pytorch/xla

Pull Requests Commits

Update ops_xla_shape_fn.h

ManfeiBai committed 1 year ago

Verified fa2dffa7

Update ops_xla_shape_fn.cpp

ManfeiBai committed 1 year ago

Verified e129294f

Update ops_lower_fn.cpp

ManfeiBai committed 1 year ago

Verified c0328a42

Delete torch_xla/csrc/ops/topk.cpp

ManfeiBai committed 1 year ago

Verified 238dc7fc

Delete torch_xla/csrc/ops/topk.h

ManfeiBai committed 1 year ago

Verified 63bbe1d5

Update aten_xla_type.cpp

ManfeiBai committed 1 year ago

Verified e7480721

Update ops_xla_shape_fn.cpp

ManfeiBai committed 1 year ago

Verified ee03ca6e

Update ops_xla_shape_fn.h

ManfeiBai committed 1 year ago

Verified 3a0cda14

Update xla_native_functions.yaml

ManfeiBai committed 1 year ago

Verified ba1895a7

Create sharded data placeholders directly (#6203)

will-cromar committed 1 year ago

Verified efa6fcfd

Remove Colab notebooks (#6155)

will-cromar committed 1 year ago

Verified 26dcba94

benchmarks: add nightly.sh to automate regular benchmarking (#6231)

cota committed 1 year ago

Verified 99fdf9a2

experiment_runner: catch AMP exception when building child environment (#6229)

cota committed 1 year ago

Verified 9c320743

Promote int to float for acos operation (consistent with Pytorch) (#6224)

ManfeiBai committed 1 year ago

Verified 22a6e6ea

Fix erf, atan2 for integers (#6216)

qihqi committed 1 year ago

Verified aff50b80

Revert "Re-land: Fix model initialization. (#6182)" (#6220)

cota committed 1 year ago

Verified bcefe806

Pin expecttest version (#6221)

JackCaoG committed 1 year ago

Verified 0f16471e

Enable core aten test and loose tolerance for aten_log10 and aten_log2 (#6222)

wonjoo-wj committed 1 year ago

Verified c9c2f3bd

Update aten_exp to do a manual type conversion (#6214)

wonjoo-wj committed 1 year ago

Verified 7339c82e

Add manual type conversion for aten_asin (#6213)

wonjoo-wj committed 1 year ago

Verified 8af15d04

Update dtype conversion for aten_tan (#6211)

wonjoo-wj committed 1 year ago

Verified 051a0be1

Enable some core aten tests (#6215)

qihqi committed 1 year ago

Verified e5039866

benchmark/experiment_runner: save accelerator_model on error (#6218)

cota committed 1 year ago

Verified 86636ba7

benchmarks/aggregate: annotate latest plot with model names (#6217)

cota committed 1 year ago

Verified dc8948d9

benchmarks/aggregate: skip first three data points (#6210)

cota committed 1 year ago

Verified 0eb9afbe

benchmarks/torchbench_model: skip benchmarks that fail to load (#6199)

cota committed 1 year ago

Verified 3ccb4ed2

Dump dynamo counters (#6201)

frgossen committed 1 year ago

Verified 9a92472e

benchmarks/aggregate: make fig height and width configurable (#6209)

cota committed 1 year ago

Verified 56b72c41

Add option to dump HLOs during benchmark runs (#6200)

frgossen committed 1 year ago

Verified 61d42ebe

experiment_runner: revamp options to dump PyTorch/XLA metrics (#6186)

frgossen committed 1 year ago

Verified ad7c1412

Older