* Later, if we detect that a tuple is the root of its group in [this
line](https://github.com/dmlc/tvm/blob/master/src/relay/pass/fuse_op
@masahi I see, thanks. Another option is probably using a copy operator if
there are duplicates.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/dmlc/tvm/issues/3039#issuecomment-484385353
@zhiics It looks like the tuple with duplicated tensors is only problematic if
it is the return value of a subfunction (i.e. a function that is lowered to
topi and compiled by TVM). If we lift the tuple out of a subfunction and put it
under the global function, it seems to work fine. The test be
