- 16 4月, 2020 9 次提交
-
-
由 Sachin Joglekar 提交于
PiperOrigin-RevId: 306667501 Change-Id: Id91da849cd77dd8861786841d8194c5e6d2a04fb
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306666603 Change-Id: I45c877249c4305a5a89c1ce5fab5d731484d3a80
-
由 Advait Jain 提交于
PiperOrigin-RevId: 306663879 Change-Id: I61bd553f79fcd0d97ab86a8952972c210d368be1
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306661503 Change-Id: I6db0405936c800ecc6338738fb1d0f995222c2d0
-
由 Blake Hechtman 提交于
PiperOrigin-RevId: 306657217 Change-Id: I1e48bb858da585fbc421092da0e6172ea3cf5d56
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306656469 Change-Id: I8591530f51c818380b76e3a7c42944503da523ac
-
由 Juhyun Lee 提交于
PiperOrigin-RevId: 306655077 Change-Id: I30b1577b5d25a51f54a75cc1b8635daa628a92b8
-
由 Juhyun Lee 提交于
PiperOrigin-RevId: 306654653 Change-Id: I0cc7a6ccb0ef2965fc8ce30f5ca6d80ce1759e18
-
由 Dan Moldovan 提交于
Internal cleanup: process global and nonlocal in a more consistent fashion during static analysis. Disconnect definitions of symbols between local functions and their parent. PiperOrigin-RevId: 306653572 Change-Id: Ia0ced96af49b28d58fa8d5a69a5d7c4bad17a663
-
- 15 4月, 2020 31 次提交
-
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306652585 Change-Id: I53312a7f02ee07893816b1f358667e324aa879a6
-
由 Mihai Maruseac 提交于
PiperOrigin-RevId: 306652032 Change-Id: Iabf3914e9dd8012bbc4814baaa7dc47ada64ca74
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306650050 Change-Id: I49df540bab790bb4e5be83fc4244871c2ac5321a
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306641868 Change-Id: I2cf43db1b1d80646d55f8ab178584498e616fa3e
-
由 Alexander Belyaev 提交于
PiperOrigin-RevId: 306631896 Change-Id: Id48abba3cce826184cba5348a1f1b9226caa2f87
-
由 Alexander Belyaev 提交于
PiperOrigin-RevId: 306631753 Change-Id: I1f00a25982905972418dd4d66447ddd42d428300
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306626034 Change-Id: Ia97b4b62cfa78f0b9f7252da1587354e341e9060
-
https://github.com/llvm/llvm-project/commit/129cf84e6953由 Benjamin Kramer 提交于
PiperOrigin-RevId: 306623856 Change-Id: I8a5a1ef0cb6152e72397dd8a93841da01b10534c
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306603118 Change-Id: I0c0788ebe2353059e5ae49e648157fffaf5a5ad0
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306601722 Change-Id: I09763375347b67d7f01b7e1d622307eec3433537
-
由 Renjie Liu 提交于
PiperOrigin-RevId: 306591715 Change-Id: I6a011b8ee753f7b541de5e3bf992d9eaab81caea
-
由 Chao Mei 提交于
PiperOrigin-RevId: 306590390 Change-Id: I98b5d17b95747420b2dd0c3c0c037a6088cd495e
-
由 TensorFlower Gardener 提交于
PiperOrigin-RevId: 306587683 Change-Id: Id3af4fca721d2c66a4893d52d4d350ba56d387ba
-
由 Mihai Maruseac 提交于
PiperOrigin-RevId: 306586051 Change-Id: Ib4e89016c825c6a3a9089b774a44dfc48ec01612
-
由 Sean Silva 提交于
This op arises during many dynamic-shape lowerings. For example, when lowering dot_general to a single batch matmul op that has a single innermost contracting dimension, one needs to transpose/reshape all the contracting dimensions to the innermost dimension and "flatten" them to a single dimension with a dynamic reshape. Also, move the folder and verifier of the non-dynamic reshape op to be next to each other. They somehow got separated. PiperOrigin-RevId: 306585385 Change-Id: I0f7769a3f3086ac6c2225a9024a76fdddef26715
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306580752 Change-Id: I68e733c099e93957eb1ccb728077563f0c1b3d2d
-
由 Scott Zhu 提交于
PiperOrigin-RevId: 306577348 Change-Id: I97f59ea373980b3303109cbf3ee7346a124db823
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306576920 Change-Id: I9154295d9c865b77674975392adb09438072b67c
-
由 Smit Hinsu 提交于
PrimitiveType::F16 maps to Eigen::half and interpreting it as float will fail. This incorrect handling is removed for now and later on this type along with bf16 and complex types can be supported by constructing appropriate ElementsAttr. PiperOrigin-RevId: 306574537 Change-Id: Icbd87b24cdc60656ac62de5a7829340f94594a9d
-
由 Gunhan Gulsoy 提交于
PiperOrigin-RevId: 306572687 Change-Id: I49448295cdd088eb114a7d518dd0db1822f2d7a6
-
由 Smit Hinsu 提交于
This enables unary ops in unary_ops_test.py that have kernel defined in tf2xla/kernels/unary_ops.cc and doesn't already have legalizations. Some tests are disabled if the op is not supported or either using unsigned int or complex constants. This also deletes unary_mlir_ops_test test now the old and new bridge tests are consolidated. PiperOrigin-RevId: 306570457 Change-Id: Idbaab1e8986c3659916ae0da28f60bf1960b9f4e
-
由 Juhyun Lee 提交于
PiperOrigin-RevId: 306569682 Change-Id: I702bd924cd8cca4103659a341f5c31ebdc3b5978
-
由 Marat Dukhan 提交于
PiperOrigin-RevId: 306569189 Change-Id: I049b590b9954ad53d04429c1a105317260d8cca4
-
由 Chao Mei 提交于
PiperOrigin-RevId: 306568160 Change-Id: I44dcddceb5a886b6337151ada0f7b758370c8d0c
-
由 Renjie Liu 提交于
PiperOrigin-RevId: 306566180 Change-Id: I23c2e7c4301c0478ad16c26d46a7ad0f0cecef70
-
由 Gaurav Jain 提交于
Unprotect should only be called on local handles. In order to test the triggering of forwarding for remote inputs to a function we add an optimization whereby EagerExecute releases the inputs of the eager operation. This enforces that a TFE_Op cannot be reused since the inputs would have been removed. This was technically already true since if the inputs were ever forwarded we should not be re-using the TFE_Op. PiperOrigin-RevId: 306564949 Change-Id: I94bd3a243658277891867802b792a4492ec0a039
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 306564339 Change-Id: I8f6484411eafc88d40034892c58e5cd12ece5e71
-
由 Feng Liu 提交于
This activation range is determined by the default min/max, scale and zero point from the UniformQuantizedType, and the activation function. PiperOrigin-RevId: 306561114 Change-Id: Ib48414263931b921295239499cc86cf2c92baa1b
-
由 Robert David 提交于
- Return immediately instead of break and returning at the end. - Don't reinterpret_cast or static_cast; just assign the pointer. - Add TF_LITE_ENSURE_STATUS around ConvertTensorType calls. - Fix a Clang warning (const auto& -> const auto*) PiperOrigin-RevId: 306560762 Change-Id: Ieeb765ae47fe337fa4fc0a57bb4177fe90069b90
-
由 Chao Mei 提交于
Support Hexagon delegate to have an option to specify the max number of partitions to be delegated, and changed the benchmark tool to support this feature. PiperOrigin-RevId: 306560419 Change-Id: I75148012edb7109e84fee213720c66a2385a09cb
-
由 Scott Zhu 提交于
PiperOrigin-RevId: 306560299 Change-Id: I100c3e23973276bc4f395c46b08705c0f277fb3b
-