maheshambule commented on a change in pull request #4952: [Frontend]
[Tensorflow] ReadVariableOp operator support
URL: https://github.com/apache/incubator-tvm/pull/4952#discussion_r385552641
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2187,6 +2187,14 @@ de
masahi commented on issue #4927: [Relay][Pass] Add inline pass
URL: https://github.com/apache/incubator-tvm/pull/4927#issuecomment-592408781
@zhiics not sure if this is one of an intended use cases, but I think adding
a test case for inlining a composite function would be a useful addition.
haiyang-tju commented on issue #4261: DepthWise Transposed Conv error
URL: https://github.com/apache/incubator-tvm/issues/4261#issuecomment-592423340
It does not support the situations of `groups!=1`, and you must set `groups`
as default (1) in `nn.ConvTranspose2d`.
---
leandron opened a new pull request #4965: [CI][Docker] pin xgboost dependency
version to 0.90
URL: https://github.com/apache/incubator-tvm/pull/4965
Pin xgboost dependency version to 0.90
* Sets xgboost dependency to be 0.90, preventing
segfaults during TVM python unit tests e
merrymercy commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385613944
##
File path: src/te/schedule/message_passing.cc
##
@@ -72,14 +121,35 @@ void PassDownDomain(const Stage& stage,
merrymercy commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385615522
##
File path: src/te/schedule/message_passing.cc
##
@@ -51,6 +51,45 @@ void Update(std::unordered_map* p_state,
}
merrymercy commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385613563
##
File path: src/te/schedule/message_passing.cc
##
@@ -62,6 +101,16 @@ void PassDownDomain(const Stage& stage,
r
merrymercy commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385615817
##
File path: src/te/schedule/message_passing.cc
##
@@ -51,6 +51,45 @@ void Update(std::unordered_map* p_state,
}
inadob commented on a change in pull request #4704: [Relay][Frontend][TFLite]
Add parser support for arg_min_max
URL: https://github.com/apache/incubator-tvm/pull/4704#discussion_r385636005
##
File path: python/tvm/relay/frontend/tflite.py
##
@@ -826,6 +828,50 @@ def _conv
leandron commented on issue #4965: [CI][Docker] pin xgboost dependency version
to 0.90
URL: https://github.com/apache/incubator-tvm/pull/4965#issuecomment-592467369
It seems the broken build reason is not specifically related to this patch.
-
inadob commented on a change in pull request #4704: [Relay][Frontend][TFLite]
Add parser support for arg_min_max
URL: https://github.com/apache/incubator-tvm/pull/4704#discussion_r385642946
##
File path: python/tvm/relay/frontend/tflite.py
##
@@ -826,6 +828,50 @@ def _conv
inadob commented on issue #4805: [Frontend][TFlite] Add parser support for
relu6, leaky_relu, relu_n1_to_1, log_softmax
URL: https://github.com/apache/incubator-tvm/pull/4805#issuecomment-592482232
> Relu and Clip implementation does not look right.
>
> We can keep the computation in
inadob edited a comment on issue #4805: [Frontend][TFlite] Add parser support
for relu6, leaky_relu, relu_n1_to_1, log_softmax
URL: https://github.com/apache/incubator-tvm/pull/4805#issuecomment-592482232
> Relu and Clip implementation does not look right.
>
> We can keep the computa
masahi commented on issue #4965: [CI][Docker] pin xgboost dependency version to
0.90
URL: https://github.com/apache/incubator-tvm/pull/4965#issuecomment-592502826
hey @leandron we are in the same boat. The discussion is happening at
https://github.com/apache/incubator-tvm/pull/4961#issueco
masahi commented on a change in pull request #4927: [Relay][Pass] Add inline
pass
URL: https://github.com/apache/incubator-tvm/pull/4927#discussion_r385684873
##
File path: src/relay/pass/inline.cc
##
@@ -0,0 +1,227 @@
+/*
+ * Licensed to the Apache Software Foundation (AS
leandron commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592512021
On a quick look, *it seems* it is behaving as expected by
`task_sphinx_precheck.sh`. It is
leandron commented on issue #4965: [CI][Docker] pin xgboost dependency version
to 0.90
URL: https://github.com/apache/incubator-tvm/pull/4965#issuecomment-592512753
> hey @leandron we are in the same boat. The discussion is happening at
[#4961
(comment)](https://github.com/apache/incubato
leandron edited a comment on issue #4961: [Torch] Upsampling op support and
enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592512021
On a quick look, *it seems* it is behaving as expected by
`task_sphinx_precheck.sh`.
leandron commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592515583
Discussion about fixing docs is happening in #4908.
---
masahi commented on a change in pull request #4927: [Relay][Pass] Add inline
pass
URL: https://github.com/apache/incubator-tvm/pull/4927#discussion_r385720594
##
File path: src/relay/pass/inline.cc
##
@@ -0,0 +1,227 @@
+/*
+ * Licensed to the Apache Software Foundation (AS
zhiics commented on a change in pull request #4927: [Relay][Pass] Add inline
pass
URL: https://github.com/apache/incubator-tvm/pull/4927#discussion_r385782213
##
File path: src/relay/pass/inline.cc
##
@@ -0,0 +1,227 @@
+/*
+ * Licensed to the Apache Software Foundation (AS
zhiics commented on a change in pull request #4927: [Relay][Pass] Add inline
pass
URL: https://github.com/apache/incubator-tvm/pull/4927#discussion_r385786133
##
File path: src/relay/pass/inline.cc
##
@@ -0,0 +1,227 @@
+/*
+ * Licensed to the Apache Software Foundation (AS
zhiics commented on issue #4927: [Relay][Pass] Add inline pass
URL: https://github.com/apache/incubator-tvm/pull/4927#issuecomment-592585040
@masahi We have annotated functions with external compilers, no difference
with composite if we need to add it.
-
zhiics edited a comment on issue #4927: [Relay][Pass] Add inline pass
URL: https://github.com/apache/incubator-tvm/pull/4927#issuecomment-592585040
@masahi We have annotated functions with external compilers, no difference
with composite if we need to add it. I don’t think we need to add it
yongfeng-nv commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385793332
##
File path: src/te/schedule/message_passing.cc
##
@@ -62,6 +101,16 @@ void PassDownDomain(const Stage& stage,
yongfeng-nv commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385793479
##
File path: src/te/schedule/message_passing.cc
##
@@ -72,14 +121,35 @@ void PassDownDomain(const Stage& stage,
yongfeng-nv commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385793595
##
File path: src/te/schedule/message_passing.cc
##
@@ -51,6 +51,45 @@ void Update(std::unordered_map* p_state,
}
yongfeng-nv commented on a change in pull request #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#discussion_r385793686
##
File path: src/te/schedule/message_passing.cc
##
@@ -51,6 +51,45 @@ void Update(std::unordered_map* p_state,
}
yongfeng-nv commented on issue #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#issuecomment-592591273
> Use const reference when possible.
> The logic looks good to me.
> cc @tqchen for double check.
Thank you for the careful review. I forgo
FrozenGene commented on issue #4803: [WIP][Frontend] Asymmetric padding of
convolution support
URL: https://github.com/apache/incubator-tvm/pull/4803#issuecomment-592592327
> So, if I understand correctly there are 2 issues here
>
> * If you try to pass the asymmetric padding via QNN
inadob opened a new pull request #4966: [Frontend][TFLite] Add parser support
for l2_normalization
URL: https://github.com/apache/incubator-tvm/pull/4966
* TF doesn't provide uint8 support
* TFLite does the normalization only if it's over the last dimension
* TFL uses only the default
comaniac commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592643432
Yes. It is due to the doc checker introduced yesterday. I'm looking into the
problem now.
anijain2305 commented on issue #4803: [WIP][Frontend] Asymmetric padding of
convolution support
URL: https://github.com/apache/incubator-tvm/pull/4803#issuecomment-592645558
Thanks for reminding. Yes, I will make QNN Conv2D accept 4D padding for now,
to make parser cleaner for now. The cha
jwfromm commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385832926
##
File path: tests/python/frontend/pytorch/test_forward.py
##
jwfromm commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385830597
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@
jwfromm commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592646686
Left minor comments but otherwise looks good. Having a custom convert map is
a neat idea tha
tqchen commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592647164
docs error should be fixed by
https://github.com/apache/incubator-tvm/pull/4967
tqchen opened a new pull request #4967: [DOCS] Fix sphinx precheck
URL: https://github.com/apache/incubator-tvm/pull/4967
cc @icemelon9 @comaniac @ZihengJiang
This is an automated message from the Apache Git Service.
To respo
pingsutw commented on issue #4950: The links not found in Depthwise Convolution
tutorial
URL: https://github.com/apache/incubator-tvm/issues/4950#issuecomment-592651732
Thanks @cchung100m.
sure, I'm going to send a PR
Th
wpan11nv opened a new pull request #4968: [TOPI][CUDA] Teak CUDA injective
scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968
- vectorizing fp32/int32 types is also beneficial.
Signed-off-by: Wei Pan
Thanks for contributing to TVM! Please refer to guideline
h
wpan11nv commented on issue #4968: [TOPI][CUDA] Teak CUDA injective scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592657153
Perf data on 1080TI with n = 134217728
Exec time without vectorization: 3.923180 ms., float32
Exec time without vectorization:
kevinthesun commented on a change in pull request #4938: [topi][relay] add
operation tan to TVM
URL: https://github.com/apache/incubator-tvm/pull/4938#discussion_r385851691
##
File path: python/tvm/relay/op/_tensor.py
##
@@ -27,6 +27,7 @@
register_broadcast_schedule("
notoraptor commented on issue #4938: [topi][relay] add operation tan to TVM
URL: https://github.com/apache/incubator-tvm/pull/4938#issuecomment-592669418
Hi @kevinthesun I rebased and added shape function for tan in latest commit.
I still have a mismatch error in
`tests/python/fronte
kevinthesun opened a new issue #4969: [RFC] Enhance TensorFlow Frontend Control
Flow Support
URL: https://github.com/apache/incubator-tvm/issues/4969
### Summary
In today’s TVM TensorFlow frontend, there is only limited support for
control flow, which resulting in difficult in covering
zhiics edited a comment on issue #4964: [Torch] Add initial control flow
support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592674477
@masahi Thank you very much for the nice work! I am familiar the
control-flow, and trying to read more about PyTorch control-flow
zhiics commented on issue #4964: [Torch] Add initial control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592674477
@masahi Thank you very much for the nice work! I am familiar the
control-flow, and trying to read more about PyTorch control-flow construc
zhiics commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592675406
Just a reminder, to support these models we need some patches for tensor
array as well. mask_rcnn seems requiring some mor
zhiics commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592677784
ahh, one more reminder, for all these models, we will have OOM problem for
pretty printing after the ANF pass. It is very
zhiics edited a comment on issue #4969: [RFC] Enhance TensorFlow Frontend
Control Flow Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592677784
ahh, one more reminder, for all these models, we will have OOM problem for
pretty printing after the ANF pass. It i
zhiics commented on a change in pull request #4927: [Relay][Pass] Add inline
pass
URL: https://github.com/apache/incubator-tvm/pull/4927#discussion_r385786133
##
File path: src/relay/pass/inline.cc
##
@@ -0,0 +1,227 @@
+/*
+ * Licensed to the Apache Software Foundation (AS
anijain2305 edited a comment on issue #4962: Support dilation in x86 NCHWc
depthwise conv
URL: https://github.com/apache/incubator-tvm/issues/4962#issuecomment-592682877
I think it is just a missing feature.
This is an automa
anijain2305 commented on issue #4962: Support dilation in x86 NCHWc depthwise
conv
URL: https://github.com/apache/incubator-tvm/issues/4962#issuecomment-592682877
I think it is just missing feature.
This is an automated messa
anijain2305 commented on issue #4805: [Frontend][TFlite] Add parser support for
relu6, leaky_relu, relu_n1_to_1, log_softmax
URL: https://github.com/apache/incubator-tvm/pull/4805#issuecomment-592684003
> The problem here was that `Clip()` from `RELAY_PASS_PATTERN ` needs the
clipping rang
zhiics edited a comment on issue #4964: [Torch] Add initial control flow
support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592674477
@masahi Thank you very much for the nice work! I am familiar the TF
control-flow, and trying to read more about PyTorch control-fl
alexwong commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385913650
##
File path: tests/python/frontend/pytorch/test_forward.py
#
alexwong commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385913650
##
File path: tests/python/frontend/pytorch/test_forward.py
#
alexwong commented on a change in pull request #4964: [Torch] Add initial
control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385917420
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -123,7 +124,8 @@ def _impl(inputs, input_ty
alexwong commented on a change in pull request #4964: [Torch] Add initial
control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385924756
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -614,6 +615,55 @@ def _impl(inputs, input_t
alexwong commented on a change in pull request #4964: [Torch] Add initial
control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385917420
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -123,7 +124,8 @@ def _impl(inputs, input_ty
alexwong commented on a change in pull request #4964: [Torch] Add initial
control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385924885
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -614,6 +615,55 @@ def _impl(inputs, input_t
This is an automated email from the ASF dual-hosted git repository.
masahi pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git
The following commit(s) were added to refs/heads/master by this push:
new a449d8b [DOCS] Fix sphinx precheck (#496
masahi merged pull request #4967: [DOCS] Fix sphinx precheck
URL: https://github.com/apache/incubator-tvm/pull/4967
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub
tqchen commented on issue #4965: [CI][Docker] pin xgboost dependency version to
0.90
URL: https://github.com/apache/incubator-tvm/pull/4965#issuecomment-592734767
the fix has been merged, please send another commit to retrigger the CI
---
tqchen edited a comment on issue #4965: [CI][Docker] pin xgboost dependency
version to 0.90
URL: https://github.com/apache/incubator-tvm/pull/4965#issuecomment-592734767
the fix https://github.com/apache/incubator-tvm/pull/4967 has been merged,
please send another commit to retrigger the C
jmorrill opened a new pull request #4970: Added CopyFromBytes and CopyToBytes
convenience methods. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970
It is not very intuitive (specially as a beginner) on how to copy a `void*`
to an `NDArray`. Currently one may have to co
masahi commented on a change in pull request #4964: [Torch] Add initial control
flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385931158
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -123,7 +124,8 @@ def _impl(inputs, input_type
masahi commented on a change in pull request #4964: [Torch] Add initial control
flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385932502
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -614,6 +615,55 @@ def _impl(inputs, input_typ
masahi commented on a change in pull request #4964: [Torch] Add initial control
flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#discussion_r385932765
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -614,6 +615,55 @@ def _impl(inputs, input_typ
jmorrill commented on issue #4968: [TOPI][CUDA] Teak CUDA injective scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592741278
I applied this commit and I get the following exception when I
"build_module" (On a 1080GTX):
```
tvm._ffi.base.TVMError: Tra
kumasento commented on issue #4847: Return empty CSourceModule when no
lowered_funcs exists in Relay mod
URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-592741362
@FrozenGene @tqchen @zhiics @mbaret Would you mind giving another read on
this? Now the approach is creatin
tqchen commented on issue #4908: [DOCS] Fix Sphinx Warnings
URL: https://github.com/apache/incubator-tvm/issues/4908#issuecomment-592741577
Thanks to @comaniac @cchung100m 's effort, we have now cleaned up all the
WARNINGS and setup a CI check. A remaining step would be to bring the sphinx
tqchen closed issue #4908: [DOCS] Fix Sphinx Warnings
URL: https://github.com/apache/incubator-tvm/issues/4908
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and u
masahi commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592743884
wow, supporting mask rcnn is a great achievement! I was also trying to
support mask rcnn in torch, but I was stuck at dyna
tqchen commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385939469
##
File path: include/tvm/runtime/ndarray.h
##
@@ -306,6 +324,
zhiics commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592746247
Yes, @yongwww had one months back
https://github.com/apache/incubator-tvm/pull/4312
---
masahi commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385942736
##
File path: tests/python/frontend/pytorch/test_forward.py
###
jmorrill commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385944181
##
File path: include/tvm/runtime/ndarray.h
##
@@ -306,6 +32
masahi commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592751982
great, will take a look. Hopefully the PR would become active soon.
--
masahi commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385946326
##
File path: tests/python/frontend/pytorch/test_forward.py
###
masahi commented on issue #4961: [Torch] Upsampling op support and enable
registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#issuecomment-592753364
@alexwong @jwfromm I removed the dependency on `packaging` and dropped
support for PT 1.2 and
wpan11nv commented on issue #4968: [TOPI][CUDA] Teak CUDA injective scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592754184
> type!
What is your test? I'll have a look. Thanks!
This i
jmorrill commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385947900
##
File path: include/tvm/runtime/ndarray.h
##
@@ -306,6 +32
jmorrill commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385947900
##
File path: include/tvm/runtime/ndarray.h
##
@@ -306,6 +32
kevinthesun commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control
Flow Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592759262
One blocking issue for dynamic NMS is that it needs to use dynamic
strided_slice. However, current fusion pass can't
tqchen commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385955400
##
File path: include/tvm/runtime/ndarray.h
##
@@ -68,19 +68,3
jmorrill commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385959768
##
File path: include/tvm/runtime/ndarray.h
##
@@ -68,19 +68
yongfeng-nv commented on issue #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#issuecomment-592769137
I don't understand the current failure:
https://ci.tvm.ai/blue/organizations/jenkins/tvm/detail/PR-4931/7/pipeline.
How do I handle it?
-
masahi commented on a change in pull request #4961: [Torch] Upsampling op
support and enable registering a user defined op conversion map
URL: https://github.com/apache/incubator-tvm/pull/4961#discussion_r385963791
##
File path: python/tvm/relay/frontend/pytorch.py
##
@@ -
jmorrill commented on issue #4968: [TOPI][CUDA] Teak CUDA injective scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592771893
> > type!
>
> What is your test? I'll have a look. Thanks!
On a 1080gtx
Using [this mxnet
model](http://insightface.ai/fil
jmorrill edited a comment on issue #4968: [TOPI][CUDA] Teak CUDA injective
scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592771893
> > type!
>
> What is your test? I'll have a look. Thanks!
On a 1080gtx
Using [this mxnet
model](http://insigh
jmorrill edited a comment on issue #4968: [TOPI][CUDA] Teak CUDA injective
scheduler
URL: https://github.com/apache/incubator-tvm/pull/4968#issuecomment-592771893
> > type!
>
> What is your test? I'll have a look. Thanks!
On a 1080gtx (cuda 10.1)
Using [this mxnet
model](h
comaniac commented on issue #4931: Tighten split's extent
URL: https://github.com/apache/incubator-tvm/pull/4931#issuecomment-592780728
This CI failure has been fixed several hours ago. Rebasing this branch to
upstream and the errors should be gone.
tqchen commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385972583
##
File path: src/runtime/ndarray.cc
##
@@ -185,6 +185,18 @@ N
jmorrill commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385980113
##
File path: src/runtime/ndarray.cc
##
@@ -185,6 +185,18 @@
MarisaKirisame commented on issue #4964: [Torch] Add initial control flow
support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592792701
can you dont call it parse? parse is for converting strings to an ast.
converting ast to ast is a converter.
nice work oth
masahi commented on issue #4969: [RFC] Enhance TensorFlow Frontend Control Flow
Support
URL: https://github.com/apache/incubator-tvm/issues/4969#issuecomment-592795910
I'm not familiar with dynamic shape support in Relay, so I cannot comment on
how it relates to fusion. But it sounds like
masahi commented on issue #4964: [Torch] Add initial control flow support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592798088
> can you dont call it parse? parse is for converting strings to an ast.
converting ast to ast is a converter.
no probelm, will do.
masahi edited a comment on issue #4964: [Torch] Add initial control flow
support
URL: https://github.com/apache/incubator-tvm/pull/4964#issuecomment-592798088
Tests passed!!
> can you dont call it parse? parse is for converting strings to an ast.
converting ast to ast is a converte
tqchen commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385990046
##
File path: src/runtime/ndarray.cc
##
@@ -185,6 +185,38 @@ N
tqchen commented on a change in pull request #4970: Added CopyFromBytes and
CopyToBytes convenience methods to NDArray. Fixed typos.
URL: https://github.com/apache/incubator-tvm/pull/4970#discussion_r385990080
##
File path: src/runtime/ndarray.cc
##
@@ -185,6 +185,18 @@ N
1 - 100 of 116 matches
Mail list logo