Libin dev #1
Merged
Libin dev #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
1. Support new op where 2. Support cast op "to bool" mode 3. Support slice op "multi step" mode Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
feat: Support new op where and more mode for op cast and slice See merge request applied-machine-learning/sysml/mace!1470
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
Fix: fix APU memory leak problem
Fix: fix output node memory be reuse issue See merge request applied-machine-learning/sysml/mace!1477
Fix: fix half error when validate layer mode. See merge request applied-machine-learning/sysml/mace!1478
Fix: fix arm-linux-gnueabihf convert problem when generator libmace_static or convert running on cpu See merge request applied-machine-learning/sysml/mace!1479
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
fix: Fix maxPooling op accuracy loss issue See merge request applied-machine-learning/sysml/mace!1480
desc:
1.Support htp fp16 for transformer net
2.Disable gpu when htp, because dma-buf not need it
3.Fix data format issue at validate mode
4.Add transpose op for onnx-htp, because Conv is NHWC,
but MatMul is NCHW
5.Add TRANSFORM_BIASADD_TO_ADD, because htp not support biasadd
6.Add TRANSFORM_SLICE_TO_STRIDED_SLICE,
change to htp StridedSlice format
7.Add some log for debug
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
feat: Support htp fp16 mode See merge request applied-machine-learning/sysml/mace!1482
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
feat: upgrade qnn to 1.11.0 See merge request applied-machine-learning/sysml/mace!1483
Signed-off-by: yuanlonglong <yuanlonglong@xiaomi.com>
fix: qnn: fix reduce and conv2d bug See merge request applied-machine-learning/sysml/mace!1484
Signed-off-by: wenshuai <wenshuai@xiaomi.com>
feat: add EltwiseType of SQR_DIFF and output qnn log by VLOG. See merge request applied-machine-learning/sysml/mace!1486
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.