Commit Graph

7427 Commits

Author SHA1 Message Date
lixian 6164578a22 add fast transpose algorithm 2020-08-16 23:00:49 +08:00
zhaozhenlong 220ec45181 fp16 concat 2020-08-16 22:20:28 +08:00
yangruoqi713 999a936c46 [MS][LITE] fix bug of arm cpu fp16 conv op 2020-08-16 22:14:49 +08:00
mindspore-ci-bot a0c12e7aa7 !4504 Add test cases for tflite models
Merge pull request !4504 from wangshaocong/lite
2020-08-16 20:17:49 +08:00
wsc c4c99ca627 Add test cases of tflite models 2020-08-16 19:48:58 +08:00
zhaozhenlong 5290cd843b fp16 transpose 2020-08-16 19:43:43 +08:00
mindspore-ci-bot 517c1d139c !4546 fix bug in formatTrans+transpose fusion
Merge pull request !4546 from hangq/master
2020-08-16 19:18:42 +08:00
mindspore-ci-bot 2d926d55d3 !4544 Add Inception_v4 and ml_object_detect Tflite Model
Merge pull request !4544 from jianghui/master
2020-08-16 19:16:01 +08:00
hangq 90b55937fe fix bug in formatTrans+transpose fusion 2020-08-16 19:01:01 +08:00
jianghui58 b79b853dc3 add inception_v4 ml_object_detect.tflite 2020-08-16 18:46:34 +08:00
mindspore-ci-bot 34b23e4ed3 !4452 Add Tflite Models
Merge pull request !4452 from jianghui/master
2020-08-16 18:40:43 +08:00
mindspore-ci-bot 2348e45815 !4541 [MS][LITE][Develop]optimize arithmetic fp16, gather and squezze support int32
Merge pull request !4541 from chenjianping/lite_dev2
2020-08-16 18:29:09 +08:00
chenjianping 3b1a048677 arithemetic support fp16,gather, squezze support int32 2020-08-16 18:21:52 +08:00
mindspore-ci-bot 5d7e02ea18 !4531 [MS][LITE]deconvolution fp16
Merge pull request !4531 from ling/deconv
2020-08-16 18:16:56 +08:00
mindspore-ci-bot 2e3f598726 !4542 change formatTrans+permute fusion to formatTrans+transpose fusion
Merge pull request !4542 from hangq/master
2020-08-16 18:07:42 +08:00
ling e603486104 deconv fp16 2020-08-16 18:02:55 +08:00
hangq f194cd6f9d change formatTrans+permute to formatTrans+transpose 2020-08-16 17:46:00 +08:00
mindspore-ci-bot 4b5549a6e9 !4537 disable arm32 assembly
Merge pull request !4537 from sunsuodong/arm32
2020-08-16 17:45:08 +08:00
mindspore-ci-bot 4bd40f3677 !4483 support fp16 models
Merge pull request !4483 from cjh9368/fp16
2020-08-16 17:44:59 +08:00
mindspore-ci-bot 8b3419c8a3 !4533 add model_emotions_0727_nosoftmax.tflite
Merge pull request !4533 from ghzl/add-emotions-model
2020-08-16 17:34:45 +08:00
mindspore-ci-bot baa6070ce0 !4536 add format_transform_op and permute_op fusion
Merge pull request !4536 from hangq/master
2020-08-16 17:14:50 +08:00
zhaizhiqiang 1b900ca44f optimize the interface of model and kernel registry 2020-08-16 17:04:11 +08:00
hangq cfef472285 add format_transform_op and permute_op fusion 2020-08-16 16:55:07 +08:00
sunsuodong 22efb0184b disable arm32 assembly 2020-08-16 16:50:48 +08:00
mindspore-ci-bot 106cadbaa4 !4530 add formatTransOp around eltwise fusion
Merge pull request !4530 from hangq/master
2020-08-16 16:38:41 +08:00
mindspore-ci-bot 3d6111c5e5 !4532 [MS][LITE][DEV]fp16 priority
Merge pull request !4532 from 张学同/to_merge
2020-08-16 16:38:08 +08:00
hangq aaaf554aed fix for eltwise and permute 2020-08-16 16:21:47 +08:00
guohongzilong 5617e21701 add model_emotions_0727_nosoftmax.tflite 2020-08-16 16:14:23 +08:00
mindspore-ci-bot 2808c0f8ea !4524 add to_format opencl op
Merge pull request !4524 from wandongdong/up
2020-08-16 16:14:21 +08:00
mindspore-ci-bot 125d021e1a !4529 [MS][LITE] optimize arm cpu fp16 conv op: add common converter functions for input and output
Merge pull request !4529 from yangruoqi713/lite
2020-08-16 16:06:14 +08:00
wandongdong 1112a7d5b9 add to_format opencl op 2020-08-16 00:58:02 -07:00
yangruoqi713 1b89036bea [MS][LITE] optimize arm cpu fp16 conv op: add common converter functions for input and output 2020-08-16 15:52:49 +08:00
mindspore-ci-bot 7c05853609 !4528 [MS][LITE][Develop]arithmetic support fp16
Merge pull request !4528 from chenjianping/lite_dev2
2020-08-16 15:48:19 +08:00
zhangxuetong 86199d522e float16 priority 2020-08-16 15:37:49 +08:00
chenjianping 6dafc9954a arithemetic support fp16 2020-08-16 15:30:30 +08:00
mindspore-ci-bot 1745c1c1d7 !4523 [MS][LITE] optimize arm cpu fp32/fp16 op: add assembly file for deconv depthwise border
Merge pull request !4523 from yangruoqi713/deconv_dw
2020-08-16 14:59:29 +08:00
mindspore-ci-bot 5270b33374 !4511 [MS][LITE]support parallel executor
Merge pull request !4511 from zhaizhiqiang/master
2020-08-16 14:59:14 +08:00
mindspore-ci-bot c7ff0f01cc !4486 rearch anf_importer and anf_exporter
Merge pull request !4486 from hangq/master
2020-08-16 14:56:37 +08:00
mindspore-ci-bot 6e7fb6601a !4495 Add Transpose and SpaceToBatch fp32 multi-thread
Merge pull request !4495 from zhongligeng/master
2020-08-16 14:54:11 +08:00
mindspore-ci-bot 45e1101214 !4525 rewrite conv3x3 conv winograd post func
Merge pull request !4525 from fuzhiye/tmp
2020-08-16 14:53:05 +08:00
mindspore-ci-bot fbc19f6b0c !4525 rewrite conv3x3 conv winograd post func
Merge pull request !4525 from fuzhiye/tmp
2020-08-16 14:53:04 +08:00
mindspore-ci-bot 1907b2073b !4518 add posttraining nets for benchmark
Merge pull request !4518 from andy_wangrui/master
2020-08-16 14:46:45 +08:00
fuzhiye 4333de375e rewrite conv3x3 conv winograd post func 2020-08-16 14:46:02 +08:00
hangq 93592f14a2 rearch anf_importer and anf_exporter 2020-08-16 13:06:31 +08:00
mindspore-ci-bot 154e338b69 !4522 rewrite fp16 conv1x1 && support global pooling
Merge pull request !4522 from fuzhiye/tmp
2020-08-16 12:18:10 +08:00
mindspore-ci-bot 7e5bdbe0e4 !4521 disable arm32 assembly
Merge pull request !4521 from sunsuodong/disable_arm32
2020-08-16 12:17:44 +08:00
yangruoqi713 59d9710447 [MS][LITE] optimize arm cpu fp32/fp16 op: add assembly file for deconv depthwise border 2020-08-16 12:16:38 +08:00
zhaizhiqiang 961b00db09 support parallel executor 2020-08-16 12:01:56 +08:00
fuzhiye cbab9e74d9 adjust fp16 1x1 conv 2020-08-16 11:54:15 +08:00
sunsuodong 00e323e47e disable arm32 assembly 2020-08-16 11:36:19 +08:00