Cherry pick (batch 1) to rel-1.5.1 #5266
tianleiwu
merged
26 commits into
rel-1.5.1
from
tlwu/1.5.1_cherry_pick_1
Fix a bug in EmbedLayerNorm fusion (#5150)
3119a109
Amdmigraphx improvements (#5158)
7cb20aab
Fix softmax_warp_backward math when is_log_softmax = True and registe…
23ac6ac3
MaxPool versioning in quantization tools. (#5194)
b4e36c72
Add API to allow configuration of the global thread pools. (#5199)
bc5b88f9
Corrects doc typos and formatting (#5201)
da143113
NCHWc optimizer fixes for quantized models (#5203)
81648ce9
Fix order of returned values in quantize_weight_per_channel (#5205)
5f3c7a36
Remove shape inference and fix save large model(>2g) issue (#5210)
51ea1595
--shm-size=1024m to fix nccl shared memory issue (#5214)
2459b4c9
Update conversion script and process to simplify creating ORT format …
35853bf7
fix _WIN32 (#5218)
0b0ae5a5
[java] Fixing the buffer semantics. (#5223)
3bd04550
Create profile for all dynamic shape input tensors (#5229)
9810c15d
Fix TransposeScaleMatMul and MatMulScaleFusion issues (#5230)
1ad0c26b
Revert to using release SafeInt repo now that it supports a build wit…
855da64d
update TensorRT docs (#5238)
f2c4ac62
Update BUILD.md for CUDA versions (#5239)
e6b8bc50
Update BUILD.md training dependency info. (#5240)
cac53ca6
Place Shape's output in CPU memory (#5245)
e93c5e38
Fix possible ios build break after update to Xcode 12 (#5246)
f52c5714
Fix reshape fusion crash (#5252)
9bbdc791
bump version to 1.5.1 (#5258)
c40e0769
Fix bug in the back to back quantization of matmul and conv (#5264)
b29ef24f
ORT DirectML EP for Iron release, ONNX 1.5 (part 2) (#5263)
1c9e7b6a
Handle missing initializers in allocation planner to fix crashes with…
06ff3c70
edgchen1
approved these changes
on 2020-09-23
pranavsharma
approved these changes
on 2020-09-23
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub