kice commented on issue #4569: Some Windows and MSVC fixes
URL: https://github.com/apache/incubator-tvm/pull/4569#issuecomment-568685746
I assume that everytime the byte array we feed `py_str` comes from different
program, which might have different encoding.
The most common encoding
optima2005 commented on issue #4511: [CONV] Asymmetric padding
URL: https://github.com/apache/incubator-tvm/pull/4511#issuecomment-568669112
rebase to master to fix conflict.
This is an automated message from the Apache Git Se
tqchen merged pull request #4566: [Relay/Topi][Op] Added native DepthToSpace
and SpaceToDepth Operators
URL: https://github.com/apache/incubator-tvm/pull/4566
This is an automated message from the Apache Git Service.
To resp
tqchen commented on issue #4566: [Relay/Topi][Op] Added native DepthToSpace and
SpaceToDepth Operators
URL: https://github.com/apache/incubator-tvm/pull/4566#issuecomment-568654294
Thanks @jwfromm @kice !
This is an automated
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from f9bc748 [DEPRECATION] Remove NNVM compiler (#4571)
add 9b92c53 [Relay/Topi][Op] Added native DepthToSpa
tqchen edited a comment on issue #4579: [RFC] Deprecate Extension VTable in
Favor of Unified Object System
URL: https://github.com/apache/incubator-tvm/issues/4579#issuecomment-568654027
No, as nnvm compiler has already been deprecated, we won't support nnvm
compiler objects. New projects
tqchen commented on issue #4579: [RFC] Deprecate Extension VTable in Favor of
Unified Object System
URL: https://github.com/apache/incubator-tvm/issues/4579#issuecomment-568654027
No, as nnvm compiler has already been deprecated, we won't support nnvm
compiler objects
tqchen commented on issue #4578: [RUNTIME] Remove Extension VTable in favor of
Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578#issuecomment-568653824
the example(apps/extension) has been updated
ghostplant commented on issue #4572: What is an acceptable way to write an
one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568647610
@tqchen Anyone who can help me answer other of my old question lists below?
Thanks!
https://discuss.
junrushao1994 commented on issue #4578: [RUNTIME] Remove Extension VTable in
favor of Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578#issuecomment-568642078
Do we have a tutorial for this somewhere around? If so, shall we also delete
this tutorial?
--
junrushao1994 commented on issue #4579: [RFC] Deprecate Extension VTable in
Favor of Unified Object System
URL: https://github.com/apache/incubator-tvm/issues/4579#issuecomment-568641741
The implication here is that we won't support nnvm objects any more, right?
If so, I think it's fine
-
FrozenGene commented on issue #4511: [CONV] Asymmetric padding
URL: https://github.com/apache/incubator-tvm/pull/4511#issuecomment-568633447
Thanks. Could you help to resolve the conflict?
cc @optima2005 @zhiics @yzhliu @Huyuwei anyone could help to review and
handle it?
--
tqchen commented on issue #4566: [Relay/Topi][Op] Added native DepthToSpace and
SpaceToDepth Operators
URL: https://github.com/apache/incubator-tvm/pull/4566#issuecomment-568628797
@kice please
https://docs.tvm.ai/contribute/code_review.html#approve-and-request-changes-explicitly
tqchen commented on issue #4496: [NNVM][RFC] Begin the removal of NNVM Compiler
URL: https://github.com/apache/incubator-tvm/issues/4496#issuecomment-568616533
NNVM compiler is removed from the mainline as of #4571 The core nnvm data
structures still will be maintained.
---
tqchen closed issue #4496: [NNVM][RFC] Begin the removal of NNVM Compiler
URL: https://github.com/apache/incubator-tvm/issues/4496
This is an automated message from the Apache Git Service.
To respond to the message, please lo
tqchen opened a new issue #4579: [RFC] Deprecate Extension VTable in Favor of
Unified Object System
URL: https://github.com/apache/incubator-tvm/issues/4579
Before the unified object protocol, we support passing additional extension
objects around by declaring a type as an extension type.
tqchen commented on issue #4578: [RUNTIME] Remove Extension VTable in favor of
Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578#issuecomment-568615838
RFC https://github.com/apache/incubator-tvm/issues/4579
--
tqchen edited a comment on issue #4578: [RUNTIME] Remove Extension VTable in
favor of Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578#issuecomment-568615194
cc @junrushao1994 @yzhliu
This is an
tqchen commented on issue #4578: [RUNTIME] Remove Extension VTable in favor of
Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578#issuecomment-568615194
cc @junrushao1994
This is an automated messa
tqchen opened a new pull request #4578: [RUNTIME] Remove Extension VTable in
favor of Unified Object system.
URL: https://github.com/apache/incubator-tvm/pull/4578
Before the unified object protocol, we support pass
additional extension objects around by declaring a type as an extension
anijain2305 commented on issue #4508: Broadcasting is broken with
LayoutTransform
URL: https://github.com/apache/incubator-tvm/issues/4508#issuecomment-568597152
https://github.com/apache/incubator-tvm/pull/4577
This is not just expand dims problem. The input shape is (1,) - if we ta
anijain2305 opened a new pull request #4577: [Relay][AlterLayout] Broadcast
with scalar shape
URL: https://github.com/apache/incubator-tvm/pull/4577
Relevant Issue - https://github.com/apache/incubator-tvm/issues/4508
Preventing insertion of layout transform and expand dims when the
jwfromm commented on a change in pull request #4566: [Relay/Topi][Op] Added
native DepthToSpace and SpaceToDepth Operators
URL: https://github.com/apache/incubator-tvm/pull/4566#discussion_r360763434
##
File path: python/tvm/relay/frontend/onnx.py
##
@@ -539,35 +539,7 @@ c
anijain2305 commented on a change in pull request #4335: [Relay] ConvertLayout
pass.
URL: https://github.com/apache/incubator-tvm/pull/4335#discussion_r361000359
##
File path: src/relay/pass/transform_layout.h
##
@@ -0,0 +1,333 @@
+/*
+ * Licensed to the Apache Software Fo
tqchen edited a comment on issue #4572: What is an acceptable way to write an
one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568567919
Please move this kind of discussions directly to https://discuss.tvm.ai/
instead :)
---
tqchen commented on issue #4572: What is an acceptable way to write an one_hot
with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568567919
Please move this kind of discussions directly to https://discuss.tvm.ai/
instead
-
tqchen closed issue #4572: What is an acceptable way to write an one_hot with
hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572
This is an automated message from the Apache Git Service.
To respond to th
anijain2305 commented on issue #4335: [Relay] ConvertLayout pass.
URL: https://github.com/apache/incubator-tvm/pull/4335#issuecomment-568566850
> I have many transpose when using tensor core due to NCHW. About 20% of the
time is doing transpose.
>
> So, I want a NCHW to NHWC converte
tqchen commented on issue #4284: [TEST][FLAKY]
darknet/test_forward.py::test_forward_conv_batch_norm
URL: https://github.com/apache/incubator-tvm/issues/4284#issuecomment-568566478
could due to random number for variance issue, introduced a solution in
https://github.com/apache/incubator-t
tqchen closed issue #4284: [TEST][FLAKY]
darknet/test_forward.py::test_forward_conv_batch_norm
URL: https://github.com/apache/incubator-tvm/issues/4284
This is an automated message from the Apache Git Service.
To respond to
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from 9bf2bee Fix llvm-enabled build by adding missing intrinsics headers
(#4575)
add f9bc748 [DEPRECATION]
tqchen merged pull request #4571: [DEPRECATION] Remove NNVM compiler
URL: https://github.com/apache/incubator-tvm/pull/4571
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
tqchen opened a new pull request #4576: [DEPRECATION] Cleanup legacy verilog
support
URL: https://github.com/apache/incubator-tvm/pull/4576
This PR cleans up the left over code for legacy verilog support which was
experimental.
The new hardware backend path is now support by VTA via TSI
tqchen commented on a change in pull request #4558: [NEWS] add v0.6 release
URL: https://github.com/apache/incubator-tvm/pull/4558#discussion_r360965535
##
File path: NEWS.md
##
@@ -26,6 +26,899 @@ Refer to the Roadmap issue for complete list on on-going
version features.
tqchen edited a comment on issue #4569: Some Windows and MSVC fixes
URL: https://github.com/apache/incubator-tvm/pull/4569#issuecomment-568542825
Sorry I wasn't being clear, I agree that it is important to keep it,
however, it might be better to implement it as follows(so condition is only
tqchen commented on issue #4569: Some Windows and MSVC fixes
URL: https://github.com/apache/incubator-tvm/pull/4569#issuecomment-568542825
Sorry I wasn't being clear, I agree that it is important to keep it,
however, it might be better to implement it as follows(so condition is only
execut
zhiics merged pull request #4575: Fix llvm-enabled build by adding missing
intrinsics headers
URL: https://github.com/apache/incubator-tvm/pull/4575
This is an automated message from the Apache Git Service.
To respond to the
This is an automated email from the ASF dual-hosted git repository.
zhic pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from 9ec0e5c remove unnecessary cast to int32 (#4573)
add 9bf2bee Fix llvm-enabled build by adding missing int
zhiics commented on issue #4575: Fix llvm-enabled build by adding missing
intrinsics headers
URL: https://github.com/apache/incubator-tvm/pull/4575#issuecomment-568524242
Thanks @dmakarov
This is an automated message from the
zhiics merged pull request #4573: [ONNX] Remove unnecessary cast of constants
to int32
URL: https://github.com/apache/incubator-tvm/pull/4573
This is an automated message from the Apache Git Service.
To respond to the messag
This is an automated email from the ASF dual-hosted git repository.
zhic pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from dfc4009 [VTA][Chisel] End-to-end Inference with Chisel VTA (#4574)
add 9ec0e5c remove unnecessary cast to
tmoreau89 merged pull request #4574: [VTA][Chisel] End-to-end Inference with
Chisel VTA
URL: https://github.com/apache/incubator-tvm/pull/4574
This is an automated message from the Apache Git Service.
To respond to the messa
This is an automated email from the ASF dual-hosted git repository.
moreau pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from 79581dd Remove nnvm (#4565)
add dfc4009 [VTA][Chisel] End-to-end Inference with Chisel VTA (#4574)
No
tmoreau89 commented on issue #4574: [VTA][Chisel] End-to-end Inference with
Chisel VTA
URL: https://github.com/apache/incubator-tvm/pull/4574#issuecomment-568522609
The PR has been merged.
This is an automated message from th
liangfu edited a comment on issue #4574: [VTA][Chisel] End-to-end Inference
with Chisel VTA
URL: https://github.com/apache/incubator-tvm/pull/4574#issuecomment-568465219
This is actually a fix to accept negative imm in ALU ops.
Please feel free to rename the title. @tmoreau89
dmakarov opened a new pull request #4575: Fix llvm-enabled build by adding
missing intrinsics headers
URL: https://github.com/apache/incubator-tvm/pull/4575
This is an automated message from the Apache Git Service.
To re
kice commented on issue #3879: Cast from float16 to uint8 was not supported by
CUDA
URL: https://github.com/apache/incubator-tvm/issues/3879#issuecomment-568471644
Also I found before CUDA 10.2, there is no support for absolute value APIs
for half-precision.
--
liangfu commented on issue #4574: [VTA][Chisel] End-to-end Inference with
Chisel VTA
URL: https://github.com/apache/incubator-tvm/pull/4574#issuecomment-568465219
This is actually a fix to accept negative imm in ALU ops.
Please feel free to rename the title. @tmoreau89
-
deepakbabel commented on issue #4478: [TOPI] implement pool3d op
URL: https://github.com/apache/incubator-tvm/pull/4478#issuecomment-568455107
@optima2005 : Thank you for the suggestion. I actually have done the
same(seperate function for 3dpooling) in tensorflow.py file(Frontend support).
liangfu opened a new pull request #4574: [VTA][Chisel] End-to-end Inference
with Chisel VTA
URL: https://github.com/apache/incubator-tvm/pull/4574
This PR brings end-to-end inference capability to Chisel version of VTA.
This
FrozenGene commented on issue #4572: What is an acceptable way to write an
one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568426268
```python
@tvm.hybrid.script
def one_hot(a):
c = output_tensor((a.shape[0] * 10, ), 'float32')
ghostplant edited a comment on issue #4572: What is an acceptable way to write
an one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568416313
@FrozenGene Yes, but the compute node doesn't support build it with
`tvm.build()`.
ghostplant commented on issue #4572: What is an acceptable way to write an
one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568416313
@FrozenGene Yes, but the compute node doesn't support build it.
---
FrozenGene commented on issue #4572: What is an acceptable way to write an
one_hot with hybrid_script?
URL: https://github.com/apache/incubator-tvm/issues/4572#issuecomment-568415359
```python
import tvm
import numpy as np
@tvm.hybrid.script
def one_hot(a):
c = output_ten
masahi opened a new pull request #4573: [ONNX] Remove unnecessary cast to int32
URL: https://github.com/apache/incubator-tvm/pull/4573
-
55 matches
Mail list logo
