deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431623053
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431623431
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
junrushao1994 opened a new pull request #5687:
URL: https://github.com/apache/incubator-tvm/pull/5687
Steps
- [x] Unify them in `src/`
- [ ] Unify them in `include/`
- [ ] Unify them in `python/` and `tests/cpp`
- [ ] Fix potential bugs
-
srkreddy1238 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431629053
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431635377
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431635461
##
File path: tests/python/frontend/tensorflow/test_forward.py
##
@@ -3179,10 +3183,342 @@ def test_forward_isfinite():
_verify_infiniten
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431636410
##
File path: tests/python/frontend/tensorflow/test_forward.py
##
@@ -3179,10 +3183,342 @@ def test_forward_isfinite():
_verify_infiniten
siju-samuel opened a new pull request #5688:
URL: https://github.com/apache/incubator-tvm/pull/5688
Fix the issue of batch size reported in
https://discuss.tvm.ai/t/yolov3-tiny-batch-input-test-failed/6796
@FrozenGene Please help to review this PR. TIA.
maheshambule commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431643670
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -3155,6 +3176,91 @@ def _convert_control_flow_operator(self, node, inputs,
attrs,
maheshambule commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r431651445
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
maheshambule commented on a change in pull request #5052:
URL: https://github.com/apache/incubator-tvm/pull/5052#discussion_r431653020
##
File path: python/tvm/contrib/target/onnx.py
##
@@ -0,0 +1,755 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more c
jroesch commented on pull request #5596:
URL: https://github.com/apache/incubator-tvm/pull/5596#issuecomment-635212257
Thanks! LGTM
This is an automated message from the Apache Git Service.
To respond to the message, please
This is an automated email from the ASF dual-hosted git repository.
jroesch 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 a072da0 [TIR][REFACTOR] std::string ->
jroesch merged pull request #5596:
URL: https://github.com/apache/incubator-tvm/pull/5596
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to
poincarelee commented on issue #4371:
URL: https://github.com/apache/incubator-tvm/issues/4371#issuecomment-635215942
Thanks a lot, this problem was solved indeed. I retried and it worked.
Thanks again.
At 2020-05-25 09:43:0
dhruvaray commented on a change in pull request #5447:
URL: https://github.com/apache/incubator-tvm/pull/5447#discussion_r431854463
##
File path: topi/include/topi/transform.h
##
@@ -1309,5 +1309,57 @@ inline Tensor one_hot(const Tensor& indices, const
PrimExpr on_value, const
dhruvaray commented on a change in pull request #5447:
URL: https://github.com/apache/incubator-tvm/pull/5447#discussion_r431854463
##
File path: topi/include/topi/transform.h
##
@@ -1309,5 +1309,57 @@ inline Tensor one_hot(const Tensor& indices, const
PrimExpr on_value, const
cchung100m commented on pull request #5596:
URL: https://github.com/apache/incubator-tvm/pull/5596#issuecomment-635418845
Thanks to @jroesch :)
This is an automated message from the Apache Git Service.
To respond to the messa
siju-samuel commented on pull request #5683:
URL: https://github.com/apache/incubator-tvm/pull/5683#issuecomment-635427158
BERT has dependancy with pip package `pytorch_pretrained_bert`. i will add
some bert testcases later. May be we can add as another tutorial also.
---
kevinthesun commented on a change in pull request #4312:
URL: https://github.com/apache/incubator-tvm/pull/4312#discussion_r432043625
##
File path: python/tvm/relay/op/_transform.py
##
@@ -99,8 +99,80 @@ def _arange_shape_func(start, stop, step):
@_reg.register_shape_func("a
kevinthesun commented on a change in pull request #4312:
URL: https://github.com/apache/incubator-tvm/pull/4312#discussion_r432044019
##
File path: python/tvm/relay/op/_transform.py
##
@@ -99,8 +99,80 @@ def _arange_shape_func(start, stop, step):
@_reg.register_shape_func("a
kevinthesun commented on a change in pull request #4312:
URL: https://github.com/apache/incubator-tvm/pull/4312#discussion_r432045059
##
File path: python/tvm/relay/op/transform.py
##
@@ -611,31 +611,41 @@ def split(data, indices_or_sections, axis=0):
return TupleWrapper(_
kevinthesun commented on a change in pull request #4312:
URL: https://github.com/apache/incubator-tvm/pull/4312#discussion_r432049322
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -614,6 +614,52 @@ def _impl(inputs, attr, params, mod):
return out
ret
tqchen closed issue #5673:
URL: https://github.com/apache/incubator-tvm/issues/5673
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to
tqchen commented on issue #5673:
URL: https://github.com/apache/incubator-tvm/issues/5673#issuecomment-635545203
Thanks for reporting the problem, let us move to the discuss thread
https://discuss.tvm.ai/t/mxnet-group-conv-not-support-very-well-by-tvm/6811/3
--
tqchen edited a comment on pull request #5681:
URL: https://github.com/apache/incubator-tvm/pull/5681#issuecomment-635592715
https://github.com/apache/incubator-tvm/blob/master/tests/python/relay/test_ir_text_printer.py
This
tqchen commented on pull request #5681:
URL: https://github.com/apache/incubator-tvm/pull/5681#issuecomment-635592715
we can create a new file tests/relay/test_text_printer.py
This is an automated message from the Apache Git
comaniac opened a new pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689
Discussion: https://discuss.tvm.ai/t/patternlang-match-constant-nodes/6835
Changelog:
* Adds `ConstantPattern` to the pattern language.
* Add unit tests of constant pattern.
* Eli
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635629161
> To make other syntactic sugar, you could add an is_const() that simiply
creats a new ConstPattern, but I don't think it's strictly necessary.
I could do that. How
comaniac edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635629161
> To make other syntactic sugar, you could add an is_const() that simiply
creats a new ConstPattern, but I don't think it's strictly necessary.
I could do th
mbaret commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635633008
Thanks for the PR so soon! Is there an example of how partition works on a
constant match? In particular, does the constant remain propagated into the
body?
On a gen
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635637682
@mbaret Originally the pass embedded constants in the function body, but
@comaniac filed #5662, and I responded with #5663, so it currently will lift
the constants to t
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635637903
> Thanks for the PR so soon! Is there an example of how partition works on a
constant match? In particular, does the constant remain propagated into the
body?
An e
mbrookhart edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635637682
@mbaret Originally the partition pass embedded constants in the function
body, but @comaniac filed #5662, and I responded with #5663, so it currently
will lift t
mbaret commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635640183
```
%2 = fn (%input: Tensor[(1, 224, 224, 3), uint8], Composite="qnn_conv2d") ->
Tensor[(1, 224, 224, 64), uint8] {
%0 = qnn.conv2d(%input, meta[relay.Constant][0]
comaniac edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635637903
> Thanks for the PR so soon! Is there an example of how partition works on a
constant match? In particular, does the constant remain propagated into the
body?
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r432158182
##
File path: tests/python/frontend/tensorflow/test_forward.py
##
@@ -3179,10 +3183,342 @@ def test_forward_isfinite():
_verify_infiniten
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r432158461
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -2896,15 +2903,29 @@ def _parse_import_prerequisites(self, graph):
"""
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r432158330
##
File path: tests/python/frontend/tensorflow/test_forward.py
##
@@ -3179,10 +3183,342 @@ def test_forward_isfinite():
_verify_infiniten
deepakbabel23 commented on a change in pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#discussion_r432158588
##
File path: python/tvm/relay/frontend/tensorflow.py
##
@@ -3155,6 +3176,91 @@ def _convert_control_flow_operator(self, node, inputs,
attrs,
deepakbabel23 commented on pull request #5617:
URL: https://github.com/apache/incubator-tvm/pull/5617#issuecomment-635643165
@srkreddy1238,
I have incorporated all your review comments. Please take a look at the
fresh changes. Also quickly evaluated if we can consider using
convert_vari
masahi commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I expect constants to be available to external codegen at compile time.
Otherwise we cannot do constant folding.
@comaniac Under the current constant impl of partitio
masahi edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I expect constants to be available to external codegen at compile time.
Otherwise we cannot do constant folding.
@comaniac Under the current constant impl of p
masahi edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I expect constants to be available to external codegen at compile time.
Otherwise we cannot do constant folding.
@comaniac Under the current implementation of
masahi edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I expect constants to be available to external codegen at compile time.
Otherwise we cannot do constant folding.
@comaniac Under the current implementation of
masahi edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I expect constants to be available to external codegen at compile time.
Otherwise we cannot do constant folding.
@comaniac Under the current implementation of
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635646234
:point_up: I'll let you guys discuss the appropriate behavior, there's seems
to be some complication to this.
In the mean time, this was my initial assumption:
masahi edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635644901
I might be missing something, but I expect constants to be available to
external codegen at compile time. Otherwise we cannot do constant folding.
Sorry I didn't put
mbrookhart edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635646234
:point_up: I'll let you guys discuss the appropriate behavior, there's seems
to be some complication to this.
In the mean time, this was my initial assump
masahi commented on pull request #5683:
URL: https://github.com/apache/incubator-tvm/pull/5683#issuecomment-635648772
You can add BERT tests to the repo and only execute them if import succeeds.
They won't be run on CI but we can test them locally. Since BERT seems popular,
it would be nic
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635649921
I made an example and tested it with old `MergeComposite` pass (master
commit 6100112a150540588ddc9abb36dea0ff961f4301). It behaves as I expected. The
partitioned composi
mbaret commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635650558
Try changing w in the pattern to a relay.const rather than a var.
This is an automated message from the Apa
masahi commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635658618
> @masahi I just checked with this PR. The unit test hits the line you
pointed out twice.
hmm something seems off to me. On the dnnl fused mobilenet tests, I think
a
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635663626
OK I can finally reproduce this case:
```
fn (%x: Tensor[(1, 3, 224, 224), float32], %b: Tensor[(3), float32]) ->
Tensor[(1, 3, 222, 222), float32] {
%1 = f
comaniac edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635663626
OK I can finally reproduce this case:
```
fn (%x: Tensor[(1, 3, 224, 224), float32], %b: Tensor[(3), float32]) ->
Tensor[(1, 3, 222, 222), float32] {
comaniac edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635663626
OK I can finally reproduce this case:
```
fn (%x: Tensor[(1, 3, 224, 224), float32], %b: Tensor[(3), float32]) ->
Tensor[(1, 3, 222, 222), float32] {
mbaret commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635665317
It's not really the case that old merge composite explicitly 'lifts'
constant nodes if they're vars, in old merge composite vars indicate inputs to
the pattern (so they don
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635666562
I think I agree with @mbaret here. I think I see a simple way to do that
behavior by default, why don't I post another PR in that direction after this
goes in?
--
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635667979
That's the behavior I just realized with above examples. In this case, what
would be the behavior if we specify `VarPattern('x') | ConstantPattern()`?
--
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635668518
I think it depends on which one matched? If it's a var, we lift it, if it's
a constant, we embed?
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635672040
OK so here is a conclusion for another follow-up PR to deal with constant
lifting.
- Pattern input `ConstantPattern`:
- Only match constant nodes and keep th
comaniac edited a comment on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635672040
OK so here is a conclusion for another follow-up PR to deal with constant
lifting.
- Pattern input `ConstantPattern`:
- Only match constant nodes and
zhiics opened a new pull request #5690:
URL: https://github.com/apache/incubator-tvm/pull/5690
Per #5650, this PR moves this fallback_device field in `PassContext` into
config.
@tqchen @junrushao1994
This is an auto
zhiics commented on issue #5650:
URL: https://github.com/apache/incubator-tvm/issues/5650#issuecomment-635674079
I realize we may also want to remove `relay.build_config` and directly use
PassContext to keep it consistent to tir, but it may invoke a lot of changes.
@tqchen thoughts?
---
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635674194
I'd liked to refine slightly:
- Pattern input ExprPattern(relay.const):
- Only match constant nodes that match the value and embed them in the
partitioned func
mbrookhart commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635675067
@comaniac The input VarPattern name is already optional, if you use
is_input() it will match any VarNode
https://github.com/apache/incubator-tvm/blob/a072da0588c5427
masahi commented on a change in pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#discussion_r432183447
##
File path: docs/langref/relay_pattern.rst
##
@@ -131,6 +136,44 @@ The next example is matching a pattern of batch_norm ->
get(0) -> relu:
masahi commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635675471
@comaniac @mbrookhart good to go?
This is an automated message from the Apache Git Service.
To respond to t
comaniac commented on pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689#issuecomment-635675585
> @comaniac @mbrookhart good to go?
Yeah I'm fine with that.
This is an automated message from the
This is an automated email from the ASF dual-hosted git repository.
masahi pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from a072da0 [TIR][REFACTOR] std::string -> String Migration in TIR nodes
(#5596)
add 95b3ad9 [PatternLang]
masahi merged pull request #5689:
URL: https://github.com/apache/incubator-tvm/pull/5689
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to g
tqchen commented on issue #5650:
URL: https://github.com/apache/incubator-tvm/issues/5650#issuecomment-635678608
I agree we should do that as well
This is an automated message from the Apache Git Service.
To respond to the me
tqchen commented on a change in pull request #5690:
URL: https://github.com/apache/incubator-tvm/pull/5690#discussion_r432187036
##
File path: tests/python/relay/test_pass_annotation.py
##
@@ -538,9 +539,10 @@ def expected():
expected_index = [2, 2, 2, 1, 1, 1, 2, 2]
tqchen commented on pull request #5683:
URL: https://github.com/apache/incubator-tvm/pull/5683#issuecomment-635679640
One middle ground would be manually construct some snippet that relates to
bert but not directly use th bert model itself
-
masahi commented on pull request #5683:
URL: https://github.com/apache/incubator-tvm/pull/5683#issuecomment-635683945
yes this is also a good option if the model is not too complicated. It is up
to @siju-samuel. See our lstm tests
https://github.com/apache/incubator-tvm/blob/master/tests/p
tqchen opened a new pull request #5691:
URL: https://github.com/apache/incubator-tvm/pull/5691
Please join us to welcome @masahi as a new PPMC member of the TVM community.
Masa is an early TVM contributor. He has been contributing various modules
such as pytorch control flow import, AMD R
tqchen opened a new pull request #5692:
URL: https://github.com/apache/incubator-tvm/pull/5692
Please join us to welcom @zhiics as a new PPMC member of the TVM community.
Zhi has contributed to various aspects of TVM. His primary contribution
includes the relay pass manager, code gen
cchung100m commented on pull request #5691:
URL: https://github.com/apache/incubator-tvm/pull/5691#issuecomment-635713551
Congratulations! @masahi :)
This is an automated message from the Apache Git Service.
To respond to the
cchung100m commented on pull request #5692:
URL: https://github.com/apache/incubator-tvm/pull/5692#issuecomment-635713660
Congratulations! @zhiics :)
This is an automated message from the Apache Git Service.
To respond to the
zhiics commented on a change in pull request #5690:
URL: https://github.com/apache/incubator-tvm/pull/5690#discussion_r432223352
##
File path: tests/python/relay/test_pass_annotation.py
##
@@ -538,9 +539,10 @@ def expected():
expected_index = [2, 2, 2, 1, 1, 1, 2, 2]
mbrookhart opened a new pull request #5693:
URL: https://github.com/apache/incubator-tvm/pull/5693
Following up from conversation in #5689, this PR implements these rules for
lifting or embedding constants in Partitioned Functions:
- Pattern input ExprPattern(relay.const):
-
tqchen commented on issue #5559:
URL: https://github.com/apache/incubator-tvm/issues/5559#issuecomment-635724547
@roastduck would you be interested in taking over the pass?
This is an automated message from the Apache Git Ser
tqchen commented on pull request #5601:
URL: https://github.com/apache/incubator-tvm/pull/5601#issuecomment-635725119
@liangfu please followup with the reviews. @gussmith23 it would be great if
you can also help to take a look
--
liangfu commented on pull request #5456:
URL: https://github.com/apache/incubator-tvm/pull/5456#issuecomment-635727096
Just a high-level question, how shall we integrate with LLVM?
The options we have
* Search for LLVM in default system library, like /usr/lib,
/usr/lib/x86_64-linux-g
liangfu edited a comment on pull request #5456:
URL: https://github.com/apache/incubator-tvm/pull/5456#issuecomment-635727096
Just a high-level question, how shall we integrate with LLVM?
The options we have
* Search for llvm-config
* Search for LLVM in default system library, lik
tobegit3hub opened a new pull request #5694:
URL: https://github.com/apache/incubator-tvm/pull/5694
Add more dtypes support for TVMDSOOp which expands the scenarios of TVM
operators in TensorFlow graph.
This is an automated
roastduck commented on issue #5559:
URL: https://github.com/apache/incubator-tvm/issues/5559#issuecomment-635739066
I met some difficulties improving this pass. For now, I'm not going to take
over it.
This pass massively utilizes low level semantics such as `PostOrderVisit`
(instead
zhiics commented on issue #5650:
URL: https://github.com/apache/incubator-tvm/issues/5650#issuecomment-635742328
One potential problem is that it may break downstream deployment.
This is an automated message from the Apache G
tqchen commented on issue #5650:
URL: https://github.com/apache/incubator-tvm/issues/5650#issuecomment-635744914
We could mark build_config as deprecation, and remove that API in the next
release cycle. In the meanwhile, we should replace all of the current usages.
---
dhruvaray commented on pull request #5329:
URL: https://github.com/apache/incubator-tvm/pull/5329#issuecomment-635745170
@FrozenGene - Could you kindly review and merge?
This is an automated message from the Apache Git Servic
tqchen commented on issue #5559:
URL: https://github.com/apache/incubator-tvm/issues/5559#issuecomment-635745118
We could certainly rewrite the pass completely, instad of the PostOrderVisit
This is an automated message from t
zhiics commented on issue #5650:
URL: https://github.com/apache/incubator-tvm/issues/5650#issuecomment-635745547
Sounds like a good plan.
This is an automated message from the Apache Git Service.
To respond to the message, pl
majiang31312 commented on issue #5686:
URL: https://github.com/apache/incubator-tvm/issues/5686#issuecomment-635759813
I have encounted the same problem.
It's triggered by 7b74a8672e1e40e7541c0007d8628586c62277e8, as vulkan use
some cuda codes in topi. Forcing the sched_warp_softmax (i
liangfu commented on a change in pull request #5601:
URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r432266454
##
File path: src/tir/transforms/bf16_legalize.cc
##
@@ -0,0 +1,384 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or mor
94 matches
Mail list logo